A comparison of two obstacle avoidance path planners for autonomous underwater vehicles

The breadth-first path planner is compared to the optimal A* path planner in terms of distance traveled and computation time. Certain modifications are added to the obstacle avoidance algorithm to increase its robustness to vehicle position uncertainties arising from the Doppler/INS system. Simulation results show that the A* path planner generates paths that are between 0.8 and 1.1 times the length of the paths generated by the breadth-first approach. Computationally, however, the A* solution takes anywhere from 1.5 to 19.8 times longer to calculate. Repeated trials in simulated, dense obstacle fields have verified that despite position uncertainties, the obstacle avoidance algorithm safely guides an autonomous underwater vehicle along a planned path to the desired end point. >