Collision Avoidance Method for USVs Based on the Artificial Potential Field and Elliptical Ship Domain

With the rapid development of unmanned vehicle technology, the shipping industry is facing increasingly severe navigation safety challenges. The significant increase in the number of ships at sea has made ports and waterways more congested. To address this, the International Maritime Organization has formulated the COLREGs to promote coordination and consensus in collision avoidance processes. However, approximately 80% of maritime accidents are caused by human error, highlighting the importance of developing unmanned surface vehicles (USVs). USVs can not only assist tugboats and pilots with boarding operations but also reduce the risk of human error. The APF method has been widely applied in collision avoidance technology for unmanned vehicles. APF technology transforms navigation into a potential field model, allowing ships to autonomously avoid obstacles. However, the traditional artificial potential field method often results in longer collision avoidance paths in dynamic environments or high-density navigation areas, thereby increasing risk. To solve this problem, this paper proposes an improved method based on the artificial potential field to guide ship collision avoidance. This method creates an elliptical field at the collision point of two ships. As the distance between the own ship and the collision point decreases, the semi-minor axis of the elliptical field lengthens. An attractive point is set at the edge of the semi-minor axis. The attractive force generated by this point and the repulsive force generated by the collision point overlap, forming a resultant force to guide the ship in collision avoidance. Compared to the traditional artificial potential field method without an attractive point, the improved method proposed in this work can reduce the collision avoidance path length by 12-24%. Most importantly, this method complies with COLREGs regulations, enabling quick alignment with existing ship collision avoidance rules and ensuring consistency. This improvement enhances the collision avoidance performance of unmanned vehicles.