Traditional trajectory tracking control algorithms have the inherent flaw of decoupling path tracking from dynamic obstacle avoidance, making it difficult to adapt to dynamic uncertainty environments. Although the TEB algorithm can dynamically avoid obstacles. it has high computational complexity and is difficult to balance real-time performance and trajectory smoothness in high-speed scenarios. We designed an improved pure pursuit local path planner that integrates pure pursuit algorithm and TEB algorithm. We constructed a kinematic model of a four-wheel differential drive robot and verified its obstacle avoidance and trajectory tracking performance based on the ROS system. The results show that the improved pure pursuit algorithm's local path is more in line with the global path and with good trajectory performance. Moreover it can avoid dynamic obstacles in a timely manner. The SCOUT MINI lite robot was used in real vehicle testing to verify the obstacle avoidance performance and operational stability of the improved pure pursuit algorithm in local path planning in closed-loop indoor scenes with corners, narrow paths, and obstacles.