Trajectory optimization method and device

Abstract

The embodiment of the invention provides a trajectory optimization method and device. The method comprises the following steps that: an original path is determined; discrete sampling is carried out onthe original path to obtain a plurality of sampling points, the maximum drivable area corresponding to each sampling point is determined, wherein the maximum drivable area is described by adopting alinear constraint condition; path optimization is performed on the original path according to the linear constraint condition to obtain an optimized path; speed optimization is performed based on theoptimization path to obtain a speed optimization curve; the optimized path is combined with the speed optimization curve, so that an optimized trajectory can be obtained. Compared with calculation ofthe distance between two convex polygons, the calculation of the maximum drivable area is lower in complexity, and therefore, path optimization can be rapidly achieved; and therefore, trajectory optimization can be quickly carried out, and the trajectory optimization efficiency is improved.

Description

Technical field [0001] The present invention relates to the field of automobile technology, in particular to a trajectory method and device. Background technique [0002] With the development of artificial intelligence, artificial intelligence has also been widely used in various fields, such as the vehicle field. Vehicles use artificial intelligence in many aspects, such as voice assistants, autonomous driving (also known as unmanned driving) and so on. Autonomous driving realizes the control of vehicle driving by combining sensors, machines and artificial intelligence to improve the safety and efficiency of the entire driving behavior. [0003] In the process of automatic driving, it is necessary to plan the trajectory of the vehicle, and then control the vehicle to drive according to the planned trajectory. At present, trajectory planning methods are mainly divided into sampling, searching and optimal control. Although the optimal control-based planner can achieve the continu...

AI Technical Summary

Problems solved by technology

Although the initial guess of the fast search helps the numerical solution process, there are still a large number of non-convex collision avoidance constraints in the formulated optimal control problem, which makes the optimization slow when the initial guess is not close to optimal

In addition, in the traditional collision detection, the vehicle needs to perform collision detection with all obstacles in the environment at each sampling position. Even if only nearby obstacles can be searched at the sampling position, the scale of collision detection can be reduced, but collision avoidance is The optimal control problem is usually expressed as the minimum distance between two convex polygons is greater than the set distance threshold, but the distance calculation between two convex polygons involves complex mathematical operations and has a lot of time consumption

Images