Simulation of optimal path planning using modified minimum spanning tree for autonomous agricultural mobile robots

Background: Autonomous driving is a major issue for precision agriculture. One of important technic is a generate optimum path for autonomous driving. Up to now, autonomous driving technology generated a path that created a least-cost path using only the distance cost. However, in real field large driving radius cause lateral deviation error of mobile robot. Therefore, in this paper, we propose a path planning algorithm base on modified Minimum Spanning Tree (mMST) that takes weight of vehicle radius of gyration into account.

Methods: In this paper we generate optimal path using mMST. First, randomly create nodes in virtual test field. Calculate distance cost, radius of gyration cost between each node. Second, from starting node select next node by considering distance and gyration. Sometime Minimum Spanning Tree (MST) algorithm generate more than 31ink on one node. In this case, path is terminating the driving. The path should be trail or circuit. Repeat the above sequence until all nodes are connected. Before field test simulation was conduct for each path.

Results: In this paper, we applied and simulated MST and mMST algorithms. The cost efficiency according to the turning radius was experimented. Two paths (weight: distance, distance and radius) were generated by each algorithm and the results were compared. Experimental results show that mMST is more superior compared with MST.

Discussions: Future challenges remain to apply this research to real fields. Real field has difficulty to driving because ground condition are uneven and crop damage must be minimized.

Conclusion: The proposed mMST method may resolved that sudden change of direction between the path point causes a large error in the driving of the vehicle. And satisfactory results have been obtained.