Coverage Path Planning for a Moving Vehicle
MetadataShow full item record
A simple coverage plan called a Conformal Lawn Mower plan is demonstrated. This plan enables a UAV to fully cover the route ahead of a moving ground vehicle. The plan requires only limited knowledge of the ground vehicle's future path. For a class of curvature-constrained ground vehicle paths, the proposed plan requires a UAV velocity that is no more than twice the velocity required to cover the optimal plan. Necessary and sufficient UAV velocities, relative to the ground vehicle velocity, required to successfully cover any path in the curvature restricted set are established. In simulation, the proposed plan is validated, showing that the required velocity to provide coverage is strongly related to the curvature of the ground vehicle's path. The results also illustrate the relationship between mapping requirements and the relative velocities of the UAV and ground vehicle. Next, I investigate the challenges involved in providing timely mapping information to a moving ground vehicle where the path of that vehicle is not known in advance. I establish necessary and sufficient UAV velocities, relative to the ground vehicle velocity, required to successfully cover any path the ground vehicle may follow. Finally, I consider a reduced problem for sensor coverage ahead of a moving ground vehicle. Given the ground vehicle route, the UAV planner calculates the regions that must be covered and the time by which each must be covered. The UAV planning problem takes the form of an Orienteering Problem with Time Windows (OPTW). The problem is cast the problem as a Mixed Integer Linear Program (MILP) to find a UAV path that maximizes the area covered within the time constraints dictated by the moving ground vehicle. To improve scalability of the proposed solution, I prove that the optimization can be partitioned into a set of smaller problems, each of which may be solved independently without loss of overall solution optimality. This divide and conquer strategy allows faster solution times, and also provides higher-quality solutions when given a fixed time budget for solving the MILP. We also demonstrate a method of limited loss partitioning, which can perform a trade-off between improved solution time and a bounded objective loss.
Cite this version of the work
Barry Gilhuly (2020). Coverage Path Planning for a Moving Vehicle. UWSpace. http://hdl.handle.net/10012/15775