AI Qing lin1, YU Jie1, HU Ke yong1,2, CHEN Qi1

 (1.Key Laboratory of E&M, Ministry of Education & Zhejiang Province, Zhejiang University of Technology, Hangzhou 310014, China；

2.Hangzhou Normal University Qianjiang College, Hangzhou 310036, China）

 

 

Abstract: Aiming at the problems of the decline in real time and robust performance of robot simultaneous localization and mapping(SLAM) in a complex environment, an optimization frame of SLAM based on improved ORB(oriented FAST and rotated BRIEF) keyframe detection and matching algorithm was proposed. After the analysis of keypoint detection, frame matching, motion estimation and loop closure detection algorithm, the relationship between keyframe loop closure matching algorithm and SLAM system was established. An improved ORB algorithm was adopted to implement the fast and efficient matching between two adjacent RGB frames. Combined camera perspective projection model with dense frames, the 3D color point clouds can be transformed from adjacent matched 2D frames. Then the relative pose between the adjacent frames was computed by improved RANSAC ICP algorithm, which can solve the mobile robot precise localization problem. The keyframe Bag of Word algorithm was the basis of loop closure detection, which can improve the mapping speed and consistency. The purpose of closure detection was to reduce redundant model structure and generate a map with consistency. The real time and robust performance were evaluated by matching speed and root mean square (RMSE) of the absolute trajectory error (ATE). The Results base on standard testing indicate that the robot can build a precise environment model where the robot can localize itself real time and robustly.

 

Key words: simultaneous localization and mapping(SLAM); keypoint detection; keyframe selection; closure detection