Novel autonomous parking positioning and navigation method based on point cloud registration

Abstract

The invention discloses a novel autonomous parking positioning and navigation method based on point cloud registration, and the method comprises the following steps: S1, building a laser point cloud map: carrying out the mapping of an underground garage through a vehicle-mounted laser radar, and generating the laser point cloud map; s2, real-time positioning: creating a laser point cloud map according to the step 1, fusing NDT point cloud registration information and inertial measurement unit information by using unscented Kalman filtering, and calculating to obtain the current pose of the vehicle; s3, planning a path: according to the current pose of the vehicle calculated in the step 2, generating a parking path by using a predefined parking planning method; and S4, tracking the trajectory: controlling the vehicle to follow the expected path by using a self-adaptive MPC transverse controller. According to the method, the underground garage laser point cloud map is created in advance, the pose of the automatic driving automobile in a coordinate system is obtained in real time, the parking path is generated through a predefined parking planning method, and finally the transverse controller and the longitudinal controller are used for controlling the automobile to track the planned path.

Description

technical field [0001] The invention relates to the technical field of automobile autonomous parking, in particular to a novel autonomous parking positioning and navigation method based on point cloud registration. Background technique [0002] At present, driven by related technologies such as computers and sensors, autonomous driving technology has developed rapidly, and has been effectively applied in fields such as adaptive cruise control (ACC) and autonomous valet parking (AVP). The primary task of an autonomous vehicle is to achieve flexible autonomous positioning and navigation based on its own sensors. However, various sensors are limited by different spatial environments. To complete autonomous positioning and navigation tasks, different solutions need to be applied in different environments. [0003] Due to the different scenarios faced by self-driving cars, some scenarios require high positioning and navigation accuracy, such as urban roads, underground garages, e...

AI Technical Summary

Problems solved by technology

The positioning accuracy of the global satellite positioning system (GPS) is generally at the meter level when the signal is good, but in environments such as buildings and underground garages, the signal cannot even be received, which does not meet the application requirements

[0004] In addition, there have been visual navigation methods based on two-dimensional codes such as ribbons, barcodes, and QR codes. This method usually uses colored ribbons laid on the ground as a guide and uses the location information contained in the two-dimensional codes for positioning to achieve navigation. The ribbons and QR codes on the screen are easy to be stained, and when the task path changes, the ribbons and QR codes need to be pasted again, which has poor flexibility and high environmental requirements

In addition, this method often requires that the direction of the optical axis of the camera is perpendicular to the plane on which the two-dimensional code is laid, and that the distance between the camera and the plane on which the two-dimensional code is laid is very close, otherwise the camera will not be able to recognize

have greater limitations

Images