Global Path Planning for Autonomous Vehicles in Orchards and Vineyards

Timo Schönegg, Turcan Tuna, Fan Yang, Gabriel Waibel, Matías Mattamala, Marco Hutter

Abstract: Autonomous vehicles have a vast potential to increase the efficiency and productivity in agriculture. A known challenge in the application of autonomous vehicles in agriculture is the requirement to cover a certain area, such as rows of a vineyard field while navigating in an unstructured environment. This paper presents a new time-constrained global path-planning approach for kinematically constrained autonomous vehicles in orchards and vineyards. The proposed framework utilizes a volumetric representation of the environment as an input to generate a navigation graph. This graph is then used by a hierarchical A* approach that enforces the kinematic vehicle constraints for a given computational time budget to find a feasible global path covering the whole field. Through simulated and real-world data, the proposed approach is validated for various environments using different vehicle configurations, showing the efficacy and the gained performance. Furthermore, the generalizability of the proposed framework to different environments is validated through real-world forest and search-and-rescue training facility data.