Autonomous Data Collection With Timed Communication Constraints for Unmanned Underwater Vehicles

This letter studies the problem of data collection where an autonomous underwater vehicle (AUV) is required to inspect several goals while staying within the communication range of an unmanned surface vehicle (USV) moving along a predefined path. A reward is associated with each inspected goal, and, so the overall objective is to increase the sum of the rewards. The proposed planner takes the AUV dynamics and the obstacles into account in order to generate collision-free and dynamically-feasible trajectories that enable the AUV to inspect multiple goals while maintaining communication with the USV. When not all the goals can be inspected, the planner generates trajectories that increase the sum of the rewards. The proposed planner constructs a roadmap equipped with time information in order to facilitate navigation and maintain communication in complex 3D environments. A crucial aspect is an effective coupling of discrete search with sampling-based motion planning. Specifically, a multi-goal discrete search is proposed to effectively compute roadmap paths that increase the sum of the rewards. These paths then guide the sampling-based expansion of a motion tree. Experiments using complex 3D environments and an AUV model defined by nonlinear second-order differential equations indicate the efficiency of the planner to inspect an increasing number of goals.