Making a Liquid Robotics Wave Glider Fully Autonomous: Field Results From DANS20 Trial

Abstract

We present the integration of an embedded autonomy engine into a Liquid Robotics Wave Glider surface vehicle, following the “backseat-driver” control paradigm. We describe the Wave Glider Autonomy Adapter software module, designed for handling the communication between a Backseat payload computer, installed by the vehicle end-user, with the manufacturer Frontseat Command and Control Unit. The module runs in the MOOS middleware environment and enables the intelligent Cooperative Autonomous Decision Making Engine (iCADME), running on the Backseat, to execute a mission involving autonomous decision-making and to issue the produced commands to the Wave Glider Frontseat autopilots, hence controlling the vehicle navigation. This makes the CMRE Wave Glider fully autonomous and capable to make decisions onboard, with no need of communication with the Command and Control (C2) centre. This reduces the need of data exchange with the C2 centre through the traditionally used satellite link. In this way, the vehicle planning and control are significantly simplified. This paves the way to the deployment of highly scalable networks, composed of cooperative Wave Gliders, capable to operate effectively in complex missions with only a limited supervision by the C2 centre. We report successfully tests of the fully autonomous Wave Glider conducted during the DANS20 trial, in December 2020. The onboard autonomy was able to successfully control the vehicle both by issuing direct waypoints commands and by using the MOOS-IvP Helm behaviour-based autonomy for producing heading reference for the Wave Glider autopilots.