Novel Motion Models for Time-Optimal Risk-Aware Motion Planning for Variable-Speed AUVs

Abstract

In this paper, we develop new motion models for curvature-constrained and variable-speed Autonomous Underwater Vehicles (AUVs) for time-optimal risk-aware motion planning. AUVs are becoming increasingly useful and cost-effective for a variety of tasks in many underwater applications including surveillance and scientific expeditions. Despite recent advances, the autonomy of AUVs is limited, especially for vehicles that operate in environments with obstacles. In particular, there is limited research for time-optimal risk-aware motion planning. In contrast, there has been significant research for finding the time-optimal paths in environments without obstacles; however, adapting these models to environments with obstacles yields sub-optimal results, since these models force the AUV to operate at extremal speeds at close proximity to obstacles. Specifically, moving at maximum speed increases the risk of collision, while moving at minimum speed dramatically increases travel time. As such, this paper presents new motion models for AUVs that enable the selection of intermediate speeds to provide a better balance between time and risk near obstacles. These models enhance the agility and maneuverability of the AUV and provide motion planners the flexibility to select appropriate speeds and therefore construct time-optimal risk-aware paths. Additionally, the models are simple to compute and are suitable for on-demand real-time computation. The performance of the proposed model is compared against existing models using our recently developed T⋆ algorithm for time-optimal risk-aware motion planning. The results show that our new model yields paths that are shorter in obstacles-rich scenarios with substantially lower risks.