System research on AGV carts for warehouse logistics based on visual navigation

In order to effectively improve factory productivity and reduce the labor of shop floor workers in handling production goods, a monocular vision-based navigation system is proposed to realize the automatic handling of AGVs. First, by scanning the QR code information pasted beside the AGV path with a monocular camera, the matrix QR code is used as an absolute position label to assist in correcting the cumulative error generated during the navigation process of the odometry and inertial measurement system. Second, a second-order multi-target template matching algorithm is proposed to improve the recognition and localization speed of the algorithm while also improving its anti-interference to the environment.