Research On VI-SLAM System For Wearable Scene

Simultaneous localization and map construction,or SLAM,refers to the use of sensor data measurement in an unknown environment to locate one’s own posture and state,and to perceive and compose the surrounding environment.The current mobile robots cannot replace professionals to complete tasks quickly and accurately in complex scenarios,but wearable devices combined with SLAM technology can achieve accurate real-time positioning of wearers under highly dynamic conditions,while sending and recording environmental information,effectively assisting Users perform special tasks in unknown environments.However,wearable computing platforms usually have very limited computing power and are more sensitive to power consumption.Moreover,compared with the regular motion pattern of the robot,there are more uncertainties in the motion of the human body,and the tracking robustness of the SLAM system is required to be higher.Therefore,in order to achieve the goal of efficiently and robustly completing positioning,navigation and mapping in wearing scenarios,the following research work is mainly carried out in this paper:Firstly,the SLAM system of binocular-inertial fusion is researched and analyzed,and the high robustness requirement of SLAM system in wearable scenarios is solved by means of binocular and inertial tight coupling.A set of binocular-inertial fusion SLAM system is designed and implemented around the tightly coupled scheme.Secondly,in order to make the algorithm more suitable for wearable computing platforms,a lightweight method based on the motion state of the camera is proposed.The data measured by the IMU(Inertial Measurement Unit)is used to estimate the motion state,and different data processing methods are adopted in different motion states,which can effectively It ensures the accuracy of the system,reduces the amount of redundant data in the system,avoids unnecessary computing tasks,and improves the performance of the system.Finally,based on binocular stereo vision and SLAM technology,the indoor 3D point cloud construction method is studied.The scene depth information obtained by binocular stereo vision and the pose data estimated by the SLAM system are combined to construct a scene 3D dense point cloud,so as to realize the intuitive display of indoor 3D scenes.