High-precision map visual positioning method for automatic driving and automatic driving method

Abstract

The invention discloses a high-precision map visual positioning method for automatic driving. The method comprises the following steps: acquiring sensor information and confirming an initial position of a vehicle; acquiring an environment image and extracting multi-scale spatial features of the image; converting the image from a pixel plane to a BEV plane and carrying out image segmentation; and extracting data from an image segmentation result and fusing the data with a GNSS positioning result to obtain a high-precision positioning result of the current vehicle. The invention further discloses an automatic driving method comprising the high-precision map visual positioning method for automatic driving. The current high-precision position of a vehicle is obtained by filtering and fusing a visual matching positioning result and a GNSS positioning result in allusion to the situation that position drifting and the like occur in weak GNSS or GNSS-free signal area positioning; therefore, the method can provide a low-cost high-precision positioning result for the autonomous vehicle, improves the positioning robustness, and is high in reliability, good in continuity, precise and stable.

Description

technical field [0001] The invention belongs to the field of digital signal processing, and particularly relates to a high-precision map visual positioning method for automatic driving and an automatic driving method. Background technique [0002] With the development of economy and technology and the improvement of people's living standards, autonomous driving technology has been widely used in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable operation of autonomous driving has become the focus of research. [0003] The core technology system of autonomous driving is divided into three modules: perception, decision-making and execution. Among them, positioning is a very important part of the perception module; positioning technology is used to determine the precise position of the vehicle on the map: only when the vehicle is accurately positioned on the map, the system can better assist ...

AI Technical Summary

Problems solved by technology

However, the current technology still has the following defects: (1) The lidar sensor itself is expensive, and in tunnels or snowy weather, etc., the effect will be affected to a certain extent, thus affecting the positioning results; therefore, the lidar sensor cannot provide low-cost and robust positioning results; (2) The matching positioning process based on the visual sensor needs to project the positioning layer from the pixel plane to the BEV (Bird's EyeView, bird's-eye view) plane for positioning; this projection process currently mostly uses IPM However, the IPM (InversePerspective Mapping, inverse perspective transformation) projection method has perspective noise, and the farther the distance from the scene is, the greater the error is; at the same time, the IPM projection method generally sets the ground as a plane, but this setting does not apply to many scenes. It is not established, so the projection error of the IPM projection method is relatively large, which leads to low positioning accuracy and cannot provide high-precision positioning results

Images