Dynamic Coverage of Unicycle Robots with Anisotropic Sensing for Time-Priority

This paper deals with dynamic coverage control of anisotropic sensors with the unicycle motion platform. By constructing the shortest coverage paths through geometric analysis, a global target allocation is realized with a generalized Voronoi partition using time-priority principle. Then, based on the compact representation of target states, a local dynamic coverage policy is derived by reinforcement learning approach, adaptable to changes in target number. It is shown via a numerical example that the proposed method speeds up the convergence of learning as well as fast coverage completion time.