多源信息融合的地外天体表面巡视定位方法研究

项目名称： 多源信息融合的地外天体表面巡视定位方法研究

项目编号： No.61773047

项目类型： 面上项目

立项/批准年度： 2018

项目学科： 自动化技术、计算机技术

项目作者： 刘传凯

作者单位： 北京航天飞行控制中心

项目金额： 16万元

中文摘要： 针对我国未来月球和火星巡视探测任务的工程需求，研究对现有的半自动化月面遥操作导航定位系统进行改造升级的关键支撑技术，借鉴和引入近年来在图像匹配、特征提取、目标检测、数据处理和平差求解方面的最新方法，从跨区域、多传感信息融合的角度，研究和突破多类扭曲与形变图像配准和多种测量定位数据融合的关键技术，减少图像自动配准的错误率、提高海量图像特征平差定位精度、增强定位求解的纠错能力，为我国未来月球和火星巡视探测任务提供理论指导和关键技术支撑。

中文关键词： 多源视觉导航理论与方法；多源信息融合方法；关联分析与综合评价

英文摘要： According to the engineering requirements for future lunar and Mars exploration mission of China, we will study the key technologies of the rover navigation and localization based on the existing semi-automated teleoperation system developed in Chang’E-3 mission. By learning from the latest advances in graph-based image matching, learning based object detection and classification, constrained weighted total least squares adjustment solving, we will investigate our methods from the aspects of the integration of three different levels of cross regional, cross space and inter planet,and expect to make breakthroughs in efficient visual odometry, accurate correspondence of deformed and distorted images, multi-constrained adjustment operations with error data and data fusion with a variety of localization results, so that we can reduce the error rate of image processing, enhance error correction capability of adjustment solving and expand the verification means of localization. Thus, the difficulties in the autonomy, automation and reliability of navigation and localization system will be overcomed and the achievments can provide theoretical guidance and key technical support for rover navigation and localization in China's future moon and Mars mission.

英文关键词： Vision-based Localization;Multi-source information fusion;Correlation analysis and comprehensive evaluation