Algorithm for planning lane changing track of automatic driving vehicle based on potential energy field heuristic search

Abstract

The invention relates to an algorithm for planning a lane changing track of an automatic driving vehicle by applying a potential energy field theory. The algorithm is applied to automatic driving lane changing track planning in a complex road environment. A potential energy field model of the vehicle risk is constructed by considering the position influence of the environmental vehicle and the vehicle motion trend; the lane line and the road boundary potential energy field in the road environment are described by Gaussian distribution and inverse proportion functions respectively. On the basis of a potential energy field, an unacceptable minimum risk threshold of a vehicle is calibrated, a vehicle track non-intrusive area is divided, and an A-star heuristic search algorithm is adopted to search for a path with the minimum risk value. According to the method, the actual trajectory of the vehicle is solved by adopting a self-adaptive model prediction control method, and the planned trajectory is tracked. According to the invention, the adaptability of automatic driving lane changing trajectory planning to complex roads and traffic environments is improved.

Description

technical field [0001] The invention belongs to the field of automatic driving trajectory planning algorithms. Background technique [0002] Commonly used lane-changing trajectory planning models for autonomous driving use polynomials for fitting, such as quintic polynomials, to calculate safe and comfortable trajectories for unmanned vehicles to complete predetermined driving tasks. Another method is the trajectory planning method based on deep learning, which imitates human driving behavior to complete trajectory planning by training human driving trajectory. These methods generally consider factors such as vehicle clearance and vehicle speed or set a series of safety rules to prevent collisions with dynamic vehicles. However, there is a lack of consideration of road conditions, such as curving road linearity, ramp acceleration lane length and other road constraints; on the other hand, the influence of traffic rules, such as right of way, lane speed limit and intersection...

AI Technical Summary

Problems solved by technology

[0004] The shortcomings of the existing research on lane-changing trajectory planning for autonomous driving are as follows: (1) Only considering the risk of interactive vehicles, lacking a comprehensive description of the complex road environment

(2) The trajectory generated by the gradient descent method based on the potential energy field may not meet the requirements of the vehicle kinematics model, so it will not be executed

Images