MEC: a robot number optimization algorithm for coverage path planning with efficient Dubins robots

Lin Li, Xing Zhou, Songchang Jin, Dan Chen, Junfeng Man, Dianxi Shi

Abstract: Multi-robot coverage path planning (MCPP) has received increasing attention in agricultural automation. In practical agricultural applications, robots are critical resources, and optimizing their number to execute a specific coverage task with a deadline is an important topic. Existing MCPP methods typically assume a fixed number of robots and each robot covers the farmland with a constant speed, which may increase task costs, thereby causing a waste of robot resources. To address this challenge, this article proposes a resource-optimized coverage path planning method. First, as a base, we formulate a generalized Dubins model, containing variable speeds instead of a constant speed, to generate the (near-)fastest Dubins path between two poses instead of the shortest Dubins path, thus improving the efficiency of a single robot to perform tasks. Next, we propose a multi-objective evolutionary algorithm for MCPP called MEC. The MEC algorithm optimally minimizes the number of robots and the single-robot maximum coverage time, by reusing the optimization information through the computing process. We validate the performance of MEC through comparison experiments. The results demonstrate that MEC can reduce the robot usage by up to 25% without the generalized Dubins model, and MEC can further reduce robot usage by at least 33.3% when incorporating the generalized Dubins model.

External IDs:dblp:journals/jksucis/LiZJCMS25