Localization method and system based on stroke sensing and visual fusion, terminal and storage medium

Abstract

The invention provides a localization method and system based on stroke sensing and visual fusion, a terminal and a storage medium. In a preparation stage, visual data and stroke data are acquired; incremental corresponding relationship conversion is conducted, wherein an initial position is determined, and a conversion relationship for an extracted increment in the stroke data is obtained by stroke sensing equipment between a later moment j and a previous moment i; constraint is conducted, wherein visual sensing equipment is constrained with the conversion relationship for the extracted increment of the stroke sensing equipment as a constraint condition. The stroke sensing equipment is adopted for sensing the length and heading angle information of the movement of a vehicle body, the length or trajectory information sensed by the stroke sensing equipment serves as the trajectory constraint of a camera, the problem of scale uncertainty of the monocular camera is solved, and visual localization error accumulation and scale drift can also be prevented. In addition, the good localization effect can be achieved on non-planar scenes instead of being limited to planar scenes simply by adopting the length constraint of stroke sensing.

Description

technical field [0001] The present invention relates to the technical field of automotive electronics, in particular to a positioning method, system, terminal and storage medium based on travel perception and vision fusion. Background technique [0002] Simultaneous localization and mapping (SLAM for short) centered on vision is a technology that simultaneously calculates the position, orientation (pose) of the camera and the three-dimensional coordinates of the environment point cloud through image matching. For the monocular camera, there is an unknown scale factor (scale) between the calculated trajectory and the real trajectory length, which makes the positioning and mapping results lack of practicability; in addition, the positioning results of the monocular camera are also prone to scale drift, resulting in It is unavailable in a wide range; the present invention uses the vehicle wheel speed pulse odometer as a scale information constraint, so that the positioning and ...

AI Technical Summary

Problems solved by technology

[0016] The above is the existing technology closest to this article, and also uses the pose constraints provided by the wheel odometer to constrain and optimize the pose of the camera, but these methods without exception require the odometer to run on a plane, and its objects are There is a robot encoder odometer with higher precision, but this method is based on the inherent odometer of the car, and its accuracy is lower than that of the robot encoder. The scheme described in [4-6] is not practical here. What is important is that the workaround based on formula (8) can also be used when the car is climbing, thus breaking through the constraints of the previous papers

Images