Reactive Obstacle Avoidance for the REMUS Autonomous Underwater Vehicle Utilizing a Forward Looking Sonar

摘要

One day fully autonomous AUV's will no longer require human interactions to complete its missions. To make this a reality, the AUV must be able to safely navigate in unfamiliar environments with unknown obstacles. This thesis builds on previous work conducted at NPS's Center for AUV Research to improve the autonomy of the REMUS class of AUVs with an implemented FLS. The first part of this thesis deals with accurate path following with the use of look-ahead pitch calculations. With the use of a SIMULINK model, constraints surrounding obstacle avoidance path planning are then explored, focusing on optimal sensor orientation issues. Two path planning methods are developed to address the issues of a limited sonar field of view and uncertainties brought on by an occlusion area. The first approach utilizes a pop-up maneuver to increase the field of view and minimize the occlusion area, while the second approach creates a path with the addition of a spline. Comparing the two methods, it was concluded that spline addition planner provided a robust optimal obstacle avoidance path and along with the look-ahead pitch controller completes the design of a back-seat driver to improve REMUS s survivability in an unknown environment. REMUS, AUV, UUV, Autonomous Underwater Vehicle, Reactive Obstacle Avoidance, Forward Looking Sonar, Vertical Plane, Pitch Controller, Spline, Gaussian, Occlusion, Optimal Sensor Orientation.