基于 AI 芯片和机器视觉的协同控制实现

Currently, there are few studies on how to realize autonomous navigation of unmanned cluster system in a networkless and GPS - free environment. For this special application scenario, this paper innovatively proposes a ground - air cooperative control system using a centralized formation approach based on machine vision. The system consists of air reconnaissance aircrafts, air - following aircrafts, ground reconnaissance vehicles and ground - following vehicles. The air reconnaissance aircrafts use VINS - MONO to navigate and provide a position reference. At the same time, the CVM - Net is used to match captured images of air reconnaissance aircrafts and ground reconnaissance vehicles to obtain the relative position of them, which is used to guide the ground - following vehicles and air reconnaissance vehicles, and YOLO - LITE detection algorithm is used to realize target real - time detection. The system uses Huawei Kirin 970 as the core hardware, which can realize real - time and no - return processing of information. Through this system, the ground and air unmanned cluster can perform autonomous navigation and coordinated control in the networkless and GPS - free environment, which has broad application prospects in the fields of emergency search and rescue and military investigation.