Trajectory Planning Algorithm Considering Obstacle Risk in Dynamic Traffic Scenarios

In the field of autonomous vehicles, efficient and safe trajectory planning is a crucial task. Planners must respond to environmental changes in real time and ensure driving safety and passenger comfort. To address this challenge, this paper proposes an innovative trajectory planning framework based on risk field modeling. First, the generation of reference lines is accelerated by identifying matching points, and an optimized cost function is developed to refine the reference line. On this basis, a discretized drivable area is constructed, and a risk field model is established to quantitatively assess risks posed by obstacles. The trajectory planning process is decoupled into separate path planning and speed planning phases. During the dynamic planning phase, collision risk assessment ensures safety when generating preliminary trajectories. Subsequently, the trajectory is optimized within the convex space boundary through quadratic programming, and the repelling line technique and adaptive lane change speed strategy are used to keep the trajectory smooth. Finally, simulation experiments validate the effectiveness of this framework, and demonstrate its ability to successfully avoid obstacles in dynamic traffic scenarios while maintaining passenger comfort.