Abstract
The agro-forestry sector is one of those that could benefit from new technologies and robotic applications to adapt to climate change effects, population growth and globalization. These applications can improve the efficiency and performance of the operations, as well as the products’ quality and traceability. Anyhow, the possibility to automate many of the applications carried out in the open field is related to the development of mobile platforms able to move and adapt to highly unstructured environments. Within this context, this work focuses on the development of a path-planning algorithm for UAVs navigating through an orchard/vineyard environment without colliding with static obstacles (e.g., plants) and/or unexpected ones (e.g., people or other robots operating in the same area). The path-planning strategy is grounded on an enhanced Boundary Node Method (BNM) and a re-planning approach is considered to recomputing and reconfiguring the path in case of unexpected obstacles that prevent the robot from following the original one. The developed method has been firstly evaluated in a simulated environment (i.e., Gazebo simulator), considering different combinations of obstacles and robot tasks. The same numerical experiments were then tested in a real emulated orchard scenario. The results showed good path-tracking capabilities in case of both known and unknown obstacles.