Straightened velocity obstacle algorithm for inland ships collision avoidance

With the advancement of artificial intelligence, inland Maritime Autonomous Surface Ships (MASS) are gaining traction. While various algorithms exist for collision avoidance, many assumptions from open waters, such as constant speed and course, do not apply to inland waterways. Ships in these environments are constrained by riverbanks, making traditional methods like DCPA/TCPA-based risk evaluation and the Linear Velocity Obstacle (LVO)algorithm ineffective. To address this issue, we propose a Straightened Velocity Obstacle (SVO) algorithm for collision avoidance in inland waters. This algorithm is based on the observation that ships maintain a relatively constant distance from the riverbank or the channel's centerline. Unlike traditional methods that use geodetic coordinates, the SVO algorithm sets the channel's centerline as the y-axis and the distance to it as the x-axis, effectively straightening the meandering channel for better collision detection in Decision Space (DS). Simulations using traffic data from inland waters validate the proposed algorithm. Scenarios involving heading and overtaking in curved channels demonstrate that the SVO algorithm outperforms traditional LVO and CPA-based methods, demonstrating its effectiveness for MASS in constrained inland environments.