Research On Application Of VSLAM Technology In Indoor Autonomous Cruise Vehicle

A series of vision sensors represented by RGB-D cameras have the advantages of low price,small size,and portability.Compared with sensors such as lidars,they are more suitable for small,low-cost mobile robots.In recent years,the research of mobile robots based on visual sensors has become a new wave of mobile robot research.Based on this background,this paper studies the VSLAM system in an indoor environment with RGB-D cameras as sensors,focusing on its real-time positioning,indoor 3D environment reconstruction,autonomous navigation and other key technologies,and builds an indoor autonomous cruiser platform.First of all,regarding the front-end visual odometer of the VSLAM system,this paper uses the small hole imaging model as the benchmark,uses the ORB algorithm to extract the feature points of the image collected by the RGB-D camera,and completes the feature matc hing through the optimized BF algorithm,and then uses P3P The algorithm realizes the initi al positioning of the camera;then,in view of the problem that pure visual odometer cannot adapt to fast movement and rotation,according to the idea of multi-sensor fusion,the Kalma n filter loosely coupled method is used to fuse visual odometer data and IMU data to realize vision-Inertial navigation odometer;then BA algorithm was used to optimize the pose in th e back-end optimization to obtain the optimal pose estimation in the long running process of the system,and for the problem of large accumulated errors when the system moves in a lar ge-scale environment,the Bow bag of words model is used to complete the loop detection;t wo different environmental maps are further generated according to the needs of the system;Finally,according to the generated map,a path planning system with~*algorithm and DW A algorithm is used to realize autonomous navigation in three-dimensional space.In order to verify the feasibility of the algorithm used in the system,an indoor autonomous cruise vehicle platform with Le TMC-520 camera,Jetson Nano processor and Ackerman rear-wheel-drive steering chassis as the core components was finally built,and a large number of experiments were carried out in the indoor environment.Test the real-time positioning,3D environment reconstruction,path planning and other functions of the indoor autonomous cruise vehicle.The final result shows that the VSLAM system designed in this paper can meet the requirements of autonomous cruise of the car.