Improving the perching capability of a vertical take-off and landing unmanned vehicle through reconfiguration

Abstract

Landing on lighting columns like nature’s birds is a desirable capability which can only extend the uses of unmanned aerial systems. This thesis investigates what the most effective form of perching on existing street furniture with a VTOL UAV and how the perch site can be recognised using low cost off the shelf sensors. Additionally to this, the UAV in question will have to execute the perch without relying on GPS data. The work conducted here covers an extensive design review which selects a bird claw like gripper to sustain the perch. In order for the UAV to know where it is in relation to the perch site without relying on GPS data, a Raspberry Pi and PiCamera were used to detect common features which are found on top of a lighting column. Using a search and perch algorithm which was developed specifically for the task of perching on lamp post projection brackets, the on-board microprocessor controlled the UAV over the perch site and gradually descended into the perch position. The perched position and approach was also tested to ensure the perching element could cope with various weather conditions. The testing was conducted in a wind tunnel with the UAV mounted in various perched positions and the moment the UAV would slip, the wind speed were measured and analysed which highlighted an interesting prediction method. During the perching development the addition of a gantry style test-rig was also developed to ease the algorithm development with minimal incidents. The final result is a search and perch algorithm which is initiated when the VTOL UAV is within the vicinity of a lamp post at which point the on-board vision processing and control system takes over, removing the burden from the UAV operator to ensure a collision free perch.

More information

Identifiers

ORCID for Stephen Prior: orcid.org/0000-0002-4993-4942

Catalogue record

Export record