Aiming at the problems that traditional trajectory tracking control algorithms of crawler robots require high model accuracy and have poor control effects

an algorithm of trajectory tracking control based on twin delayed deep deterministic policy gradient(TD3)in deep reinforcement learning is proposed.Firstly

the kinematic model of the crawler robot is established.Secondly

an algorithm of trajectory tracking control based on TD3 is proposed.The TD3 algorithm is used as the update strategy of the agent.In order to complete the training of the agent

the reward function and neural network structure are designed

and the hyperparameters are optimized.Finally

the proposed algorithm is simulated and compared with the deep deterministic policy gradient(DDPG)algorithm

linear quadratic regulator(LQR)and model predictive control(MPC)algorithm.The results show that the algorithm of trajectory tracking control algorithm based on TD3 has higher tracking precision and better stability in three typical working conditions.