Diagonal parking space detection method and device and automatic parking method and system

Abstract

The invention provides a diagonal parking space detection method and device and an automatic parking method and system. The diagonal parking space detection method comprises the following steps: determining a target diagonal parking space in a plurality of diagonal parking spaces arranged along a reference line; obtaining a plurality of radar ranging values between a vehicle to be parked and a plurality of reflection points on a first reference vehicle and a second reference vehicle in a set time; forming a two-dimensional data image with a vehicle running distance as an abscissa and the radarranging values as an ordinate; generating a two-dimensional gray scale image according to the two-dimensional data image as input of a parking depth neural network, and respectively outputting coordinates of three characteristic points and an included angle between the target diagonal parking space and the reference line after treatment of the parking depth neural network; and calculating the width of the diagonal parking space. According to the method, diagonal parking spaces meeting the parking condition can be automatically searched for, the accuracy and the robustness of characteristic points of the diagonal parking spaces are improved by using the depth neural network, and the defect that only parallel parking spaces and vertical parking spaces can be processed in the prior art is overcome.

Description

technical field [0001] The invention relates to the technical field of automotive electronics, in particular to a detection method and device for diagonal parking spaces, and an automatic parking method and system. Background technique [0002] At present, the research on the parking algorithm of parallel parking spaces and vertical parking spaces is mainly carried out in the field of automatic parking. For inclined parking spaces with uncertain angles, since the extraction of feature points is much more complicated, it is necessary to use sensors to detect the inclination angle and angle of the parking space before parking. However, due to the limited information obtained by sensors, it is difficult to extract parking angle information through conventional algorithms. As a technology that has only become popular in recent years, artificial intelligence deep learning technology is currently lacking in applications in the field of automatic parking. [0003] A known tilted pa...

AI Technical Summary

Problems solved by technology

However, the applicable object of this scheme is side parking, and the calculated tilt angle is only used to correct the coordinates of the reference point, and cannot be used to calculate diagonal parking spaces

Images