ࡱ >
2 5 % & ' ( ) * + , - . / 0 1 0 bjbjR|R| ; 0 0 ]
6 6 6 6 6 J J J 8 d | J d b ( ( ( ( ) + #, D c c c c c c c $ f li V d 9 6 g, ) ) g, g, d 6 6 ( ( Vd !3 !3 !3 g, ^ 6 ( 6 ( yU ~ !3 g, c !3 !3 = h > ( =(uQ / N ;> eU ld 0 d K> | i 1 i > > f i 6 -D 8 g, g, !3 g, g, g, g, g, d d !3 g, g, g, d g, g, g, g, i g, g, g, g, g, g, g, g, g, : Take-over time in highly automated vehicles: non-critical transitions to and from manual control
Alexander Eriksson, Neville A StantonTransportation Research Group, Faculty of Engineering and the Environment, University of Southampton, Boldrewood campus, SO16 7QF, UK
Running head: Take-over time in automated vehicles
Manuscript type: Research article
Word count: 4791
Corresponding author: Alexander Eriksson Transportation Research Group, Faculty of Engineering and the Environment, University of Southampton, Boldrewood campus, SO16 7QF, UK. Email: HYPERLINK "mailto:Alexander.eriksson@soton.ac.uk" Alexander.eriksson@soton.ac.uk
Acknowledgements: This research has been conducted as a part of the European Marie Curie ITN project HFAuto - Human Factors of Automated driving (PITN-GA-2013-605817)
Biographies
Alexander Eriksson, MSc, received his Master of Science degree in Cognitive Science from Linkping University in 2014 and is currently a Marie Curie Research Fellow in the EU funded Marie Curie International Training Network on Human Factors in Highly Automated Vehicles (HF-Auto) within the Faculty of Engineering and the Environment at the University of Southampton where he is undertaking his PhD research. His primary research focus is on human-automation interaction, specifically in how automated vehicles hands back control to the driver in terms of information presentation and cues.
Professor Neville Stanton, PhD, DSc, is a Chartered Psychologist, Chartered Ergonomist and a Chartered Engineer and holds the Chair in Human Factors Engineering in the Faculty of Engineering and the Environment at the University of Southampton. He is leading the EPSRC/JLR funded project on Human Interaction: Designing Automated Vehicles (HI:DAVe) and is a partner in the EU funded Marie Curie International Training Network on Human Factors in Highly Automated Vehicles (HF-Auto).
Abstract
Objective: The aim of this study was to review existing research into driver control transitions and to determine the time it takes drivers to resume control from a highly automated vehicle in non-critical scenarios. Background: Contemporary research has moved from an inclusive design approach to only adhering to mean/median values when designing control transitions in automated driving. Research into control transitions in highly automated driving has focused on urgent scenarios where drivers are given a relatively short time span to respond to a request to resume manual control. We found a paucity in research into more frequent scenarios for control transitions, such as planned exits from highway systems.Method: Twenty six drivers drove two scenarios with an automated driving feature activated. Drivers were asked to read a newspaper, or to monitor the system, and to relinquish, or resume, control from the automation when prompted by vehicle systems.Results: Significantly longer control transition times were found between driving with and without secondary tasks. Control transition times were substantially longer than those reported in the peer-reviewed literature.Conclusion: We found that drivers take longer to resume control when under no time-pressure compared to that reported in the literature. Moreover, we found that drivers occupied by a secondary task exhibit larger variance, and slower responses to requests to resume control. Workload scores implied optimal workload.Application: Intra- and inter-individual differences need to be accommodated by vehicle manufacturers and policy makes alike to ensure inclusive design of contemporary systems and safety during control transitions.
Keywords: Automation, Automated Driving, Control Transitions, Take-Over Requests, Driving Performance, Task Regulation
Prcis: This study reviews the literature for non-critical control transitions in highly automated driving and contrasts the reported results with driver-paced control transitions. The results show increased response times compared to the literature, and when engaged in secondary tasks compared with no task engagement. The study also reports on transition times from manual to automated driving for the first time.
Topic Choice: Surface Transportation
Introduction
Highly automated vehicles are becoming an engineering reality and will become commonplace on our roads in the very near future ADDIN EN.CITE Walker201513(Walker et al., 2015)13136Walker, G. H.,Stanton, N. A. Salmon, P. M. Human Factors in Automotive Engineering and Technology2015Ashgate: Aldershot(Walker et al., 2015). For example, Tesla released its Autopilot feature in 2015, with BMW, Mercedes and Audi quickly following with similar technologies ADDIN EN.CITE safecarnews.com20151364(Audi, 2014; BMW, 2013; safecarnews.com, 2015)1364136412safecarnews.comIntelligent Drive Concept for new Mercedes-Benz GLC2015http://safecarnews.com/intelligent-drive-concept-for-new-mercedes-benz-glc_ju6145/Audi201416616616612AudiAudi piloted driving2014http://www.audi.com/com/brand/en/vorsprung_durch_technik/content/2014/10/piloted-driving.html201623 MayBMW201316716716717BMWConnectedDrive2013http://www.bmw.com/com/en/insights/technology/connecteddrive/2013/active_assist/(Audi, 2014; BMW, 2013; safecarnews.com, 2015). It is a common misconception that these features are highly automated when they are in fact classified as conditional driving automation ADDIN EN.CITE SAE International20163443SAE Level 3`, '(SAE Level 3, 'SAE International, 2016)3443344358SAE International,Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor VehiclesJ3016_2016092016SAE International(SAE Level 3, 'SAE International, 2016). This means that they come with limitations, such as the features may only be intended for use under certain Operational Design Domains, for example, on highways, as well as requiring driver monitoring and intervention ADDIN EN.CITE ADDIN EN.CITE.DATA (Stanton et al., 1997; Wolterink et al., 2011).
When using a driver assistance system that is able to automate the driving task to such an extent that hands- and feet-free driving is possible ADDIN EN.CITE SAE International20163443SAE Level 3`, '(SAE Level 3, 'SAE International, 2016)3443344358SAE International,Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor VehiclesJ3016_2016092016SAE International(SAE Level 3, 'SAE International, 2016), the driver becomes decoupled from the operational and tactical levels of control ADDIN EN.CITE ADDIN EN.CITE.DATA (Michon, 1985; Stanton & Young, 2005), leaving the high level strategic goals to be dealt with by the driver (until the point of resuming manual control). This is a form of driver-initiated automation, where the driver is in control of when the system becomes engaged or disengaged ADDIN EN.CITE ADDIN EN.CITE.DATA (Banks & Stanton, 2015, 2016; Lu & de Winter, 2015). Indeed, according to ADDIN EN.CITE Bainbridge1983388Bainbridge (1983)388388017Bainbridge, L.Bainbridge, L
Univ London Univ Coll,Dept Psychol,London Wc1e 6bt,England
Univ London Univ Coll,Dept Psychol,London Wc1e 6bt,EnglandIronies of AutomationAutomaticaAutomaticaAutomaticaAutomaticaAutomaticaAutomatica775-77919619830005-1098WOS:A1983RY94200025<Go to ISI>://WOS:A1983RY9420002510.1016/0005-1098(83)90046-8EnglishBainbridge (1983), two of the most important tasks for humans in automated systems are monitoring the system to make sure it performs according to expectations and to be ready to resume control when the automation deviates from expectation ADDIN EN.CITE Stanton19961505(Stanton & Marsden, 1996)15051505017Stanton, N. A.Marsden, P.Stanton, NA
Univ Southampton,Dept Psychol,Highfield,Southampton So17 1bj,Hants,England
Univ Southampton,Dept Psychol,Highfield,Southampton So17 1bj,Hants,England
Univ Salford,Dept Math & Comp Sci,Salford M5 4wt,Lancs,EnglandFrom fly-by-wire to drive-by-wire: Safety implications of automation in vehiclesSafety ScienceSafety SciSafety Science35-492411996Oct0925-7535WOS:A1996VY90300003<Go to ISI>://WOS:A1996VY90300003Doi 10.1016/S0925-7535(96)00067-7English(Stanton & Marsden, 1996). Research has shown that vehicle automation has a negative effect on mental workload and situation awareness ADDIN EN.CITE ADDIN EN.CITE.DATA (Endsley & Kaber, 1999; Kaber & Endsley, 1997; Stanton et al., 1997; Stanton & Young, 2005; Young & Stanton, 2002), and that reaction times increase as the level of automation increases ADDIN EN.CITE Young200795(Young & Stanton, 2007)9595017Young, M. S.Stanton, N. A.School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK. m.young@brunel.ac.ukBack to the future: brake reaction times for manual and automated vehiclesErgonomicsErgonomicsErgonomicsErgonomicsErgonomicsErgonomics46-58501Accidents, Traffic/prevention & controlAdultAutomation*Automobile Driving/statistics & numerical dataComputer SimulationDecelerationEquipment DesignHumansMale*Motor Vehicles/statistics & numerical data*Reaction TimeSafety2007Jan 150014-0139 (Print)
0014-0139 (Linking)17178651http://www.ncbi.nlm.nih.gov/pubmed/1717865110.1080/00140130600980789(Young & Stanton, 2007). This becomes problematic when the driver is expected to regain control when system limits are exceeded, as a result of a sudden automation failure. Failure-induced transfer of control has been extensively studied ADDIN EN.CITE ADDIN EN.CITE.DATA (see Desmond et al., 1998; Molloy & Parasuraman, 1996; Stanton et al., 1997; Stanton et al., 2001; Strand et al., 2014; Young & Stanton, 2007). In one failure-induced control-transition-scenario, ADDIN EN.CITE Stanton19971496Stanton et al. (1997)14961496017Stanton, N. A.Young, M.McCaulder, B.Stanton, NA
Univ Southampton, Dept Psychol, Southampton SO17 1BJ, Hants, England
Univ Southampton, Dept Psychol, Southampton SO17 1BJ, Hants, England
Univ Southampton, Dept Psychol, Southampton SO17 1BJ, Hants, EnglandDrive-by-wire: The case of driver workload and reclaiming control with adaptive cruise controlSafety ScienceSafety SciSafety Science149-159272-3automationworkloaddrivingadaptive cruise controlcollisionshuman factors1997Nov-Dec0925-7535WOS:000071257900006<Go to ISI>://WOS:000071257900006Doi 10.1016/S0925-7535(97)00054-4EnglishStanton et al. (1997) found that more than a third of drivers failed to regain control of the vehicle following an automation failure whilst using Adaptive Cruise Control. Other research has shown that it takes approximately one second for a driver manually driving to respond to an unexpected and sudden braking event in traffic ADDIN EN.CITE ADDIN EN.CITE.DATA (Summala, 2000; Swaroop & Rajagopal, 2001; Wolterink et al., 2011). ADDIN EN.CITE Young200795Young and Stanton (2007)9595017Young, M. S.Stanton, N. A.School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK. m.young@brunel.ac.ukBack to the future: brake reaction times for manual and automated vehiclesErgonomicsErgonomicsErgonomicsErgonomicsErgonomicsErgonomics46-58501Accidents, Traffic/prevention & controlAdultAutomation*Automobile Driving/statistics & numerical dataComputer SimulationDecelerationEquipment DesignHumansMale*Motor Vehicles/statistics & numerical data*Reaction TimeSafety2007Jan 150014-0139 (Print)
0014-0139 (Linking)17178651http://www.ncbi.nlm.nih.gov/pubmed/1717865110.1080/00140130600980789Young and Stanton (2007) report brake reaction times of 2.130.55 seconds for drivers using Adaptive Cruise Control (SAE Level 1), and brake reaction times of 2.480.66 seconds for drivers with Adaptive Cruise Control and Assistive Steering (SAE Level 2). By contrasting the results from ADDIN EN.CITE Young200795Young and Stanton (2007)9595017Young, M. S.Stanton, N. A.School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK. m.young@brunel.ac.ukBack to the future: brake reaction times for manual and automated vehiclesErgonomicsErgonomicsErgonomicsErgonomicsErgonomicsErgonomics46-58501Accidents, Traffic/prevention & controlAdultAutomation*Automobile Driving/statistics & numerical dataComputer SimulationDecelerationEquipment DesignHumansMale*Motor Vehicles/statistics & numerical data*Reaction TimeSafety2007Jan 150014-0139 (Print)
0014-0139 (Linking)17178651http://www.ncbi.nlm.nih.gov/pubmed/1717865110.1080/00140130600980789Young and Stanton (2007) where drivers experienced an automation failure whilst a lead vehicle suddenly braked, with ADDIN EN.CITE Summala20001494Summala (2000)14941494017Summala, HeikkiBrake Reaction Times and Driver Behavior AnalysisTransportation Human FactorsTransportation Human Factors217-2262320001093-974110.1207/sthf0203_2Summala (2000) it seems like it takes an additional 1.1-1.5 seconds to react to sudden events requiring braking whilst driving with Driver Assistance Automation (SAE Level 1) and Partial Driving Automation (SAE Level 2). This increase, in combination with headways as short as 0.3 seconds ADDIN EN.CITE Willemsen2015158(Willemsen et al., 2015)158158010Willemsen, DehliaStuiver, ArjanHogema, JeroenAutomated Driving Functions Giving Control Back to the Driver: A Simulator Study on Driver State Dependent Strategies24th International Technical Conference on the Enhanced Safety of Vehicles (ESV)15-01092015(Willemsen et al., 2015) coupled with evidence that drivers are poor monitors ADDIN EN.CITE Molloy19961252(Molloy & Parasuraman, 1996)12521252017Molloy, R.Parasuraman, R.Catholic Univ Amer,Cognit Sci Lab,Washington,Dc 20064Monitoring an automated system for a single failure: Vigilance and task complexity effectsHuman FactorsHum FactorsHum FactorsHuman factorsHum FactorsHuman factors311-322382search1996Jun0018-7208WOS:A1996UZ12800011<Go to ISI>://WOS:A1996UZ1280001110.1518/001872096779048093English(Molloy & Parasuraman, 1996), could actually cause accidents. Evidently, automating the driving task seem to have a detrimental effect on driver reaction time ADDIN EN.CITE Young200795(Young & Stanton, 2007)9595017Young, M. S.Stanton, N. A.School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK. m.young@brunel.ac.ukBack to the future: brake reaction times for manual and automated vehiclesErgonomicsErgonomicsErgonomicsErgonomicsErgonomicsErgonomics46-58501Accidents, Traffic/prevention & controlAdultAutomation*Automobile Driving/statistics & numerical dataComputer SimulationDecelerationEquipment DesignHumansMale*Motor Vehicles/statistics & numerical data*Reaction TimeSafety2007Jan 150014-0139 (Print)
0014-0139 (Linking)17178651http://www.ncbi.nlm.nih.gov/pubmed/1717865110.1080/00140130600980789(Young & Stanton, 2007). Therefore, as ADDIN EN.CITE Cranor2008541Cranor (2008)541541047Lorrie Faith CranorA framework for reasoning about the human in the loopProceedings of the 1st Conference on Usability, Psychology, and Security1-152008San Francisco, CaliforniaUSENIX Association1387650Cranor (2008) and ADDIN EN.CITE Eriksson2016839Eriksson and Stanton (2016)83983947A. ErikssonN. A. StantonThe Chatty Co-Driver: A Linguistics Approach to Human-Automation-InteractionIEHF20162016Daventry, UKEriksson and Stanton (2016) proposed, the driver needs to receive appropriate feedback if they are to successfully re-enter the driving control loop. Recent research efforts have been made to determine the optimal Take-Over-Request lead time (TORlt: the lead-time from a take-over request (TOR) to a critical event, such as a stranded vehicle) and Take Over reaction time (TOrt: the time is takes the driver to take back control of the vehicle from the automated system when a TOR has been issued) with times varying from 0-30 seconds for TORlt and 1.14-15 seconds for TOrt as shown in Table 1. A total of 25 papers reported either TORlt, or TOrt and were included in the review (see REF _Ref450137424 \h Table 1).
Table SEQ Table \* ARABIC 1. Papers included in the review. Modalities for the Take-over request is coded as: A = Auditory, V = Visual, H = Haptic and B = Brake Jerk.
PaperTORltTOrtModality1 ADDIN EN.CITE Gold2016918Gold et al. (2016)918918017Gold, C.Korber, M.Lechner, D.Bengler, K.Institute of Ergonomics, Technical University of Munich, Munich, Germany gold@lfe.mw.tum.de.
Institute of Ergonomics, Technical University of Munich, Munich, Germany.Taking Over Control From Highly Automated Vehicles in Complex Traffic Situations: The Role of Traffic DensityHum FactorsHum FactorsHuman factorsautonomous drivingdriver behaviorhuman-automation interactionmental workloadphoning while drivingvehicle automation2016Mar 161547-8181 (Electronic)
0018-7208 (Linking)26984515http://www.ncbi.nlm.nih.gov/pubmed/26984515http://hfs.sagepub.com/content/early/2016/03/16/0018720816634226.long10.1177/0018720816634226Gold et al. (2016)72.47-3.61-2 ADDIN EN.CITE Krber20151049Krber et al. (2015)10491049017Krber, MoritzWeigerber, ThomasKalb, LuisBlaschke, ChristophFarid, MehdiPrediction of take-over time in highly automated driving by two psychometric testsDynaDyna195-2018219320150012-7353dx.doi.org/10.15446/dyna.v82n193.53496Krber et al. (2015)3-A3 ADDIN EN.CITE Louw20151156Louw et al. (2015b)11561156010Louw, TyronMerat, NatashaJamson, HamishEngaging with Highly Automated Driving: To be or Not to be in the Loop?8th International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design, At Salt Lake City, Utah, USA2015Louw et al. (2015b)6.52.18-2.47A4 ADDIN EN.CITE ADDIN EN.CITE.DATA Zeeb et al. (2016)2.5, 4-V5 ADDIN EN.CITE Dambck201256Dambck et al. (2012)5656017Dambck, DBengler, KFarid, MTnert, Lbernahmezeiten beim hochautomatisierten FahrenTagung Fahrerassistenz. MnchenTagung Fahrerassistenz. Mnchen16152012Dambck et al. (2012)4, 6, 8-A6 ADDIN EN.CITE Kerschbaum2015152Kerschbaum et al. (2015)152152010Kerschbaum, PhilippLorenz, LutzBengler, KlausA transforming steering wheel for highly automated carsIntelligent Vehicles Symposium (IV), 2015 IEEE1287-12922015IEEEKerschbaum et al. (2015)72.22-3.09V A7 ADDIN EN.CITE Belderbos201515Belderbos (2015)1515032Belderbos, CAGAuthority Transition Interface: A Human Machine Interface for Taking over Control from a Highly Automated Truck2015TU Delft, Delft University of TechnologyBelderbos (2015)105.86-5.87V A H8 ADDIN EN.CITE Walch2015155Walch et al. (2015)155155010Walch, MarcelLange, KristinBaumann, MartinWeber, MichaelAutonomous driving: investigating the feasibility of car-driver handover assistanceProceedings of the 7th International Conference on Automotive User Interfaces and Interactive Vehicular Applications11-182015ACM1450337368Walch et al. (2015)4, 61.90-2.75V A9 ADDIN EN.CITE Lorenz2014151Lorenz et al. (2014)151151010Lorenz, LutzKerschbaum, PhilippSchumann, JosefDesigning take over scenarios for automated driving How does augmented reality support the driver to get back into the loop?Proceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting1681-16855812014SAGE Publications1541-9312Lorenz et al. (2014)72.86-3.03V A10 ADDIN EN.CITE Merat20141150Merat et al. (2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006Merat et al. (2014)010-15-11 ADDIN EN.CITE Naujoks2014148Naujoks et al. (2014)148148010Naujoks, FrederikMai, CNekum, AAhram, TKarwowski, WMarek, TThe effect of urgency of take-over requests during highly automated driving under distracted conditions5th International Conference on Applied Human Factors and Ergonomics AHFE 2014201419-23 JulyKrakow, PolandNaujoks et al. (2014)-2.29-6.90V A12 ADDIN EN.CITE Schmig2015154Schmig et al. (2015)154154017Schmig, NadjaHargutt, VolkerNeukum, AlexandraPetermann-Stock, InaOthersen, InaThe interaction between highly automated driving and the development of drowsiness2015Schmig et al. (2015)12-V A 13 ADDIN EN.CITE Louw2015150Louw et al. (2015a)150150017Louw, TLKountouriotis, GCarsten, OMerat, NDriver Inattention During Vehicle Automation: How Does Driver Engagement Affect Resumption Of Control?2015Louw et al. (2015a)3-V A14 ADDIN EN.CITE Zeeb2015157Zeeb et al. (2015)157157017Zeeb, KathrinBuchner, AxelSchrauf, MichaelWhat determines the take-over time? An integrated model approach of driver take-over after automated drivingAccident Analysis & PreventionAccident Analysis & Prevention212-22178Automated drivingEye movementsDriver distractionDriver take-overVisual attentionDriving simulator20155//0001-4575http://www.sciencedirect.com/science/article/pii/S0001457515000731http://dx.doi.org/10.1016/j.aap.2015.02.023Zeeb et al. (2015)2.5, 3, 3.5, 121.14V A15 ADDIN EN.CITE Mok2015149Mok et al. (2015)149149010Mok, BrianJohns, MishelLee, Key JungMiller, DavidSirkin, DavidIve, PageJu, WendyEmergency, Automation Off: Unstructured Transition Timing for Distracted Drivers of Automated VehiclesIntelligent Transportation Systems (ITSC), 2015 IEEE 18th International Conference on2458-24642015IEEE1467365955Mok et al. (2015)2, 5, 8-A16 ADDIN EN.CITE Gold2014145Gold et al. (2014)145145010Gold, ChristianLorenz, LutzBengler, KlausInfluence of automated brake application on take-over situations in highly automated driving scenariosProceedings of the FISITA 2014 World Automotive Congress.(in print)2014Gold et al. (2014)51.67-2.22V A17 ADDIN EN.CITE Radlmayr20141357Radlmayr et al. (2014)13571357017Radlmayr, J.Gold, C.Lorenz, L.Farid, M.Bengler, K.How Traffic Situations and Non-Driving Related Tasks Affect the Take-Over Quality in Highly Automated DrivingProceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting2063-206758120141541-931210.1177/1541931214581434Radlmayr et al. (2014)71.55-2.92V A18 ADDIN EN.CITE Dogan2014144Dogan et al. (2014)144144010Dogan, EbruDeborne, RenaudDelhomme, PatriciaKemeny, AndrasJonville, PatriciaEvaluating the shift of control between driver and vehicle at high automation at low speed: The role of anticipationTransport Research Arena (TRA) 5th Conference: Transport Solutions from Research to Deployment2014Dogan et al. (2014)3-V A19 ADDIN EN.CITE Gold2013959Gold et al. (2013)959959010Gold, ChristianDambck, DanielLorenz, LutzBengler, KlausTake over! How long does it take to get the driver back into the loop?Proceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting1938-19425712013SAGE Publications1541-9312Gold et al. (2013)5, 72.06-3.65V A20 ADDIN EN.CITE van den Beukel2013153van den Beukel and van der Voort (2013)153153010van den Beukel, Arie Pvan der Voort, Mascha CThe influence of time-criticality on Situation Awareness when retrieving human control after automated drivingIntelligent Transportation Systems-(ITSC), 2013 16th International IEEE Conference on2000-20052013IEEEvan den Beukel and van der Voort (2013)1.5, 2.2, 2.8-A21 ADDIN EN.CITE Melcher2015146Melcher et al. (2015)146146017Melcher, VivienRauh, StefanDiederichs, FrederikWidlroither, HaraldBauer, WilhelmTake-Over Requests for Automated DrivingProcedia ManufacturingProcedia Manufacturing2867-28733Highly automated drivingTake-Over RequestsTORReaction timeTransition strategies2015//2351-9789http://www.sciencedirect.com/science/article/pii/S2351978915007891http://dx.doi.org/10.1016/j.promfg.2015.07.788Melcher et al. (2015)103.42-3.77V A B22 ADDIN EN.CITE Naujoks2014147Naujoks and Nekum (2014)147147010Naujoks, FrederikNekum, ADe Waard, DBrookhuis, KWiczorek, RNocera, RBrouwer, RBarham, PWeikert, CKluge, AnnetteGerbino, WTofetti, ATiming of in-vehicle advisory warnings based on cooperative perceptionHuman Factors and Ergonomics Society Europe Chapter Annual Meeting1-132014Naujoks and Nekum (2014)0, 1, 2, 3, 4-V A23 ADDIN EN.CITE Feldhtter2016143Feldhtter et al. (2016)143143047Feldhtter, AGold, C.Schneider, SBengler, KHow the Duration of Automated Driving Influences Take-Over Performance and Gaze BehaviorArbeit in komplexen Systemen - Digital, vernetz, human?! 62. Kongress der Gesellschaft fr Arbeitswissenschaft2016Aachen, GermanyFeldhtter et al. (2016)61.88-2.24A24 ADDIN EN.CITE Payre20161253Payre et al. (2016)12531253017Payre, W.Cestac, J.Delhomme, P.VEDECOM Institute, Versailles, France w.payre@hotmail.fr.
VEDECOM Institute, Versailles, FranceFrench Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.
French Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.Fully Automated Driving: Impact of Trust and Practice on Manual Control RecoveryHum FactorsHum FactorsHuman factors229-41582fully automated drivingmanual control recoverypracticetrust2016Mar1547-8181 (Electronic)
0018-7208 (Linking)26646299http://www.ncbi.nlm.nih.gov/pubmed/2664629910.1177/0018720815612319Payre et al. (2016)2, 304.30-8.70V A25 ADDIN EN.CITE Krber20161082Krber et al. (2016)10821082017Krber, MoritzGold, ChristianLechner, DavidBengler, KlausThe influence of age on the take-over of vehicle control in highly automated drivingTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour19-323920161369847810.1016/j.trf.2016.03.002Krber et al. (2016)72.41-3.66A
The review showed that the mean TORlt was 6.375.36 seconds (Figure 1) with a mean reaction time of 2.961.96 seconds. The most frequently used TORlts tended to be; 3 seconds with a mean TOrt of 1.140.45 [studies 2, 13, 14, 18, 22], 4 seconds with a mean TOrt of 2.050.13 [studies 4, 8, 22], 6 seconds with a mean TOrt of 2.692.21 [studies 5, 8, 23], and 7 seconds with a mean TOrt of 3.041.6 [studies 1, 6, 9, 17, 19, 25] as shown in REF _Ref451171596 \h Figure 2.
Figure SEQ Figure \* ARABIC 1. The TORlt used in the reviewed papers. Several papers used a multitude of TOR lead times and thus contributed on several points of the graph.
Take Over reaction times stay fairly consistent around 2-3.5 seconds in most control transitions, with a few outliers, as seen in REF _Ref451171596 \h Figure 2. ADDIN EN.CITE Belderbos201515Belderbos (2015)1515032Belderbos, CAGAuthority Transition Interface: A Human Machine Interface for Taking over Control from a Highly Automated Truck2015TU Delft, Delft University of TechnologyBelderbos (2015), ADDIN EN.CITE Merat20141150Merat et al. (2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006Merat et al. (2014), ADDIN EN.CITE Naujoks2014148Naujoks et al. (2014)148148010Naujoks, FrederikMai, CNekum, AAhram, TKarwowski, WMarek, TThe effect of urgency of take-over requests during highly automated driving under distracted conditions5th International Conference on Applied Human Factors and Ergonomics AHFE 2014201419-23 JulyKrakow, PolandNaujoks et al. (2014) and ADDIN EN.CITE Payre20161253Payre et al. (2016)12531253017Payre, W.Cestac, J.Delhomme, P.VEDECOM Institute, Versailles, France w.payre@hotmail.fr.
VEDECOM Institute, Versailles, FranceFrench Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.
French Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.Fully Automated Driving: Impact of Trust and Practice on Manual Control RecoveryHum FactorsHum FactorsHuman factors229-41582fully automated drivingmanual control recoverypracticetrust2016Mar1547-8181 (Electronic)
0018-7208 (Linking)26646299http://www.ncbi.nlm.nih.gov/pubmed/2664629910.1177/0018720815612319Payre et al. (2016) show longer TOrt compared to the rest of the reviewed papers. ADDIN EN.CITE Merat20141150Merat et al. (2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006Merat et al. (2014) and ADDIN EN.CITE Naujoks2014148Naujoks et al. (2014)148148010Naujoks, FrederikMai, CNekum, AAhram, TKarwowski, WMarek, TThe effect of urgency of take-over requests during highly automated driving under distracted conditions5th International Conference on Applied Human Factors and Ergonomics AHFE 2014201419-23 JulyKrakow, PolandNaujoks et al. (2014) had the control transition initiated without any lead time whereas ADDIN EN.CITE Belderbos201515Belderbos (2015)1515032Belderbos, CAGAuthority Transition Interface: A Human Machine Interface for Taking over Control from a Highly Automated Truck2015TU Delft, Delft University of TechnologyBelderbos (2015) and ADDIN EN.CITE Payre20161253Payre et al. (2016)12531253017Payre, W.Cestac, J.Delhomme, P.VEDECOM Institute, Versailles, France w.payre@hotmail.fr.
VEDECOM Institute, Versailles, FranceFrench Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.
French Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.Fully Automated Driving: Impact of Trust and Practice on Manual Control RecoveryHum FactorsHum FactorsHuman factors229-41582fully automated drivingmanual control recoverypracticetrust2016Mar1547-8181 (Electronic)
0018-7208 (Linking)26646299http://www.ncbi.nlm.nih.gov/pubmed/2664629910.1177/0018720815612319Payre et al. (2016) did. ADDIN EN.CITE Merat20141150Merat et al. (2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006Merat et al. (2014) showed that there is a 10-15 second time lag between the disengagement of the automated driving system and resumption of control by the driver. Notably, the control transition was system initiated and lacked a pre-emptive TOR which may have caused the increase in TOrt. Similarly, ADDIN EN.CITE Naujoks2014148Naujoks et al. (2014)148148010Naujoks, FrederikMai, CNekum, AAhram, TKarwowski, WMarek, TThe effect of urgency of take-over requests during highly automated driving under distracted conditions5th International Conference on Applied Human Factors and Ergonomics AHFE 2014201419-23 JulyKrakow, PolandNaujoks et al. (2014) observed a 6.9 second TOrt from when a TOR was issued and the automation disconnected until the driver resumed control in situations where automation became unavailable due to missing line markings, the beginning of a work zone or entering a curve. Based on personal communication with the author, the vehicle would have crossed the lane markings after approximately 13 seconds, and would have reached the faded lane markings approximately 10 seconds after the TOR. The velocity in ADDIN EN.CITE Naujoks2014148Naujoks et al. (2014)148148010Naujoks, FrederikMai, CNekum, AAhram, TKarwowski, WMarek, TThe effect of urgency of take-over requests during highly automated driving under distracted conditions5th International Conference on Applied Human Factors and Ergonomics AHFE 2014201419-23 JulyKrakow, PolandNaujoks et al. (2014) was 50 Kph, which is fairly slow compared to most other TOR studies that use speeds over 100 Kph [studies 1, 3, 4, 6, 9, 10, 12, 14, 17, 19, 21, 23, 24, 25] and may have had an effect on the perceived urgency.
Figure SEQ Figure \* ARABIC 2. Take Over reaction time averages for all the conditions in the reviewed studies. Some studies had more than one take over event and is therefore featured multiple times.
ADDIN EN.CITE Belderbos201515Belderbos (2015)1515032Belderbos, CAGAuthority Transition Interface: A Human Machine Interface for Taking over Control from a Highly Automated Truck2015TU Delft, Delft University of TechnologyBelderbos (2015) showed TOrts of 5.861.57 to 5.874.01 when drivers were given a TORlt of 10 seconds during unsupervised automated driving. ADDIN EN.CITE Payre20161253Payre et al. (2016)12531253017Payre, W.Cestac, J.Delhomme, P.VEDECOM Institute, Versailles, France w.payre@hotmail.fr.
VEDECOM Institute, Versailles, FranceFrench Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.
French Institute of Science and Technology for Transportation, Development, and Networks, Versailles, France.Fully Automated Driving: Impact of Trust and Practice on Manual Control RecoveryHum FactorsHum FactorsHuman factors229-41582fully automated drivingmanual control recoverypracticetrust2016Mar1547-8181 (Electronic)
0018-7208 (Linking)26646299http://www.ncbi.nlm.nih.gov/pubmed/2664629910.1177/0018720815612319Payre et al. (2016) utilised two different TORlts, 2, and 30 seconds. These TORlts produced significant differences in TOrt, the 2 second TORlt produced a TOrt of 4.31.2 seconds and the two scenarios that used the 30 second TORlt produced TOrts of 8.72.7 seconds and 6.82.5 seconds respectively. The shorter TOrt of the two 30 second TOR events occurred after the 2 second emergency TOR and could have been affected by the urgency caused by the short lead time in the preceding, shorter TOR.
ADDIN EN.CITE Merat20141150Merat et al. (2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006Merat et al. (2014) concluded, based on their observed TOrt, that there is a need for a timely and appropriate notification of an imminent control transition. This observation is in line with the current SAE-guidelines which state that the driver Is receptive to a request to intervene and responds by performing dynamic driving task fallback in a timely manner ADDIN EN.CITE SAE International2016344320(SAE International, 2016, p. 20)3443344358SAE International,Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor VehiclesJ3016_2016092016SAE International(SAE International, 2016, p. 20). In initial efforts to determine how long in advance the driver needs to be notified before a control transition is initiated, ADDIN EN.CITE Dambck201256Dambck et al. (2012)5656017Dambck, DBengler, KFarid, MTnert, Lbernahmezeiten beim hochautomatisierten FahrenTagung Fahrerassistenz. MnchenTagung Fahrerassistenz. Mnchen16152012Dambck et al. (2012) and ADDIN EN.CITE Gold2013959Gold et al. (2013)959959010Gold, ChristianDambck, DanielLorenz, LutzBengler, KlausTake over! How long does it take to get the driver back into the loop?Proceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting1938-19425712013SAGE Publications1541-9312Gold et al. (2013) explored a set of TOR lead times. ADDIN EN.CITE Dambck201256Dambck et al. (2012)5656017Dambck, DBengler, KFarid, MTnert, Lbernahmezeiten beim hochautomatisierten FahrenTagung Fahrerassistenz. MnchenTagung Fahrerassistenz. Mnchen16152012Dambck et al. (2012) utilised three TORlts, 4, 6, and 8 seconds and found that given an 8 second lead time, drivers did not differ significantly from manual driving. This was confirmed by ADDIN EN.CITE Gold2013959Gold et al. (2013)959959010Gold, ChristianDambck, DanielLorenz, LutzBengler, KlausTake over! How long does it take to get the driver back into the loop?Proceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting1938-19425712013SAGE Publications1541-9312Gold et al. (2013) who reported that drivers need to be warned at least 7 seconds in advance of a control transition to safely resume control. These findings seem to have been the inspiration for the TORlt of some recent work utilising timings around 7 seconds [studies 1, 6, 9, 17].
A caveat of a number of the reviewed studies is that the lead time given in certain scenarios such as; disappearing lane markings, construction zones, and merging motorway lanes is surprisingly short, from 0 to 12 seconds (c.f. Table 1), and will likely be longer in on road use cases [studies 4, 5, 11, 14, 15, 21]. The reason for this is the increasing accuracy of contemporary GPS hardware and associated services, such as Google Maps. Such services are already able to direct lane positioning whilst driving manually, as well as notifying drivers of construction zones and alternate, faster routes. Thus, there is no evident gain of having short lead times in such situations.
Several of the studies reviewed have explored the effect of TORs in different critical settings by issuing the TOR immediately preceding a time critical event [studies 1, 2, 3, 4, 6, 7, 8, 9, 13, 16, 17, 19, 20, 23, 24, 25]. These studies have explored how drivers manage critical situations in terms of driving behaviour, workload, and scanning behaviour. Whilst it is of utmost importance to know how quickly a driver can respond to a TOR and what the shortest TOR-times are in emergencies, there is a paucity of research exploring the time it takes a driver to resume control in normal, non-critical, situations. We argue that if the design of normal, non-critical, control transitions are designed based on data obtained in studies utilising critical situations, there is a risk of unwanted consequences such as: drivers not responding optimally due to too short lead time (suboptimal responses are acceptable in emergencies as drivers are tasked with avoiding danger), drivers being unable to fully regain situation awareness, and sudden, dramatic, increases in workload. Arguably, these consequences should not be present in every transition of control as it poses a safety risk for the driver, as well as other road users. Therefore, the aim of this study is to establish driver take-over time in normal traffic situations when, e.g. the vehicle is leaving its Operational Design Domain as these will account for most of the situations ADDIN EN.CITE Nilsson2014224(Nilsson, 2014; SAE International, 2016)2242246Nilsson, JosefSafe Transitions to Manual Driving From Faulty Automated Driving System2014Chalmers University of Technology917597035XSAE International201634433443344358SAE International,Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor VehiclesJ3016_2016092016SAE International(Nilsson, 2014; SAE International, 2016). We also explore how TOR take-over time is affected by a non-driving secondary task, as this was expected to increase the reaction time ADDIN EN.CITE Merat2012223(Merat et al., 2012)22322317Merat, NatashaJamson, A HamishLai, Frank CHCarsten, OliverHighly automated driving, secondary task performance, and driver stateHuman Factors: The Journal of the Human Factors and Ergonomics SocietyHuman Factors: The Journal of the Human Factors and Ergonomics Society762-77154520120018-7208(Merat et al., 2012).
Moreover, none of the papers included in the review mentioned the time it takes drivers to transition from manual to automated driving. Gaining an understanding on the time required to toggle an automated driving system on is important in situations such as: entering an area dedicated to automated vehicles or engaging the automated driving mode in preparation for joining a platoon as proposed by the SARTRE project ADDIN EN.CITE Robinson2010221(Robinson et al., 2010)22122110Robinson, TomChan, EricCoelingh, ErikOperating platoons on public motorways: An introduction to the sartre platooning programme17th world congress on intelligent transport systems1212010(Robinson et al., 2010). Therefore, the aim of this study was to establish the time it takes a driver to switch to automated driving when automated driving features become available. Ultimately, this research aims to provide guidance about the lead-time required to get the driver back into, and out of, the manual vehicle control loop.
Method
Participants
Twenty-six participants (10 females, 16 males) between 20 and 52 years of age (M = 30.27 SD = 8.52) with a minimum one year and an average 10.57 years (SD = 8.61) of driving experience were asked to take part in the trial. Upon recruiting participants, their informed consent was obtained. The study complied with the American Psychological Association Code of Ethics and had been approved by the University of Southampton Ethics Research and Governance Office (ERGO number 17771).
Equipment
The experiment was carried out in a fixed based driving simulator located at the University of Southampton. The simulator was a Jaguar XJ 350 with pedal and steering sensors provided by Systems Technology Inc. as part of STISIM Drive M500W Version 3 (http://www.stisimdrive.com/m500w) providing a projected 140 field of view. Rear view- and side-mirrors were provided through additional projectors and video cameras. The original Jaguar XJ instrument cluster was replaced with a 10.6 Sharp LQ106K1LA01B Laptop LCD panel connected to the computer via a RTMC1B LCD controller board to display computer generated graphics components for TORs. The default configuration of the instrument cluster is shown in REF _Ref448915354 \h Figure 3.
Figure SEQ Figure \* ARABIC 3. The instrument cluster in its default configuration
When a TOR was issued the engine speed dial was hidden and the request was shown in its place. The symbol asking for control resumption is shown in REF _Ref451760089 \h \* MERGEFORMAT Figure 5 and the symbol used to prompt the driver to re-engage the automation is shown in REF _Ref451759957 \h \* MERGEFORMAT Figure 4.
Figure SEQ Figure \* ARABIC 4. The icon shown when the automation becomes available. The icon was coupled with a computer generated voice message stating automation available.
Figure SEQ Figure \* ARABIC 5. The take-over request icon shown on the instrument cluster. The icon was coupled with a computer generated voice message stating "please resume control.
The mode switching human machine interface was located on a Windows tablet in the centre console, consisting of two buttons, used either to engage, or to disengage the automation. To enable dynamic dis-engagement and re-engagement of the automation, bespoke algorithms were developed and are reported elsewhere ADDIN EN.CITE Eriksson2016825c.f. (c.f. Eriksson et al., 2016)82582547Eriksson, A.Spychala, A.De Winter J C. F.Stanton, N. A.A Highly Automated Driving Algorithm Implementation on STISIMAHFE 20162016Florida, OL10.13140/RG.2.1.3226.9203(c.f. Eriksson et al., 2016).
Experiment Design
The experiment had a repeated-measures, within-subject design with three conditions; Manual, Highly Automated Driving (HAD) and Highly Automated Driving with a secondary task. The conditions were counterbalanced to counteract order effects. For the automated conditions, participants drove at 70 mph on a 30 kilometre, three lane highway with some curves, with oncoming traffic in the opposing three lanes separated by a barrier and moderate traffic conditions. The route was mirrored between the two automated conditions to reduce familiarity effects whilst keeping the roadway layout consistent.
In the secondary task condition, drivers were asked to read (in their head) an issue of National Geographic whilst the automated driving system was engaged in order to remove them from the driving (and monitoring) task. During both conditions, drivers were prompted to either resume control from, or relinquish control to, the automated driving system. The control transition requests were presented as both a visual cue (c.f. REF _Ref451759957 \h \* MERGEFORMAT Figure 4 and REF _Ref451760089 \h \* MERGEFORMAT Figure 5) and an auditory message, in line with previous research [studies 6,7,8,9,11,12,13,14,16,17,18,19,22,24], in the form of a computer generated, female voice stating please resume control or automation available. No haptic feedback was included in this study, despite the findings from ADDIN EN.CITE Petermeijer2016215Petermeijer et al. (2016)215215017Petermeijer, Sebastiaan M.de Winter, Joost C. F.Bengler, Klaus J.Vibrotactile Displays: A Survey With a View on Highly Automated DrivingIEEE Transactions on Intelligent Transportation SystemsIEEE Transactions on Intelligent Transportation Systems897-90717420161524-9050
1558-001610.1109/tits.2015.2494873Petermeijer et al. (2016) and ADDIN EN.CITE Scott2008218Scott and Gray (2008)21821817Scott, JJGray, RobertA comparison of tactile, visual, and auditory warnings for rear-end collision prevention in simulated drivingHuman Factors: The Journal of the Human Factors and Ergonomics SocietyHuman Factors: The Journal of the Human Factors and Ergonomics Society264-27550220080018-7208Scott and Gray (2008) showing shorter reaction times when vibrotactile feedback was used. The motivation for excluding the haptic modality was that it was under-represented in the review, with only 1 paper in the review utilising a form of haptic feedback. Furthermore, ADDIN EN.CITE Petermeijer2016215Petermeijer et al. (2016)215215017Petermeijer, Sebastiaan M.de Winter, Joost C. F.Bengler, Klaus J.Vibrotactile Displays: A Survey With a View on Highly Automated DrivingIEEE Transactions on Intelligent Transportation SystemsIEEE Transactions on Intelligent Transportation Systems897-90717420161524-9050
1558-001610.1109/tits.2015.2494873Petermeijer et al. (2016) concluded that haptic feedback is best suited for warnings, and as the current experimental design explored non-critical warnings, no motivation for including haptics could be found. The interval in which these requests were issued ranged from 30-45 seconds, thus allowing for approximately 24 control transitions of which half were to manual control.
Procedure
Upon arrival, participants were asked to read an information sheet, containing information regarding the study, the right to at any point abort their trial without any questions asked. After reading the information sheet the participants were asked to sign an informed consent form. They were also told that they were able to override any system inputs via the steering wheel, throttle or brake pedals. Drivers were reminded that they were responsible for the safe operation of the vehicle regardless of its mode (manual or automated), and thus needed to be able to safely resume control in case of failure. This is in accordance with current legislation ADDIN EN.CITE United Nations19681487(United Nations, 1968)14871487010United Nations,United NationsConvention on road trafficAmendment 11968Done at Vienna on 8 November 1968. Amendment 1. Retrieved from http://www.unece.org/fileadmin/DAM/trans/conventn/crt1968e.pdf.http://www.unece.org/fileadmin/DAM/trans/conventn/crt1968e.pdfhttps://treaties.un.org/doc/Publication/UNTS/Volume%201042/v1042.pdf(United Nations, 1968) and recent amendments to the Vienna Convention of Road Traffic. They were informed that the system may prompt them to either resume or relinquish control of the vehicle, and that when such a prompt was issued they were required to adhere to the instruction, but only when they felt safe doing so. This instruction was intended to reduce the pressure on drivers to respond immediately and to reinforce the idea that they were ultimately responsible for safe vehicle operation.
At the end of each driving condition, participants were asked to fill out the NASA-RTLX ADDIN EN.CITE Byers1989519(Byers et al., 1989)51951917Byers, James CBittner, ACHill, SGTraditional and raw task load index (TLX) correlations: Are paired comparisons necessaryAdvances in industrial ergonomics and safety IAdvances in industrial ergonomics and safety I481-4851989(Byers et al., 1989). They were also offered a short break before continuing the study. Reaction time data were logged for each transition to and from manual control.
Dependent variables
The following metrics were collected for each condition per participant.
Reaction time to the control transition request was recorded from the onset of the TOR. The control transition request was presented in the instrument cluster coupled with a computer generated voice to initiate a change in mode to and from manual control and was recorded in milliseconds.
Driving performance as measured by Standard Deviation of Steering Angular rate (degrees / second).
Subjective workload scores were collected via the NASA-TLX sub-scales at the end of each driving condition. Overall workload score was calculated through the summation of sub-scales ADDIN EN.CITE Byers1989519(Byers et al., 1989; Hart & Staveland, 1988)51951917Byers, James CBittner, ACHill, SGTraditional and raw task load index (TLX) correlations: Are paired comparisons necessaryAdvances in industrial ergonomics and safety IAdvances in industrial ergonomics and safety I481-4851989Hart198822482248224817Hart, Sandra GStaveland, Lowell EDevelopment of NASA-TLX (Task Load Index): Results of empirical and theoretical researchAdvances in psychologyAdvances in psychology139-1835219880166-4115(Byers et al., 1989; Hart & Staveland, 1988).
Analysis
The dependent measures were tested for normal distribution using the Kolmogorov-Smirnov test, which revealed that the data was non-normally distributed. To assess driving performance after control was handed back to the driver, a measure of the standard deviation of the absolute steering angular rate was used to capture corrective steering actions ADDIN EN.CITE Fisher20112268 ch 40`, pp 10(Fisher et al., 2011 ch 40, pp 10)226822686Fisher, Donald LRizzo, MatthewCaird, JeffreyLee, John DHandbook of driving simulation for engineering, medicine, and psychology2011CRC Press1420061011(Fisher et al., 2011 ch 40, pp 10). Furthermore, as the TOrt data is reaction time data, the median TOrt values for each participant was calculated after which Wilcoxon signed-rank test was used to analyse the time and workload data. The box plots in REF _Ref455412101 \h Figure 6 and REF _Ref455412121 \h Figure 8 had their outlier thresholds adjusted to accommodate the log-normal distribution of the TOrt data by using the LIBRA library for MatLab ADDIN EN.CITE Verboven2005165(Verboven & Hubert, 2005)165165017Verboven, SabineHubert, MiaLIBRA: a MATLAB library for robust analysisChemometrics and intelligent laboratory systemsChemometrics and intelligent laboratory systems127-13675220050169-7439(Verboven & Hubert, 2005) and its method for robust boxplots for non-normally distributed data by ADDIN EN.CITE Hubert2008164Hubert and Vandervieren (2008)164164017Hubert, MiaVandervieren, EllenAn adjusted boxplot for skewed distributionsComputational statistics & data analysisComputational statistics & data analysis5186-5201521220080167-9473Hubert and Vandervieren (2008). Effect sizes were calculated as: r = a b s ( Z / "N ) .
R e s u l t s
T h e r e s u l t s s h o w e d t h a t i t t o o k a p p r o x i m a t e l y 4 . 2 - 4 . 4 1 . 9 6 - 1 . 8 0 s e c o n d s ( m e d i a n ) t o s w i t c h t o a u t o m a t e d d r i v i n g , s e e R E F _ R e f 4 5 1 7 5 9 4 9 7 \ h \ * M E R G E F O R M A T T a b l e 2 . N o s i g n i f i c a n t d i f f e r e n c e s b e t w e e n t h e t w o c o n d i t i o n s c o u l d b e found when drivers transitioned from manual to automated driving (Z = -0.673, p = 0.5, r = 0.13). Control transition times from manual to automated driving in the two conditions is shown in REF _Ref455412101 \h Figure 6 and individual transition times for each participant are available in the supplementary material.
Figure SEQ Figure \* ARABIC 6. Adjusted box-plot of control transition times from manual driving to Automated Driving. The dashed horizontal line indicates the max/min values assuming a normal distribution.
Figure SEQ Figure \* ARABIC 7. A distribution plot of TOrt when drivers were prompted to engage the automation. The asterisk* marks the median value, the X axis contains 160 bins.
The results showed a significant increase in control transition time of ~1.5 seconds when drivers were prompted to resume control whilst engaged in a secondary task (Z = -4.43, p < 0.01, r = 0.86). It took drivers approximately 4.461.63 seconds to resume control when not occupied by a secondary task, and 6.062.39 seconds to resume control when engaged in a secondary task as shown in REF _Ref451759497 \h \* MERGEFORMAT Table 2.
Table SEQ Table \* ARABIC 2. Descriptive statistics of the control transition times (in milliseconds) from Automated Driving to manual control, and from manual control to Automated Driving as well as descriptive statistics from the presented TOrts from the reviewed articles.
Meta-reviewFrom Automated to ManualFrom Manual to AutomatedNo secondary taskSecondary taskNo secondary taskSecondary taskMedian2470 ms4567 ms6061 ms4200 ms4408 msIQR1415 ms1632 ms2393 ms1964 ms1800 msMin1140 ms1975 ms3179 ms2822 ms2926 msMax15000 ms25750 ms20994 ms23884 ms23221 ms
Figure SEQ Figure \* ARABIC 8. Adjusted boxplot of the Take Over reaction time when switching from automated to manual control in the two experimental conditions contrasted with the TOrt of the reviewed papers.
Figure SEQ Figure \* ARABIC 9 A distribution plot of TOrt when drivers were prompted to resume manual control. The asterisk* marks the median value, the X axis contains 160 bins. The amplitude of the reviewed papers is caused by the low number of values provided by the reviewed papers.
The results from analysing the driving performance data from 0 to 18 seconds post control transition showed non-significant differences in Standard Deviation of Absolute Steering Angular Rate (Table 3) between the two task conditions.
Table SEQ Table \* ARABIC 3. Standard deviation of Angular Rate (Degrees/second) for the two task conditions from 0-18 seconds post take-over.
No secondary taskWith secondary tasktimeM (SD)M (SD)Zpr0-3s0.08 (0.06)0.08 (0.06)-0.110.910.023-6s0.03 (0.03)0.03 (0.03)-0.140.890.036-9s0.02 (0.02)0.06 (0.17)-1.480.130.299-12s0.02 (0.02)0.11 (0.44)-0.980.330.1912-15s0.02 (0.02)0.11 (0.45)-0.390.690.0815-18s0.03 (0.07)0.13 (0.51)-0.420.670.08
The analysis of the subjective ratings for driver mental workload showed that the secondary task condition has marginally higher scores overall, as shown in Table 4. Only temporal demand had a statistically significant difference (Z= -3.11, p < 0.05, r = 0.61), with higher rated demand in the secondary task condition as shown in REF _Ref448744445 \h Figure 10.
Table SEQ Table \* ARABIC 4. Overall workload scores as well as individual workload ratings for the two conditions. ** = Significant at the 0.01 level.
Without secondary taskWith secondary taskVariableMedian (IQR)Median (IQR)ZprOverall Workload5 (7.33)6.2 (6.5)-1.9530.0510.38Mental demand7.5 (10)10.5 (9)-1.410.160.28Physical demand4 (5)5.5 (6)-1.930.0540.38Temporal demand3 (6)8.5 (8)-3.110.00**0.61Performance6 (7)4.5 (5)-0.470.630.09Effort5 (9)7 (8)-0.230.820.05Frustration4 (9)6.5 (7)-1.040.30.2
Figure SEQ Figure \* ARABIC 10. Boxplot of Subjective estimations of workload in the two conditions.
Discussion
Relinquishing control to automation
In this study we subjected drivers to multiple control transitions between manual and automated control in a highway scenario. Upon reviewing the literature, no mention of how long the driver takes to engage an automated driving system was found, making this study a first-of-a-kind. We found that drivers take between 2.82-23.8 seconds (Median = 4.2-4.4) to engage automated driving when the system indicates that the feature is available. No significant differences between the two conditions was found, but as REF _Ref455412101 \h \* MERGEFORMAT Figure 6 shows, there was large range in the time it takes to relinquish control. It is clear from REF _Ref454285905 \h \* MERGEFORMAT Figure 7 that designing for the median, or average driver effectively exclude a large part of the user group, which could have severe implications for drivers who fall outside of the mean or median. It has been common practice in Human Factors and Anthropometrics to design for 90% of the population, normally through accommodating the range between the 5th percentile female and the 95th percentile male ADDIN EN.CITE Porter2004192(Porter et al., 2004)192192017Porter, J. MarkCase, KeithMarshall, RussellGyi, DianeSims ne Oliver, RuthBeyond Jack and Jill: designing for individuals using HADRIANInternational Journal of Industrial ErgonomicsInternational Journal of Industrial Ergonomics249-26433320040169814110.1016/j.ergon.2003.08.002(Porter et al., 2004). Thus, it is important that vehicle manufacturers are made aware of the intra-individual differences, as such differences have a large effect on the larger traffic system if drivers are expected to toggle Automated Driving systems within a certain time frame. An example of potential situations where the driver would need to toggle the automated driving system could be in HAD-dedicated areas. Moreover, it may be that the time it takes to engage the automated driving system depends on external factors such as, perceived safety, weather conditions, traffic flow rates, presence of vulnerable road users, roadworks, and so on. If the driver deems a situation unsafe, or has doubts as to how well the automation would perform in a situation, the driver may hold off on completing a transition until the driver feels that the system can comfortably handle the situation.
Resuming control from automation
Previous research was reviewed and it was found that most studies utilised system paced transitions, where the automated driving system warns in advance of failure or reduced automation support with relatively short lead times, from 3 seconds [studies 2, 13, 14, 18, 22] to 7 seconds [studies 1, 6, 9, 17, 19, 25]. It has previously been shown that whilst it takes approximately 2.471.42 seconds on average, it can take up to 15 seconds to respond to such an event ADDIN EN.CITE Merat20141150(Merat et al., 2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006(Merat et al., 2014). We argue that this use case, albeit important, does not reflect the primary use case for control transitions in highly automated driving. When comparing the range of TOrt in the literature to the user paced (no secondary task) condition in this study, a great deal of overlap can be seen. The observed values in the current study are closer to the higher values observed by ADDIN EN.CITE Merat20141150Merat et al. (2014)11501150017Merat, NatashaJamson, A. HamishLai, F. Frank C. H.Daly, MichaelCarsten, Oliver M. J.Transition to manual: Driver behaviour when resuming control from a highly automated vehicleTransportation Research Part F: Traffic Psychology and BehaviourTransportation Research Part F: Traffic Psychology and Behaviour1-926, Part AHighly automated drivingLevel 3 automationDriver behaviourEye trackingTransition time20141369-8478http://www.sciencedirect.com/science/article/pii/S1369847814000722http://ac.els-cdn.com/S1369847814000722/1-s2.0-S1369847814000722-main.pdf?_tid=22fc304a-38f2-11e4-a615-00000aab0f6c&acdnat=1410357514_7b34a0150d2100cede167c8226cddc10http://dx.doi.org/10.1016/j.trf.2014.05.006Merat et al. (2014) whilst the median range of 4.56-6.06s is closer to the range of times suggested by ADDIN EN.CITE Gold2013959Gold et al. (2013)959959010Gold, ChristianDambck, DanielLorenz, LutzBengler, KlausTake over! How long does it take to get the driver back into the loop?Proceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting1938-19425712013SAGE Publications1541-9312Gold et al. (2013) and ADDIN EN.CITE Dambck201256Dambck et al. (2012)5656017Dambck, DBengler, KFarid, MTnert, Lbernahmezeiten beim hochautomatisierten FahrenTagung Fahrerassistenz. MnchenTagung Fahrerassistenz. Mnchen16152012Dambck et al. (2012). It is evident that there is a large spread in the TOrt, which when designing driving automation should be considered, as the range of performance is more important than the median or mean, as these exclude a large portion of drivers.
When subjecting drivers to TORs without time restrictions we found that drivers take between 1.97-25.75 seconds (Median = 4.56) to resume control from automated driving in normal conditions, and between 3.17-20.99 seconds (Median = 6.06) to do so whilst engaged in a secondary task preceding the control transition. This shows that there is a median 2 second difference in control transition times in the reviewed manuscripts compared to the user paced control transitions. There was a large effect of secondary task engagement on TOrt, showing an increase in driver control resumption times when engaged in a secondary task. This might be explained in part by the nature of the secondary task, as the driver had to allocate time to put down the magazine they were asked to read whilst the automated driving feature was activated. It could also be partly attributed to driver task adaptation, by holding off transferring control until they have had time to switch between the reading task and driving task. This is supported by research indicating that drivers tend to adapt to external factors such as traffic complexity to allow for more time to make decisions ADDIN EN.CITE Eriksson2014865(Eriksson et al., 2014)8658655Eriksson, AlexanderLindstrm, AndersSeward, AlbertSeward, AlexanderKircher, KatjaCan user-paced, menu-free spoken language interfaces improve dual task handling while driving?Human-Computer Interaction. Advanced Interaction Modalities and Techniques394-4052014Springer3319072293(Eriksson et al., 2014), by for example slowing down when engaged in secondary tasks ADDIN EN.CITE Cooper2009496(Cooper et al., 2009)49649617Cooper, Joel MVladisavljevic, IvanaMedeiros-Ward, NathanMartin, Peter TStrayer, David LAn investigation of driver distraction near the tipping point of traffic flow stabilityHuman Factors: The Journal of the Human Factors and Ergonomics SocietyHuman Factors: The Journal of the Human Factors and Ergonomics Society261-26851220090018-7208(Cooper et al., 2009) or the expectation of resuming control ADDIN EN.CITE Young200795(Young & Stanton, 2007)9595017Young, M. S.Stanton, N. A.School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK. m.young@brunel.ac.ukBack to the future: brake reaction times for manual and automated vehiclesErgonomicsErgonomicsErgonomicsErgonomicsErgonomicsErgonomics46-58501Accidents, Traffic/prevention & controlAdultAutomation*Automobile Driving/statistics & numerical dataComputer SimulationDecelerationEquipment DesignHumansMale*Motor Vehicles/statistics & numerical data*Reaction TimeSafety2007Jan 150014-0139 (Print)
0014-0139 (Linking)17178651http://www.ncbi.nlm.nih.gov/pubmed/1717865110.1080/00140130600980789(Young & Stanton, 2007). In light of these results, there is a case for adaptive automation that modulates TORlt by, for example, detecting whether the driver gaze is off road for a certain time-period, providing the driver with a few additional seconds before resuming control.
Furthermore, the 1.5 second increase of control resumption time when engaged in a reading task is similar to the reaction time increase caused by the introduction of automated driving features observed by ADDIN EN.CITE Young200795Young and Stanton (2007)9595017Young, M. S.Stanton, N. A.School of Engineering and Design, Brunel University, Uxbridge, Middlesex, UB8 3PH, UK. m.young@brunel.ac.ukBack to the future: brake reaction times for manual and automated vehiclesErgonomicsErgonomicsErgonomicsErgonomicsErgonomicsErgonomics46-58501Accidents, Traffic/prevention & controlAdultAutomation*Automobile Driving/statistics & numerical dataComputer SimulationDecelerationEquipment DesignHumansMale*Motor Vehicles/statistics & numerical data*Reaction TimeSafety2007Jan 150014-0139 (Print)
0014-0139 (Linking)17178651http://www.ncbi.nlm.nih.gov/pubmed/1717865110.1080/00140130600980789Young and Stanton (2007) compared to reaction time in manual driving ADDIN EN.CITE Summala20001494(Summala, 2000)14941494017Summala, HeikkiBrake Reaction Times and Driver Behavior AnalysisTransportation Human FactorsTransportation Human Factors217-2262320001093-974110.1207/sthf0203_2(Summala, 2000). Therefore, a further increase of reaction times when drivers are engaged in other tasks will have to be expected, but measures must be taken to reduce the increase in reaction time, for example through the addition of informative displays, to reduce the risk of accidents ADDIN EN.CITE Eriksson2016839(Cranor, 2008; Eriksson & Stanton, 2016)83983947A. ErikssonN. A. StantonThe Chatty Co-Driver: A Linguistics Approach to Human-Automation-InteractionIEHF20162016Daventry, UKCranor2008541541541047Lorrie Faith CranorA framework for reasoning about the human in the loopProceedings of the 1st Conference on Usability, Psychology, and Security1-152008San Francisco, CaliforniaUSENIX Association1387650(Cranor, 2008; Eriksson & Stanton, 2016).
There was a significant increase in perceived temporal demand when drivers were tasked with reading whilst the automation was engaged. This increase in perceived temporal demand may have been caused by the TOR, and the driver not being fully sure as to how long the vehicle could manage before a forced transition would occur (this was not a possibility in the current experiment). This increase in perceived temporal demand could also be attributed to the pace of the experiment, and that the drivers were required to pick up, and put down the magazine whenever a control transition was issued. Overall there are little differences in workload, and the median workload in both conditions was approximately at the halfway point on the scale, implying optimal loading ADDIN EN.CITE Stanton20111488(Stanton et al., 2011)14881488017Stanton, N. A.Dunoyer, A.Leatherland, A.Transportation Research Group, School of Civil Engineering, University of Southampton, Highfield, Southampton, UK. n.stanton@soton.ac.ukDetection of new in-path targets by drivers using Stop & Go Adaptive Cruise ControlAppl ErgonApplied ergonomicsAppl ErgonApplied Ergonomics592-601424AdultAttention/*physiologyAutomation/*instrumentation/methodsAutomobile Driving/*psychology*Automobiles*AwarenessCognitionFemaleHumansMalePsychometricsStatistics, NonparametricTask Performance and AnalysisWorkload/psychologyYoung Adult2011May1872-9126 (Electronic)
0003-6870 (Linking)20870216http://www.ncbi.nlm.nih.gov/pubmed/2087021610.1016/j.apergo.2010.08.016(Stanton et al., 2011).
In light of these results, combined with the non-significant, small effect of task condition on driving performance indicated that the drivers were able to self-regulate the control transition process by adapting the time needed to resume control ADDIN EN.CITE ADDIN EN.CITE.DATA (Eriksson et al., 2014; Kircher et al., 2014) and therefore maintaining optimal levels of workload, minimizing the severity of the after-effects observed in studies by e.g. ADDIN EN.CITE Gold2013959Gold et al. (2013)959959010Gold, ChristianDambck, DanielLorenz, LutzBengler, KlausTake over! How long does it take to get the driver back into the loop?Proceedings of the Human Factors and Ergonomics Society Annual MeetingProceedings of the Human Factors and Ergonomics Society Annual Meeting1938-19425712013SAGE Publications1541-9312Gold et al. (2013).
Conclusions
Relinquishing control to automation
The literature on control transitions in highly automated driving is absent in research reports on transitions from manual to automated vehicle control. In a first-of-a-kind study, we found that it takes drivers between 2.8 and 23.8 seconds to switch from manual to automated control. This finding has some implications for the safety of drivers merging into automated driving dedicated lanes, or other infrastructure whilst in manual mode. Such an event may require certain adaptations for traffic already occupying such a lane. Adaptations may include increasing time-headway or reducing speed to accommodate the natural variance in human behaviour to avoid collisions, or discomfort for road users in such a lane. Moreover, it may be that part of the variance could be reduced by designing merging zones on straight, uncomplicated road sections as drivers may otherwise hold off transferring control to the automated driving system until the driver feels it safe to hand control to the automated driving system.
Resuming control from automation
A review of the literature found that most papers tend to report the mean Take Over reaction time and often fail to report standard deviation and range (c.f. REF _Ref451171596 \h Figure 2), thus the variance in control transition times remain unknown (median, and interquartile range for each participant in this study can be found in the supplementary material). Additionally, the reviewed papers tended to give drivers a lead time of between 0 and 30 seconds between the presentation of a TOR and a critical event with the main part of the reviewed papers using a 3 or 7 second lead time. In this study we found that the range of time in which drivers resume control from the automated driving system was between 1.9 and 25.7 seconds depending on task engagement. The spread of TOrt in the two conditions in this study indicates that mean, or median values do not tell the entire story when it comes to control transitions. Notably, the distribution of TOrt approaches platykurtic (c.f. REF _Ref454370707 \h \* MERGEFORMAT Figure 9) when drivers are engaged in a secondary task. This implies that vehicle manufacturers must adapt to the circumstances, providing more time to drivers who are engaged in secondary tasks, whilst in highly automated driving mode to avoid excluding drivers at the tail of the distribution. In light of this, designers of automated vehicles should not focus on the mean, or median, driver when it comes to control transition times. Rather they should strive to include the larger range of control transitions times so they do not exclude users that fall outside the mean or median. Moreover, policy-makers should strive to accommodate these inter- and intra-individual differences in their guidelines for sufficiently comfortable transition times. When drivers were allowed to self-regulate the control transition process, little differences could be found in both driving performance and workload between the two conditions. This lends further support to the argument for designing for the range of transition-times rather than the mean or median in non-critical situations.
Lastly, based on the large decrease in TOrt kurtosis when drivers were engaged in a secondary task, it may also be the case that future automated vehicles need to adapt the TORlt to account for drivers engaged in other, non-driving, tasks and even adapt TORlt to accommodate external factors such as traffic density and weather.
Key Points
Large differences between control transitions times reported in the literature and the no-secondary task condition were found.
Drivers take longer to resume control from automation when engaged in a secondary task prior to the control transition.
An inclusive design approach is needed to accommodate the observed variance as the mean or median response times are not sufficient when it comes to designing control transitions in automated driving.
References
ADDIN EN.REFLIST Audi. (2014). Audi piloted driving. from HYPERLINK "http://www.audi.com/com/brand/en/vorsprung_durch_technik/content/2014/10/piloted-driving.html" http://www.audi.com/com/brand/en/vorsprung_durch_technik/content/2014/10/piloted-driving.html
Bainbridge, L. (1983). Ironies of Automation. Automatica, vol 19, no 6, pp 775-779.
Banks, V. A., & Stanton, N. A. (2015). Contrasting models of driver behaviour in emergencies using retrospective verbalisations and network analysis. Ergonomics, vol 58, no 8, pp 1337-1346.
Banks, V. A., & Stanton, N. A. (2016). Keep the driver in control: Automating automobiles of the future. Appl Ergon, vol 53 Pt B, pp 389-395.
Belderbos, C. (2015). Authority Transition Interface: A Human Machine Interface for Taking over Control from a Highly Automated Truck. TU Delft, Delft University of Technology.
BMW. (2013). ConnectedDrive.
Byers, J. C., Bittner, A., & Hill, S. (1989). Traditional and raw task load index (TLX) correlations: Are paired comparisons necessary. Advances in industrial ergonomics and safety I, pp 481-485.
Cooper, J. M., Vladisavljevic, I., Medeiros-Ward, N., Martin, P. T., & Strayer, D. L. (2009). An investigation of driver distraction near the tipping point of traffic flow stability. Human Factors: The Journal of the Human Factors and Ergonomics Society, vol 51, no 2, pp 261-268.
Cranor, L. F. (2008). A framework for reasoning about the human in the loop. Paper presented at the Proceedings of the 1st Conference on Usability, Psychology, and Security.
Dambck, D., Bengler, K., Farid, M., & Tnert, L. (2012). bernahmezeiten beim hochautomatisierten Fahren. Tagung Fahrerassistenz. Mnchen, vol 15, pp 16.
Desmond, P. A., Hancock, P. A., & Monette, J. L. (1998). Fatigue and automation-induced impairments in simulated driving performance. Human Performance, User Information, and Highway Design, no 1628, pp 8-14.
Dogan, E., Deborne, R., Delhomme, P., Kemeny, A., & Jonville, P. (2014). Evaluating the shift of control between driver and vehicle at high automation at low speed: The role of anticipation. Paper presented at the Transport Research Arena (TRA) 5th Conference: Transport Solutions from Research to Deployment.
Endsley, M. R., & Kaber, D. B. (1999). Level of automation effects on performance, situation awareness and workload in a dynamic control task. Ergonomics, vol 42, no 3, pp 462-492.
Eriksson, A., Lindstrm, A., Seward, A., Seward, A., & Kircher, K. (2014). Can user-paced, menu-free spoken language interfaces improve dual task handling while driving? Human-Computer Interaction. Advanced Interaction Modalities and Techniques (pp. 394-405): Springer.
Eriksson, A., Spychala, A., F., D. W. J. C., & Stanton, N. A. (2016). A Highly Automated Driving Algorithm Implementation on STISIM. Paper presented at the AHFE 2016.
Eriksson, A., & Stanton, N. A. (2016). The Chatty Co-Driver: A Linguistics Approach to Human-Automation-Interaction. Paper presented at the IEHF2016.
Feldhtter, A., Gold, C., Schneider, S., & Bengler, K. (2016). How the Duration of Automated Driving Influences Take-Over Performance and Gaze Behavior. Paper presented at the Arbeit in komplexen Systemen - Digital, vernetz, human?! 62. Kongress der Gesellschaft fr Arbeitswissenschaft.
Fisher, D. L., Rizzo, M., Caird, J., & Lee, J. D. (2011). Handbook of driving simulation for engineering, medicine, and psychology: CRC Press.
Gold, C., Dambck, D., Lorenz, L., & Bengler, K. (2013). Take over! How long does it take to get the driver back into the loop? Paper presented at the Proceedings of the Human Factors and Ergonomics Society Annual Meeting.
Gold, C., Korber, M., Lechner, D., & Bengler, K. (2016). Taking Over Control From Highly Automated Vehicles in Complex Traffic Situations: The Role of Traffic Density. Hum Factors.
Gold, C., Lorenz, L., & Bengler, K. (2014). Influence of automated brake application on take-over situations in highly automated driving scenarios. Paper presented at the Proceedings of the FISITA 2014 World Automotive Congress.(in print).
Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. Advances in psychology, vol 52, pp 139-183.
Hubert, M., & Vandervieren, E. (2008). An adjusted boxplot for skewed distributions. Computational statistics & data analysis, vol 52, no 12, pp 5186-5201.
Kaber, D. B., & Endsley, M. R. (1997). Out-of-the-loop performance problems and the use of intermediate levels of automation for improved control system functioning and safety. Process Safety Progress, vol 16, no 3, pp 126-131.
Kerschbaum, P., Lorenz, L., & Bengler, K. (2015). A transforming steering wheel for highly automated cars. Paper presented at the Intelligent Vehicles Symposium (IV), 2015 IEEE.
Kircher, K., Larsson, A., & Hultgren, J. A. (2014). Tactical driving behavior with different levels of automation. IEEE Transactions on Intelligent Transportation Systems, vol 15, pp 158-167.
Krber, M., Gold, C., Lechner, D., & Bengler, K. (2016). The influence of age on the take-over of vehicle control in highly automated driving. Transportation Research Part F: Traffic Psychology and Behaviour, vol 39, pp 19-32.
Krber, M., Weigerber, T., Kalb, L., Blaschke, C., & Farid, M. (2015). Prediction of take-over time in highly automated driving by two psychometric tests. Dyna, vol 82, no 193, pp 195-201.
Lorenz, L., Kerschbaum, P., & Schumann, J. (2014). Designing take over scenarios for automated driving How does augmented reality support the driver to get back into the loop? Paper presented at the Proceedings of the Human Factors and Ergonomics Society Annual Meeting.
Louw, T., Kountouriotis, G., Carsten, O., & Merat, N. (2015a). Driver Inattention During Vehicle Automation: How Does Driver Engagement Affect Resumption Of Control?
Louw, T., Merat, N., & Jamson, H. (2015b). Engaging with Highly Automated Driving: To be or Not to be in the Loop? Paper presented at the 8th International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design, At Salt Lake City, Utah, USA.
Lu, Z., & de Winter, J. C. F. (2015). A Review and Framework of Control Authority Transitions in Automated Driving. Procedia Manufacturing, vol 3, pp 2510-2517.
Melcher, V., Rauh, S., Diederichs, F., Widlroither, H., & Bauer, W. (2015). Take-Over Requests for Automated Driving. Procedia Manufacturing, vol 3, pp 2867-2873.
Merat, N., Jamson, A. H., Lai, F. C., & Carsten, O. (2012). Highly automated driving, secondary task performance, and driver state. Human Factors: The Journal of the Human Factors and Ergonomics Society, vol 54, no 5, pp 762-771.
Merat, N., Jamson, A. H., Lai, F. F. C. H., Daly, M., & Carsten, O. M. J. (2014). Transition to manual: Driver behaviour when resuming control from a highly automated vehicle. Transportation Research Part F: Traffic Psychology and Behaviour, vol 26, Part A, pp 1-9.
Michon, J. A. (1985). A critical view of driver behavior models: What do we know, what should we do? In L. Evans & R. C. Schwing (Eds.), Human behavior and traffic safety. New York: Plenum Press.
Mok, B., Johns, M., Lee, K. J., Miller, D., Sirkin, D., Ive, P., et al. (2015). Emergency, Automation Off: Unstructured Transition Timing for Distracted Drivers of Automated Vehicles. Paper presented at the Intelligent Transportation Systems (ITSC), 2015 IEEE 18th International Conference on.
Molloy, R., & Parasuraman, R. (1996). Monitoring an automated system for a single failure: Vigilance and task complexity effects. Human Factors, vol 38, no 2, pp 311-322.
Naujoks, F., Mai, C., & Nekum, A. (2014, 19-23 July). The effect of urgency of take-over requests during highly automated driving under distracted conditions. Paper presented at the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Krakow, Poland.
Naujoks, F., & Nekum, A. (2014). Timing of in-vehicle advisory warnings based on cooperative perception. Paper presented at the Human Factors and Ergonomics Society Europe Chapter Annual Meeting.
Nilsson, J. (2014). Safe Transitions to Manual Driving From Faulty Automated Driving System: Chalmers University of Technology.
Payre, W., Cestac, J., & Delhomme, P. (2016). Fully Automated Driving: Impact of Trust and Practice on Manual Control Recovery. Hum Factors, vol 58, no 2, pp 229-241.
Petermeijer, S. M., de Winter, J. C. F., & Bengler, K. J. (2016). Vibrotactile Displays: A Survey With a View on Highly Automated Driving. IEEE Transactions on Intelligent Transportation Systems, vol 17, no 4, pp 897-907.
Porter, J. M., Case, K., Marshall, R., Gyi, D., & Sims ne Oliver, R. (2004). Beyond Jack and Jill: designing for individuals using HADRIAN. International Journal of Industrial Ergonomics, vol 33, no 3, pp 249-264.
Radlmayr, J., Gold, C., Lorenz, L., Farid, M., & Bengler, K. (2014). How Traffic Situations and Non-Driving Related Tasks Affect the Take-Over Quality in Highly Automated Driving. Proceedings of the Human Factors and Ergonomics Society Annual Meeting, vol 58, no 1, pp 2063-2067.
Robinson, T., Chan, E., & Coelingh, E. (2010). Operating platoons on public motorways: An introduction to the sartre platooning programme. Paper presented at the 17th world congress on intelligent transport systems.
SAE International. (2016). Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles, J3016_201609: SAE International.
safecarnews.com. (2015). Intelligent Drive Concept for new Mercedes-Benz GLC. from HYPERLINK "http://safecarnews.com/intelligent-drive-concept-for-new-mercedes-benz-glc_ju6145/" http://safecarnews.com/intelligent-drive-concept-for-new-mercedes-benz-glc_ju6145/
Schmig, N., Hargutt, V., Neukum, A., Petermann-Stock, I., & Othersen, I. (2015). The interaction between highly automated driving and the development of drowsiness.
Scott, J., & Gray, R. (2008). A comparison of tactile, visual, and auditory warnings for rear-end collision prevention in simulated driving. Human Factors: The Journal of the Human Factors and Ergonomics Society, vol 50, no 2, pp 264-275.
Stanton, N. A., Dunoyer, A., & Leatherland, A. (2011). Detection of new in-path targets by drivers using Stop & Go Adaptive Cruise Control. Appl Ergon, vol 42, no 4, pp 592-601.
Stanton, N. A., & Marsden, P. (1996). From fly-by-wire to drive-by-wire: Safety implications of automation in vehicles. Safety Science, vol 24, no 1, pp 35-49.
Stanton, N. A., Young, M., & McCaulder, B. (1997). Drive-by-wire: The case of driver workload and reclaiming control with adaptive cruise control. Safety Science, vol 27, no 2-3, pp 149-159.
Stanton, N. A., & Young, M. S. (2005). Driver behaviour with adaptive cruise control. Ergonomics, vol 48, no 10, pp 1294-1313.
Stanton, N. A., Young, M. S., Walker, G. H., Turner, H., & Randle, S. (2001). Automating the Drivers Control Tasks. INTERNATIONAL JOURNAL OF COGNITIVE ERGONOMICS.
Strand, N., Nilsson, J., Karlsson, I. C. M., & Nilsson, L. (2014). Semi-automated versus highly automated driving in critical situations caused by automation failures. Transportation Research Part F: Traffic Psychology and Behaviour.
Summala, H. (2000). Brake Reaction Times and Driver Behavior Analysis. Transportation Human Factors, vol 2, no 3, pp 217-226.
Swaroop, D., & Rajagopal, K. R. (2001). A review of Constant Time Headway Policy for Automatic Vehicle Following. Paper presented at the IEEE Intelligent Transportation Systems Conference Proceedings, Oakland (CA), USA.
United Nations. (1968). Convention on road traffic, Done at Vienna on 8 November 1968. Amendment 1. Retrieved from HYPERLINK "http://www.unece.org/fileadmin/DAM/trans/conventn/crt1968e.pdf" http://www.unece.org/fileadmin/DAM/trans/conventn/crt1968e.pdf.
van den Beukel, A. P., & van der Voort, M. C. (2013). The influence of time-criticality on Situation Awareness when retrieving human control after automated driving. Paper presented at the Intelligent Transportation Systems-(ITSC), 2013 16th International IEEE Conference on.
Verboven, S., & Hubert, M. (2005). LIBRA: a MATLAB library for robust analysis. Chemometrics and intelligent laboratory systems, vol 75, no 2, pp 127-136.
Walch, M., Lange, K., Baumann, M., & Weber, M. (2015). Autonomous driving: investigating the feasibility of car-driver handover assistance. Paper presented at the Proceedings of the 7th International Conference on Automotive User Interfaces and Interactive Vehicular Applications.
Walker, G. H., Stanton, N. A., & Salmon, P. M. (2015). Human Factors in Automotive Engineering and Technology: Ashgate: Aldershot.
Willemsen, D., Stuiver, A., & Hogema, J. (2015). Automated Driving Functions Giving Control Back to the Driver: A Simulator Study on Driver State Dependent Strategies. Paper presented at the 24th International Technical Conference on the Enhanced Safety of Vehicles (ESV).
Wolterink, W. K., Heijenk, G., & Karagiannis, G. (2011). Automated Merging in a Cooperative Adaptive Cruise Control (CACC) System. Paper presented at the Fifth ERCIM Workshop on eMobility, Vilanova i la Geltr, Catalonia, Spain.
Young, M. S., & Stanton, N. A. (2002). Malleable attentional resources theory: a new explanation for the effects of mental underload on performance. Hum Factors, vol 44, no 3, pp 365-375.
Young, M. S., & Stanton, N. A. (2007). Back to the future: brake reaction times for manual and automated vehicles. Ergonomics, vol 50, no 1, pp 46-58.
- : A ` a g
? @ n o p r s *
+
^
_
`
~
ѿ}yn}e[N h' h[T 5B*ph j h; 0J Uh' h[T 0J j h`U Uh; j h; Uh' h[T h' h[T B*ph h' 5B*ph h{| 5B*ph h-> 5B*ph hzn 5B*ph h' h' 5B*ph hj> 5 h' 5h\1p hc 5h8 h8 6 h\1p 5 hc 5 h1 5 h9 5 h 5 a
@ b s
( 4
h i k t , D j x d gd{| gd{| d hd d d -D M
[$\$^hgdWX d gd\ gd
d gd
d d d -D M
[$\$gd\ gd\1p gd\
' ( 4 G L d
h i k s t } ~ ǹ~iVF hzi B*PJ ^J nH ph tH %hzi hzi B*PJ ^J nH ph tH (hzi hzi 5B*PJ ^J nH ph tH (hzi h R! 5B*PJ ^J nH ph tH hv\ h 5hv\ h( 5 h' 5 h' h' 5B*OJ QJ ph h
B*CJ PJ aJ ph h
h
CJ aJ h
CJ aJ h#f h
CJ aJ hH\ h
5h' h' 6h[T h' h' N X Z
$ : = B ` s ʸߨߘߨ߅r_rLL9L%h9 h>\ B*PJ ^J nH ph tH %h9 h2 B*PJ ^J nH ph tH %h9 hf B*PJ ^J nH ph tH %h9 h] B*PJ ^J nH ph tH %h9 hzi B*PJ ^J nH ph tH hP(9 B*PJ ^J nH ph tH hj> B*PJ ^J nH ph tH "hzi 5B*PJ ^J nH ph tH (hzi h R! 5B*PJ ^J nH ph tH hzi B*PJ ^J nH ph tH hf B*PJ ^J nH ph tH A B H J ] c f q t % ' ; < Ʊvvvvvvfى hP(9 B*PJ ^J nH ph tH %h9 h5 B*PJ ^J nH ph tH %h9 hj> B*PJ ^J nH ph tH (h9 hzi 5B*PJ ^J nH ph tH (h9 h R! 5B*PJ ^J nH ph tH %h9 hWX B*PJ ^J nH ph tH %h9 hzi B*PJ ^J nH ph tH %h9 h2 B*PJ ^J nH ph tH < E " $ 1 \ b
) + , K S صص؍}mZ؍}Gm %h9 ha D B*PJ ^J nH ph tH %h9 h>\ B*PJ ^J nH ph tH h
= B*PJ ^J nH ph tH hgc B*PJ ^J nH ph tH %h9 hT' B*PJ ^J nH ph tH (h9 h5 5B*PJ ^J nH ph tH hP(9 B*PJ ^J nH ph tH %h9 hj> B*PJ ^J nH ph tH %h9 hWX B*PJ ^J nH ph tH (h9 hWX 5B*PJ ^J nH ph tH S U ` b + , 4 6 B V ] ^ h j k o p s u ~ ¯yuquqmumuqumumumiuiuibU hv hv 5B*ph h' h' h2 h' hv\ ha D h8 h8 h8 h8 h8 5%hj> h R! B*PJ ^J nH ph tH hj> B*PJ ^J nH ph tH %h9 hj> B*PJ ^J nH ph tH (h9 hWX 5B*PJ ^J nH ph tH (h9 h R! 5B*PJ ^J nH ph tH %h9 hWX B*PJ ^J nH ph tH / B C D S j k w x Ż{w{skgk^kZ h9;1 hI" mH nH uhxt j hI" UhI" hP(9 h[T h_} h hZQ h hZQ h[T hZQ h;} hZQ hmk hmk h? h;} h( 5hV1 h( 5*h\ B*ph h[ h{| \h{| 5B*ph h{| B*ph hf B*ph hP(9 B*ph hf B*ph hv B*ph # , - . # # # # # # # $ 1$ @$ S$ _$ `$ k' l' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' +( -( .( 6( 8( C( I( L( ĽĽđđđđxx hWK h ^J hWK h"% ^J h9 h"% h9 h h9 hP(9 h9 h mH nH uj h9 h Uh9 h h9 hP h9 haS h9 hG- mH nH uhxt j h9 hG- Uh9 hG- h9 hLI hP(9 h_} hI" .L( M( S( b( d( p( q( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( !) A) Z) [) f, g, , , , ɿɲϤwnc\cPc\ h9 h mH nH uh9 h j h9 h UhWK h ^J hWK hc" ^J hWK hP(9 ^J hWK hmk ^J hWK hLI ^J hWK h ^J hWK h ^J mH nH uj hxt U^J j hxt U^J
hxt ^J j hWK h"% U^J hWK hC ^J hWK h ^J hWK h"% ^J hWK h ^J x ( c} ~ ~ ~ ~ (~ )~ +~ W kdr $$If l \ Ld_Z!
t 6 4 4
l a p( ytWK d $If gdWK gdq gd\ , , , , , , - - -
- /- 0- :- >- w- x- y- - - - - - - - - - . . . &. '. (. 7. 8. L. M. N. O. . |o|ahWK h/b ^J mH nH ujA hxt U^J j hWK h/b U^J hWK hOk ^J hWK hx! ^J hWK hP(9 ^J hWK h9;1 ^J hWK h[T ^J hWK h_ ^J mH nH uj
hxt U^J j hxt U^J
hxt ^J j hWK hC U^J hWK hC ^J hWK h ^J &. . . . . . . 3 3 3 3 3 3 3 4 ,4 34 64 74 >4 e4 4 4 4 4 4 4 o: p: : : : : ʶʭפגwjj\jS hWK hu ^J hWK h_ ^J mH nH uj hWK h_ U^J hWK h_ ^J hWK h+_ ^J hWK h"B ^J hWK hF- ^J hWK h ^J hWK h ^J hWK hP(9 ^J hWK h4 ^J mH nH u
hxt ^J j hWK h4 U^J hWK hLI ^J hWK h/b ^J hWK hx! ^J j hWK h/b U^J : : : : : : : ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; =C >C UC VC WC XC C C C C C -D .D =D >D RD һح}vrnrjr haO hu hOk hOk hOk hWK haO ^J mH nH uj hWK haO U^J hWK hF- ^J
h ^J
haO ^J hWK hR ^J mH nH uj% hxt U^J j hxt U^J
hxt ^J j hWK h] U^J hWK h[T ^J hWK h9 ^J hWK h] ^J %RD SD TD UD D D D D D E E E E E K K K K K K L L L L (L 2L @L IL mL nL oL L L Ļvmmd[RhWK h9FD ^J hWK hJ ^J hWK h ^J hWK h"B ^J hWK h ^J hWK h, ^J hWK h, ^J mH nH u
hxt ^J j hWK h, U^J hWK h7 ^J hWK h9 ^J hWK h, ^J hWK h] ^J hWK hm` ^J mH nH uj hWK h] U^J j hxt U^J jG hxt U^J L L L L L L L L L L
M M M M 2M 3M 4M 5M wM xM zM {M T T U U
U U $U &U