Speed Planning for Autonomous Vehicles at Intersections

This paper studies the problem of speed planning for autonomous vehicles at intersections, which is of vital importance to the safety and efficiency of autonomous vehicles. Firstly, we formulate the problem in s-t graph. Then we design a numerical calculation method of speed planning with the help of s - t and v - t graph. With the subtle modification on Rapidly-exploring Random Tree (RRT) method, approximate optimal solution is obtained by exploring s - t space. We evaluate the proposed method with randomly generated scenarios, and the simulation results show that the autonomous vehicle controlled by the proposed method passes all of the test scenarios safely. Our main contributions are as follows: (1) RRT is modified to explore in s - t graph for speed planning. (2) The proposed st-RRT method could be carried out over 130 times per second. (3) The solution of the proposed method could reach approximately optimal concerning pass times, and even global optimal in some specific scenarios.