Enhanced probabilistic roadmap for robot navigation in virtual greenhouse environment

Abstract

Inefficient navigation capability in dynamic agricultural field limits the application of agricultural robot in the field. Probabilistic roadmap however has the robustness for outdoor navigation. A path planning algorithm was established upon an enhanced probabilistic roadmap and this was implemented in a virtual greenhouse environment. A smoothing algorithm for the robot’s navigation has been proposed to improve the existing algorithm in producing an optimal path. A simulation was conducted using a crop inspection mobile robot and tested with suitable turning trajectories for lane changing. Several trajectories were initiated and compared based on travel time, distance and controller error in order to choose the best trajectories for crop inspection. The proposed smoothing algorithm was able to smooth out the initial paths in order to create an optimal path for the robot with error less than 0.1 m.