| dc.description.abstract | Visual following is an important part in the field of robot, which can be used in storage,
handling, security, military and other fields. Due to the problems that the traditional algorithm cannot track the target effectively when the background is relatively complex, it has high resolution requirements for the tracking target and the external environment, so it can only carry out auxiliary tracking, and the large amount of calculation fails to meet the real-time requirements. This paper adopts the YOLOV4 Tiny algorithm to design a ROS-based mobile robot visual following system. The Jetson Nano development board is used to provide the computing performance of 472 GFLOPS, since it uses a quad-core 64-bit ARM CPU and 128 cores to integrate NVIDIA GPU. Thus, the speed of calculation is improved and the real-time requirement of the algorithm is satisfied. | ru |
