Research On Indoor Positioning And Navigation Of Mobile Robot Based On Monocular Visual-inertial Information Fusion

Vision sensors have been widely used in all aspects of modern society because of its advantages of obtaining rich environmental information,strong visualization and high stability.Monocular cameras have the advantages of small size,low price and unlimited by the range of distance.Because of these advantages,It has become a hot spot in scientific research.Due to the limitations of monocular vision positioning and navigation,inertial information can complement each other.In this paper,a new and lightweight indoor positioning and navigation method based on robot-based monocular vision-inertial information fusion is proposed.Through the effective combination of monocular vision and inertial information,the mobile robot can achieve simple and accurate positioning and navigation in the indoor.The specific research content is as follows:Firstly,the concept of intelligent robot and the importance of indoor positioning and navigation of robot are expounded.This paper summarizes the widely used indoor positioning and navigation methods,introduces the visual positioning and navigation technology,and analyzes the research status of the related methods at home and abroad.Secondly,the inertial information self-positioning navigation technology is introduced,and its principles,advantages and disadvantages are analyzed.The principle of monocular vision self-positioning is developed.On the basis of the monocular vision mathematical model,the motion of the camera is estimated by combining the feature description and the principle of antipolar geometry.The indoor self-positioning navigation system of mobile robot based on monocular vision is designed.Based on ROS system,how to design and build a mobile robot to meet the needs of its own experiment is described in detail.Thirdly,an indoor positioning and navigation method based on monocular vision-inertial information fusion with robot as reference frame is proposed.Considering the difference of the output rate between inertial information and monocular visual data and the computational complexity,the dual asynchronous Kalman filter is proposed to fuse the position and attitude data of the robot and a simple nonlinear expression of the pose information of the robot is given by using the approximate idea.Finally,the mobile robot platform required by our experiment is designed and built.On this basis,the indoor positioning and navigation experiment of the mobile robot is carried out.The results show that the system is effective and accurate.