Adaptive DWA algorithm with decision tree classifier for dynamic planning in USV navigation

Unmanned surface vehicles (USVs) plays a vital role in marine exploration, with obstacle avoidance technology being one of the key techniques for ensuring the safe navigation of USVs. This paper aims to investigate a decision tree-based dynamic window approach (DT-DWA) algorithm to address the obstacle avoidance strategy problem in autonomous navigation of USV. The primary objective of this algorithm is to fuse decision tree classifier into DWA algorithm, and consider some characteristics that affect decision judgment, which can achieve adaptively adjust weights. Secondly, we combine Distance of Closest Point of Approach (DCPA) and Time to the Closest Point of Approach (TCPA) as the basis for USV to make obstacle avoidance decision. Finally, the rationality and effectiveness of the proposed algorithm is demonstrated through simulation experiments.