Research on path planning of robot based on deep reinforcement learning

摘要

In this paper, to avoid the problem of local optimization and slow convergence in complex environment, a reinforcement learning algorithm is proposed to solve the problem. A robot path planning model is built and its feasibility is verified by simulation. In addition, this paper proposes a deep environment to neural network for robot camera to establish a deep reinforcement learning path planning model, and establishes a deep recursive Q-network (DRQN) and Deep Dueling Q-network(DDQN) respectively. In the comparison of the final simulation results, DRQN needs to consume more computation time, but can achieve better results with higher accuracy.