A Distributed Approach for Robotic Coverage Path Planning Under Steep Slope Terrain Conditions

This article proposes a novel algorithm to determine the optimal coverage path of a mobile robot for uniform seed dispersion in any tract of agricultural land. The robot is required to operate under steep terrain conditions that are too risky for conventional, human operated equipment. Using data from an field experiment with the real robot, machine learning based function approximators are trained to estimate the minimum energy paths between adjacent points. Exemplar-based clustering is used to identify a suitable subset of way-points that ensure full coverage of a tract of land. An optimal cyclic tour is computed from the way points using a new asymmetric TSP algorithm proposed for this specific application. The clustering and the cyclic path planning algorithms can be implemented entirely through local message passing in a field-deployed sensor network. Simulations with synthetic as well as real topographic data establish the overall effectiveness of the proposed method.