Obstacle avoidance and target search of an Autonomous Surface Vehicle for 2016 Maritime RobotX challenge

This paper describes algorithms for Autonomous Surface Vehicle(ASV) obstacle avoidance and target search task. This work is primarily designed for task mission of 2016 Maritime RobotX competition. In this task, ASV must avoid obstacle buoys, while it is searching for totem-shaped buoy. To deal with such problem, algorithms for both perception and motion planning stage was designed. In perception stage, 2D scanning LIDAR and monocular vision sensor are used to detect any floating objects on the water. To recognize target (totem buoy), HSV color space information was used. Detected object (both obstacle and totem buoy) information is tracked by using Kalman filter. In action planning stage, both deliberative and reflexive action [2] planning are designed. In deliberative action planning stage, based on the Kalman filter tracked obstacle information, a grid map can be generated. Using the grid map and A∗ search algorithm, desired path for searching totem can be calculated. In reflexive action planning stage, once ASV accidently enters hazardous region, where it has high probability of getting collide, it is designed to reflexively escape the region by making pure sway motion. An ASV experiment was performed to validate proposed method.