距离缺失多特征融合的无人机感知与规避研究

项目名称： 距离缺失多特征融合的无人机感知与规避研究

项目编号： No.61473230

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 潘泉

作者单位： 西北工业大学

项目金额： 85万元

中文摘要： 随着军民领域对无人机的需求日益强烈，未来将呈现多种类型无人机、有人机空域共享的局面，空域将日趋密集。无人机感知与规避(SAA)技术成为当前国际无人机技术研究的前沿领域，美国防部于2011年成立无人机SAA专家组，美欧针对SAA领域尚存的大量技术空白给予了高度关注。基于视觉的SAA是SAA系统的核心技术之一，本项目拟通过对空间小目标的结构特征提取和质心检测，构建具有视觉结构特征与运动特征相融合的视觉反馈误差矩阵，从而得到视觉伺服框架下的距离无关交互矩阵；建立基于飞机运动方程和安全飞行包络的视觉伺服控制方程，得到距离量测缺失下的非合作目标SAA策略。综合评估距离缺失与基于距离量测的两类SAA系统的差异，构建视觉、激光雷达、ADS-B三通道综合的SAA系统；并进行典型场景下距离缺失非合作目标的小型视觉SAA系统基本功能、核心算法和性能验证。为SAA系统发展提供新的视觉SAA方法和系统实现方案。

中文关键词： 感知与规避；无人机导航；多特征融合

英文摘要： As for the intense need for UAV in both military and civil applications,in the future the national airspace will be consist of both manned and unmanned Vehicles. UAV sense and avoid(SAA)technology has become one of the most promising area for UAV research and US DoD organize a SAA technology panel which focus on the technology gap of SAA in 2011. Vision-based SAA is one of the core technology of SAA system.We plan to establish a vision feedback error matrix with motin feature and stucture feature by feature extaction and centroid detection.The we can obtain a Depth-independent interaction Martrix in the framewok of Visual Servoing.By combining both UAV kinematics and SAA safe envelop we can setting up a Visual Servoing control algorithm which can be treat as a Depth-independent un-cooperative SAA strategy.By comparing and evaluating the depth dependent and independent SAA method, we finally will build up a three channel SAA system with Vision,Lidar and ADS-B sensors. This SAA system will be tested on typical collision scene with its basic function,core algorithm and performance evaluation.All the work can provide novel method and system framework for SAA development.

英文关键词： Sense and Avoid;UAV Navigation;Multi-Feature Fusion