Construction of Autonomous Vehicle Trajectory Tracking System Under Artificial Intelligence Technology

In response to the problem of difficulty in accurately tracking the actual path of traditional unmanned vehicles, this paper introduces the deep Q-network algorithm to study it. This article introduces the current research achievements in trajectory tracking of autonomous vehicles, and establishes the state space and action space of vehicles through the analysis of lateral and longitudinal forces, and constructs a vehicle dynamics model. This article optimizes the problems of classical deep Q-network algorithms, such as long training time and complex evaluation. Experimental testing shows that after 50 iterations, the reward of the deep Q-network algorithm model is 7.41, while the reward of the improved deep Q-network algorithm model is increased to 9.92, indicating a significant improvement in training effectiveness and tracking accuracy.