Integrated Virtual Simulation and Test System for Vision-Based Applications of UAVs

Vision-based tasks of Unmanned Aerial Vehicles (UAVs) attracted significant research attention. Nowadays, testing a complex vision-based system using real experiments is expensive and time-consuming. This paper focuses on designing and implementing an integrated virtual simulation and test system for vision-based algorithm verification in visual perception, communication transmission, and closed-loop control of UAVs. First, a unified modular modeling process is proposed to construct the test system, which consists of a virtual environment subsystem, a vision algorithm subsystem, and a corresponding control subsystem. Although the vision algorithm and control subsystems are the same real experiments and virtual simulation, the virtual environment subsystem focuses on ensuring simulation credibility as the only difference between real and virtual experimentation. Then, some essential components are verified, including the camera model, image process, and communication of the visual sensor network. Finally, two case studies are performed to validate the efficiency and easy deployment, illustrating it as a promising platform for verifying and validating advanced guidance, navigation, and control algorithms.