Autonomous collision avoidance method for inland river unmanned vessels based on backward reachable set

This paper constructs a safe area model for unmanned vessels based on the concept of Backward Reachable Set, and combines Backward Reachable Set and model predictive control to carry out research on autonomous collision avoidance for unmanned vessels in inland river scenarios. Taking the relative position and relative dynamics system of the vessel and the target vessel as inputs, the calculated Backward Reachable Set is regarded as the dynamic safety region of the vessel, and the model predictive control (MPC) is utilized to inject the Backward Reachable Set into the low-level controller in advance through the high-level design of the preset trajectory, which enables the vessel to complete autonomous collision avoidance in a safe and effective manner, while not approaching excessively to the target vessel, and avoids the failure of decision-making in the extreme scenarios. The concept of Backward Reachable Set is introduced on the basis of the existing obstacle avoidance methods, considering the uncertain environment in the inland navigation scenario, and the practicality of the method is verified through model analysis and scenario experiments, which provides a new idea for the subsequent research on autonomous obstacle avoidance methods for inland waterway vessels.