Pose prediction method, pose prediction device and robot

Abstract

The invention discloses a pose prediction method, a pose prediction device, a robot and a computer readable storage medium. The method is applied to the robot provided with binocular cameras, and the binocular cameras comprise a first camera and a second camera. The method comprises the following steps: searching at least one pair of matched feature points in a first image collected by the first camera at the current moment and a second image collected by the second camera at the current moment; searching at least one pair of matched line features in the first image and the second image; and obtaining a predicted pose of the robot based on the at least one pair of line features, at least one pair of feature points and inertial data output by an inertial measurement unit of the robot. Through the scheme of the invention, more accurate and robust robot pose prediction can be realized.

Description

technical field [0001] The present application belongs to the technical field of visual algorithms, and in particular relates to a pose prediction method, a pose prediction device, a robot, and a computer-readable storage medium. Background technique [0002] In the process of robot positioning and mapping, it is necessary to obtain accurate pose prediction of the robot. In the prior art, the visual sensor is often relied on, and the feature point method based on feature point matching or the direct method based on photometric consistency are used to construct constraint conditions and realize the pose prediction of the robot. However, in some indoor scenes, such as low-texture environments or dark-light environments, it is difficult to extract effective feature points using the feature point matching-based feature point method, and the usability of the direct method based on photometric consistency is also greatly reduced. That is, in weak texture and dark light scenes, it...

AI Technical Summary

Problems solved by technology

However, in some indoor scenes, such as low-texture environments or dark-light environments, it is difficult to extract effective feature points using the feature point method based on feature point matching, and the usability of the direct method based on photometric consistency is also greatly reduced.

That is, in weak texture and dark light scenes, it may not be possible to accurately predict the pose of the robot

Images