Research on PPO algorithm in solving AUV path planning problems
Zheng Wang, Zhiyuan Hu, Yang Yang, Yang Yin · 2021 2nd International Seminar on Artificial Intelligence, Networking and Information Technology (AINIT) · 2021
In the face of a complex three-dimensional environment, the computational complexity of the traditional path planning algorithms is extremely increased. But the performance of path planning results is relatively poor. Deep reinforcement learning (DRL) algorithms don’t rely on accurate environmental models, and its overall efficiency is much higher than traditional algorithms. Aiming at the problem of AUV path planning in a three-dimensional environment, proximal policy optimization algorithm (PPO) is selected. The PPO algorithm with improved network structure is used to train the model based on the established AUV model and gym environment. Through simulation experiments, the accuracy and effectiveness of the algorithm is verified.