Metrology-aware Path Planning for Agricultural Mobile Robots in Dynamic Environments

Path planning for long-term tasks is currently an important research area for robotic applications in uncertain environments, such as farms. Here, a robot automatically follows an optimal path generated by the robot planner, and the robot planner must consider the uncertainties in the environment to adapt and improve its motion plan during run-time. In some cases, the robot must change its motion direction multiple times in a long-term task. Therefore, the robot has to decide which place to visit next to adapt to unexpected events, such as avoiding obstacles not recognized by the robot or some robot features and environmental conditions. The robot will need to either plan a short path and then back to the global path or, if better for its overall performance, create a new global path to the target and automatically follow the new path. This study proposes a path planning method to generate the unmanned ground vehicle path to visit all predefined locations on a farm. The proposed method is an extended version of our previously developed method, called the Boundary Node Method and Path Enhancement Method. Then, this method is combined with the online path optimization scheme for re-planning and scheduling the new path at run-time to improve the robot’s path and adapt to unexpected events. Different simulation environments are examined to demonstrate the performance of the proposed method. The results show that the robot can re-plan an optimal path toward the goal points to avoid unexpected threats and uncertain obstacles.