QED: using Quality-Environment-Diversity to evolve resilient robot swarms
QED: using Quality-Environment-Diversity to evolve resilient robot swarms
In quality-diversity algorithms, the behavioural diversity metric is a key design choice that determines the quality of the evolved archives. Although behavioural diversity is traditionally obtained by describing the observed resulting behaviour of robot controllers evaluated in a single environment, it is often more easily induced by introducing environmental diversity, i.e., by manipulating the environments in which the controllers are evaluated. This paper proposes Quality-Environment-Diversity, an algorithm that repeatedly generates a random environment according to a probability distribution over environmental features (e.g., number of obstacles, arena size and robot sensor and actuator characteristics), evaluates the controller in that environment, and then describes the controller in terms of the features of that environment, the environment descriptor. Our study compares Quality-Environment-Diversity to three baseline task-specific and generic behavioural descriptors, in 5 different robot swarm benchmark tasks. For each task, the quality of the evolved archives is assessed by their capability to provide high-performing compensatory behaviours following injection of 250 unique faults to the robots of the swarm. The evolved archives achieve a median 2-to 3-fold reduction in the impact of the faults on the performance of the swarm. A qualitative analysis of evolved archives is done by visualising the relation between diversity of compensatory behaviours, here called useful behavioural diversity, and fault recovery metrics. The resulting signatures indicate that, due to the diversity of environments inducing useful behavioural diversity, archives evolved by QED provide robot swarm controllers that are capable of recovering from high-impact faults.
Actuators, Legged locomotion, Quality-diversity algorithms, Robot kinematics, Robot sensing systems, Swarm robotics, Task analysis, behavioural diversity, evolutionary robotics, fault recovery, swarm robotics.
Bossens, David
633a4d28-2e59-4343-98fe-283082ba1873
Tarapore, Danesh
fe8ec8ae-1fad-4726-abef-84b538542ee4
6 November 2020
Bossens, David
633a4d28-2e59-4343-98fe-283082ba1873
Tarapore, Danesh
fe8ec8ae-1fad-4726-abef-84b538542ee4
Bossens, David and Tarapore, Danesh
(2020)
QED: using Quality-Environment-Diversity to evolve resilient robot swarms.
IEEE Transactions on Evolutionary Computation.
(doi:10.1109/TEVC.2020.3036578).
Abstract
In quality-diversity algorithms, the behavioural diversity metric is a key design choice that determines the quality of the evolved archives. Although behavioural diversity is traditionally obtained by describing the observed resulting behaviour of robot controllers evaluated in a single environment, it is often more easily induced by introducing environmental diversity, i.e., by manipulating the environments in which the controllers are evaluated. This paper proposes Quality-Environment-Diversity, an algorithm that repeatedly generates a random environment according to a probability distribution over environmental features (e.g., number of obstacles, arena size and robot sensor and actuator characteristics), evaluates the controller in that environment, and then describes the controller in terms of the features of that environment, the environment descriptor. Our study compares Quality-Environment-Diversity to three baseline task-specific and generic behavioural descriptors, in 5 different robot swarm benchmark tasks. For each task, the quality of the evolved archives is assessed by their capability to provide high-performing compensatory behaviours following injection of 250 unique faults to the robots of the swarm. The evolved archives achieve a median 2-to 3-fold reduction in the impact of the faults on the performance of the swarm. A qualitative analysis of evolved archives is done by visualising the relation between diversity of compensatory behaviours, here called useful behavioural diversity, and fault recovery metrics. The resulting signatures indicate that, due to the diversity of environments inducing useful behavioural diversity, archives evolved by QED provide robot swarm controllers that are capable of recovering from high-impact faults.
Text
QED
- Accepted Manuscript
More information
Accepted/In Press date: 1 January 2020
Published date: 6 November 2020
Keywords:
Actuators, Legged locomotion, Quality-diversity algorithms, Robot kinematics, Robot sensing systems, Swarm robotics, Task analysis, behavioural diversity, evolutionary robotics, fault recovery, swarm robotics.
Identifiers
Local EPrints ID: 445078
URI: http://eprints.soton.ac.uk/id/eprint/445078
ISSN: 1089-778X
PURE UUID: aa29c432-e88d-4431-86d5-82486d190f87
Catalogue record
Date deposited: 19 Nov 2020 17:30
Last modified: 17 Mar 2024 03:46
Export record
Altmetrics
Contributors
Author:
David Bossens
Author:
Danesh Tarapore
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics