Robust stereoscopic vision inertial pre-integration SLAM (Simultaneous Localization and Mapping) method

Abstract

The invention discloses a robust stereoscopic vision inertial pre-integration SLAM (Simultaneous Localization and Mapping) method and belongs to the technical field of visual navigation, wherein inertial sensor measurement value is used for carrying out pre-integration forecasting carrier navigation information, the information is fused with data of a stereoscopic vision sensor to judge the imagematching quality, the pre-integration information is fused to output different results according to a judgement result, the constructed map can be multiplexed, and accumulated errors are eliminated. According to the method in the invention, the problem of visual SLAM matching errors and failure in a bad environment can be effectively solved, the high-robustness and high-precision carrier navigation information is obtained, a precise map can be constructed for multiplexing, and a good popularization prospect is achieved.

Description

technical field [0001] The invention relates to a robust stereo vision inertial pre-integration SLAM method, which belongs to the technical field of visual navigation. Background technique [0002] Real-time localization and mapping (SLAM, Simultaneous Localization and Mapping) technology is a major research hotspot in the field of robot autonomous navigation technology, and it is also a necessary key technology in the application of robots in engineering practice. The visual sensor can perceive the light of the outside world without its own signal transmission, and has high autonomy. Visual SLAM has broad development prospects and great potential in robot SLAM. [0003] At present, most visual SLAMs have high requirements on the external environment, and need to use environmental information for visual calculations. In the entire SLAM process, the environment needs to be kept in a controllable state, otherwise it will cause visual calculations to fail and lead to navigation...

AI Technical Summary

Problems solved by technology

[0003] At present, most visual SLAMs have high requirements on the external environment, and need to use environmental information for visual calculations. In the entire SLAM process, the environment needs to be kept in a controllable state, otherwise it will cause visual calculations to fail and lead to navigation. Failed to output information and cause harm to the robot

However, in the real environment, there are often changes in light and shade and imaging noise caused by motion.

In the robot SLAM method, it is very sensitive to changes in the external environment, and its changes will directly lead to a decrease in accuracy, resulting in large errors and even system crashes.

Images