Automatic driving trajectory generation method in high-speed scene

Abstract

The invention relates to a method for generating an automatic driving trajectory in a high-speed scene, comprising the steps of: constructing a two-dimensional coordinate system; obtaining original road data and original vehicle state data; obtaining optimal path data; obtaining several sampling points; generating candidate trajectories; calculating all sampling points Cost value between points; filter out the best path data and output; get the target trajectory and output it to the control system for execution. The invention adopts the third-order Bezier curve, so that the unmanned vehicle can reach the target point along a smooth trajectory when changing lanes, and the curvature of the curve is continuous; for the high-speed automatic driving scene, the Dijkstra algorithm is improved accordingly. On the basis of the path, the amount of calculation is greatly reduced, which meets the real-time requirements under high-speed conditions, and the results are continuous; the vehicle motion is constrained by the Cost function model, and the optimal solution is obtained; after considering the constraints in the actual motion of the vehicle Build road and motion models to better suit practical application needs.

Description

technical field [0001] The invention relates to the technical field of automatic driving, in particular to a method for generating automatic driving trajectories in high-speed scenarios. Background technique [0002] As the central brain of the vehicle, the unmanned vehicle decision-making planning system outputs driving behavior instructions through the decision-making layer after receiving the environmental perception information, and then performs trajectory planning. Trajectory planning generates a series of trajectory curves from the current state to the next goal state. Trajectories need to satisfy the vehicle's motion and dynamic constraints to avoid collisions with obstacles in a realistic urban environment. [0003] The current mainstream trajectory generation methods are divided into two categories: the first category is based on curve interpolation methods, such as straight lines and arc segments, polynomial curves, clothoid curves, and spline curves, etc., which...

AI Technical Summary

Problems solved by technology

The advantage of this type of method is that it is intuitive, simple, and easy to think of, but the disadvantages are very obvious:

[0004] 1. Although the straight line and arc segment are simple to implement and have low calculation load, the curvature is discontinuous and the node segments are connected;

[0005] 2. The polynomial curve has the advantages of low calculation amount and continuous curvature, but the continuous curvature requires more than four orders, and it is difficult to determine the key parameters;

[0006] 3. The clothoid curve is suitable for local planning, but the integration is time-consuming, the curvature is continuous but not smooth, and depends on the trajectory point;

[0007] 4. The spline curve also has the advantages of low calculation amount and continuous curvature, but the result may not be the optimal solution

[0009] 1. Although the improved Dijkstra algorithm is suitable for finding the shortest path in a structured environment, it does not meet the real-time requirements due to the large amount of calculation when used alone, and the results are not continuous;

[0010] 2. The advantage of the A* algorithm is that it is instructive and can quickly obtain the global optimal trajectory. However, the search step size is difficult to determine, and the efficiency is low in complex and large-scale planning environments.

This method seems to solve the above problems, but its theoretical basis is based on simplifying the unmanned vehicle to a point to establish the road and motion model. The defect of this is that the vehicle attitude information is ignored, and the decision planning Taking into account the behavior instructions of the layer, the unmanned vehicle function of the trajectory planning method based on this method is simple, and at the same time, it cannot effectively control the vehicle under complex working conditions.

Images