Research On Navigation Method Of Indoor Mobile Robot Based On Vision

In order to ensure long-term stable operation of mobile robots,navigation systems that can know their location information in real time become very important.Nowadays,with the continuous advancement of computer technology,visual SLAM technology has made great breakthroughs and is favored by many scholars.Monocular vision SLAM has become the focus of research in the field of visual SLAM due to its small size and easy installation.However,the monocular camera itself has the problem of scale uncertainty.This paper designs a vision-based multi-sensor fusion positioning system,which combines the information of IMU and monocular camera data in tight coupling mode to obtain high-precision,high-stability navigation.system.The main work is as follows:Firstly,the pinhole imaging model of the camera is given.The pose transformation of the camera between two frames is described by the polar constraint.The attitude angle and velocity solving equation of the carrier in space are derived by quaternion method.The commonly used coordinate system is navigated and the coordinate transformation matrix is given.Secondly,the pre-integrated IMU information and the visual information tracked by the KLT optical flow method are tightly coupled,and the merged state vector is given,and the optimized objective function is optimized by loopback detection and pose analysis.The global consistency of the positioning trajectory is guaranteed,and the positioning algorithm of visual inertia fusion is realized.Thirdly,the kinematic control model of the omnidirectional mobile robot is established and the speed closed loop control is completed.The internal parameters and distortion coefficients of the Kinect v2 camera were calibrated by Zhang's calibration method.The JY-61 sent the data to the PC ROS under the serial port,and the Topic of the IMU data was successfully released.At the same time,the design of the Android mobile APP was completed.To control the movement of the mobile robot.Finally,the construction of the mobile robot software and hardware platform is completed.The experimental building is selected as the environment for simulating the mobile robot working in the factory.The visual inertial navigation fusion positioning experiment in the indoor environment is carried out.The experimental results show that the visual inertial fusion algorithm has better performance.Robustness and positioning accuracy.