Research on positioning and navigation of medical robot based on RGB-D visual SLAM

Abstract

With the increasing demand for medical service robots and the development of RGB-D-based visual SLAM technology, there is a growing interest in research on medical service robots that can navigate autonomously. This paper uses mathematical language to introduce the camera imaging principle, image feature matching algorithms, and pose estimation algorithms in visual SLAM in detail. We have selected an experimental scene in a targeted manner and have combined the improved ORB feature points and feature mismatch elimination algorithm (RANSAC) to enhance the original algorithm. We analyze and demonstrate the advantages and disadvantages of the algorithm before and after the improvement, and provide insights into future developments in autonomous navigation for medical service robots.