Autonomous vertical parking method based on high precision map

Abstract

The invention provides an autonomous vertical parking method based on a high-precision map. The method comprises the steps that a vehicle acquires parking space information based on the high-precisionmap and environmental information of a parking lot and road edge road measurement sensors, position attributes of the vehicle to be parked and positional relationship attributes of the vehicle in a global coordinate system are determined, switching points are determined accordingly, the switching points are fitted through an arc or spline curve, a cost function is constructed under the conditionthat the requirements for vertical parking path constraints and the minimum parking space constraint are met, and a trajectory with the smallest cost function is selected as the final optimal trajectory. By means of the method, the smoothness of vehicle movement control can be ensured, shaking of a steering wheel of the vehicle in the parking process is eliminated, and safe parking is achieved.

Description

technical field [0001] The invention relates to the technical field of automatic driving, in particular to an autonomous vertical parking method based on a high-precision map. Background technique [0002] The parking problem in the city is mainly divided into parking on the road edge and parking lot. Autonomous vertical parking trajectory planning is an important technical component of the parking system. The current position and attitude of the neighborhood to the position and attitude of the parking end point provide the required input. [0003] At present, the vertical parking trajectory mainly considers the minimum turning radius of the vehicle, and uses geometric drawing to express the parking trajectory as a mathematical model. The method of generating the optimal trajectory offline under the consideration of obstacles and vehicle kinematics constraints is too large. Single, it is impossible to achieve real-time planning for vertical parking, which is a complex auton...

AI Technical Summary

Problems solved by technology

[0003] At present, the vertical parking trajectory mainly considers the minimum turning radius of the vehicle, and uses geometric drawing to express the parking trajectory as a mathematical model. The method of generating the optimal trajectory offline under the consideration of obstacles and vehicle kinematics constraints is too large. Single, it cannot achieve real-time planning for complex autonomous driving scenarios such as vertical parking in a narrow space environment, nor can it adapt to more complex parking environments

Through the neural network and the fuzzy controller, only imitating the driving behavior of skilled drivers is used as the prior information of the reference trajectory. The trajectory varies from driver to driver, and lacks flexibility and cannot dynamically adjust the vertical parking path.

Images