Trajectory Planning for Autonomous Vehicles at Intersections Based on the Spatial Structure

Intersection trajectory planning constitutes a fundamental challenge for autonomous vehicles, with spatial structure elements such as geometry, static obstacles, and conflict points critically determining both path geometry and speed profiles. Distinguished from existing studies applied in simple crossroad scenarios where the spatial structure is often overlooked, this study proposes a flexible and effective method for intersection that explicitly incorporates spatial structure. Our approach decomposes the problem into path planning and speed planning. First, we analyze and classify the intersection’s spatial structure components that affect planning into visible and invisible elements. Then, the cubic quasiuniform Basis‐spline (B‐spline) curve and cost function are applied to generate the optimal, static‐obstacle‐free path. In the subsequent speed planning stage, the traffic rule of conflict points is used to generate the reference speed, and quadratic programming is applied to generate the dynamic‐obstacle‐free speed curve in the spatiotemporal (ST) graph. To validate the effectiveness of this approach, experiments are conducted in four static intersections and six dynamic scenarios with different spatial structures. Compared with other methods like A ∗ and dynamic programming, this approach is more applicable and efficient. The results demonstrate the flexibility and practicality of this method, which can govern the continuity and smoothness of road connection points and ensure safety under the constraints of different intersection elements.