Design of Radial Basis Function Network and State-Dependent LQT for Path Planning and Tracking of Autonomous Underwater Vehicle (AUV) to Intercept A Moving Target

The autonomous underwater vehicle (AUV) has been implemented in various fields, such as to monitor deep and long-distance sea conditions, underwater mapping, long time surveys, neutralization of seabed mines, and military applications. In the military field, one of the AUV applications is a torpedo. A torpedo has a mission to go to a given target, frequently a moving target. To meet this objective, the torpedo must have a detection system to find out the target position. However, it is impossible to be implemented because of the limited space and energy available in a torpedo. Therefore, it requires a reliable system to predict a moving target, to form the path to hit the target, avoid obstacles in the path, and the controller that can follow the path determined by the path planning algorithm. Therefore, this study developed various scenarios to simulate the conditions. The results show that the radial base function network can form a path that is free from obstacles and able to form a path that can intercept a moving target while the state-dependent LQT algorithm is able to perform a path tracking properly so that the AUV can reach a predetermined target. With the speed of 1.4 m/s, the AUV could intercept a moving target with the speed of 0.2 m/s and the face angle of -1.7 radians at t = 75.1 seconds. In another scenario, the AUV intercepted a moving target with the speed of 0.5 m/s and the face angle of -1.18 radians at t = 73.2 seconds.