Continuous Curvature Path Planning for Headland Coverage With Agricultural Robots

ABSTRACT We introduce a methodology for headland coverage planning for autonomous agricultural robot systems, which is a complex problem often overlooked in agricultural robotics. At the corners of the headlands, a robot faces the risk to cross the border of a field while turning. Though potentially dangerous, current papers about corner turns in headlands do not tackle this issue. Moreover, they produce paths with curvature discontinuities, which are not feasible by non‐holonomic robots. This paper presents an approach to strictly adhere to field borders during the headland coverage, and three types of continuous curvature turn planners for convex and concave corners. The turning planners are evaluated in terms of path length and uncovered area to assess their effectiveness in headland corner navigation. Through empirical validation, including extensive tests on a coverage path planning benchmark as well as real‐field experiments with an autonomous robot, the proposed approach demonstrates its practical applicability and effectiveness. In simulations, the mean coverage area of the fields went from 94.73%, using a constant offset around the field, to 97.29% using the proposed approach. Besides providing a solution to the coverage of headlands in agricultural automation, this paper also extends the covered area on the mainland, thus increasing the overall productivity of the field.