DF -FS based path planning algorithm with sparse waypoints in unstructured terrain
Ning Li, Xijun Zhao, Ruina Dang, Bo Su · 2018
For autonomous navigation in unstructured terrain, it's difficult to plan the path with traditional algorithms based on searching guidance waypoints such as ARA* and Field D* as a result of the poor positioning signal and the environmental uncertainty especially when dense waypoints are not given. Because of irregular road boundaries, there are also problems for path planning based road detection methods. To solve these problems, the depth first forward search (DF-FS) based path planning algorithm which combines forward searching along the road and model-based prediction is proposed. DF-FS based planner takes road boundary constraint, road direction and heading of vehicle as the basis of path planning, while using model prediction optimizes planned path to improve navigation stability. The experiment results show that in unstructured terrain with sparse waypoints or even without waypoints in the one-way road, the autonomous navigation can be guaranteed effectively. Meanwhile the proposed algorithm has a strong robustness to the uncertainty of positioning and environmental perception, so it is suitable for autonomous navigation in rough terrain.