仿鹰隼视觉的无人机软式自主空中加油技术研究

This project mainly focuses on solving the problem of accurate recognition and measurement for probe-and-drogue autonomous aerial refueling (AAR) of unmanned aerial vehicles (UAVs) under complicated dynamic environments, which is crucial for AAR system. Based on the imitation of intelligent perception and target recognition mechanisms of falcon vision system, biological retina contrast sensitivity based region detection contour extraction models are established. The retina-nucleus mapping mechanism is explored for dynamic drogue identification and tracking. Then, a biological multi-resolution regional cooperative sensing device is developed and a multi-area coordination based pose measurement method is presented. Furthermore, the integrated hardware in loop simulation system and verification platform based on dual UAVs are exploited to test and verify all above techniques. This project can make a breakthrough of target recognition, tracking and measurement under high dynamic and strong interference environments for probe-and-drogue AAR, which can provide technical support for AAR of UAVs.

本项目面向自主空中加油背景，针对复杂动态环境下无人机自主空中加油准确识别和精确测量问题，模拟鹰隼视觉智能感知与目标识别机制，建立仿鹰隼视网膜对比度感知的区域检测和轮廓提取模型，研究仿鹰隼视网膜-核团映射的加油锥套动态识别和跟踪方法，开发仿鹰隼视觉多分辨率区域协同感知装置，提出基于多区域协同的加油锥套位姿测量方法，并在无人机软式自主空中加油验证平台上进行集成验证，突破高动态强干扰下软式自主空中加油目标识别、跟踪和测量技术，为我国无人机自主空中加油提供技术支撑。

本项目面向自主空中加油背景，针对复杂动态环境下无人机自主空中加油准确识别和精确测量问题，模拟鹰隼视觉智能感知与目标识别机制，建立了仿鹰隼视网膜对比度感知的区域检测和轮廓提取模型，研究了仿鹰隼视网膜-核团映射的加油锥套动态识别和跟踪方法，开发了仿鹰隼视觉多分辨率区域协同感知装置，提出了基于多区域协同的加油锥套位姿测量方法，并在无人机软式自主空中加油验证平台上进行了集成验证，突破了高动态强干扰下软式自主空中加油目标识别、跟踪和测量技术，相关研究成果可为我国无人机自主空中加油提供技术支撑。