Obstacle detection and avoidance are crucial for the movement of autonomous vehicles; therefore, many algorithms for local path planning have been proposed in recent years. Among which, rapidly-exploring random tree (RRT) have received considerable attention due to the fact that they do not require complex mathematical modeling and can quickly identify obstacle avoidance paths in low-complexity environments. However, the performance of path planning decreases significantly as the complexity of the obstacle distribution increases. To conquer this problem, we propose in this paper a path planning algorithm that closely follows the obstacles. In addition to the path planning strategy for complex environments, a path smoothing algorithm is also proposed so that the carrier dynamics can be conformed. After extensive simulation tests, it is proved that the proposed algorithm can not only effectively reduce the number of nodes created, and thus reduce the burden of memory requirement, but also significantly outperforms other algorithms in terms of the average number of iterations, the average time for path planning, and the average path length, which is of high practical value.