Design Of Camera Location Methods Based On Apriltag

Vision-based positioning and mapping solutions are widely used in autonomous driving and various intelligent robots.In actual scenes,the existing visual positioning algorithm will be affected by various objective factors.For example,when there are large changes in lighting,camera movement is too fast,or jitter is too severe,it will seriously affect the initialization of the algorithm and the accuracy and positioning of the algorithm.stability.In order to ensure the accuracy and stability of the algorithm in a complex and changeable environment,this paper uses April Tag to assist the camera positioning initialization,track the feature points through the optical flow method,and combine the camera with the IMU sensor,thereby improving the success rate of algorithm initialization,So that the algorithm meets the real-time requirements and has good robustness in the case of poor image quality.The main research results obtained are as follows:Firstly,for the problem that visual SLAM can only obtain the pose of the camera relative to the initial frame,and there is a problem of scale uncertainty,a scheme that is initialized by April Tag assisted algorithm is proposed to establish the sparse locality while acquiring the absolute pose information of the camera.The map provides enough 3D feature points for subsequent positioning.Secondly,the correspondence between the feature points of each frame is obtained by optical flow tracking,and the Bundle Adjustment based on the direct method is used to solve and optimize the camera pose.By analyzing the principle of nonlinear optimization algorithm,the principle of corner tracking based on the optical flow method and the principle of pose optimization based on the direct method,a Forward-Addtive optical flow+Gauss-Newton nonlinear optimization algorithm was designed for motion tracking With building plans.Then,in order to make up for the deficiencies of pure vision-based positioning,tracking,and mapping that are easily affected by changes in camera imaging quality,movement speed,and lighting conditions,the camera + IMU sensor fusion method is adopted,through the sliding window-based visual inertial mileage calculation method,Control the BA to optimize the scale to obtain a more stable trajectory and improve the accuracy of the camera positioning algorithm.Finally,the requirements analysis of the solution of assisting initialization by April Tag and fusion positioning of camera and IMU is carried out,and then using a monocular fisheye camera to test the initialization algorithm of camera positioning based on April Tag,and applying the visual inertia odometer to monocular/stereo camera +IMU positioning.The experimental results show that the use of April Tag-assisted visual positioning can improve the success rate of camera initialization;the positioning based on visual inertial odometer has better stability in the environment with less texture.