| 摘要: |
| 视觉导航对于飞行器的自主性和智能化发展具有重要意义。随着各类技术，特别是人工智能技术的飞速发展，飞行器视觉导航进入了一个新的时代。通过分析飞行器视觉导航的演进历程及其在不同发展阶段的技术特点和应用，揭示了飞行器视觉导航的发展趋势和重点。根据输入信息和输出导航参数的特点，飞行器视觉导航技术的发展可分为地形匹配导航、景象匹配导航、空间计算导航和空间智能认知导航4个阶段。重点归纳整理了各阶段具有代表性的导航方式、信息处理特点及应用情况，简要介绍了团队在飞行器空间计算导航方面的研究。讨论了当前视觉导航与人工智能的融合情况，并对后续发展进行了展望。随着数据不断积累、算法持续迭代及端侧算力不断提升，基于空间大模型的“空间智能认知导航”在未来可能“让飞行器像人一样导航”，作为一种全新的飞行器自主导航方法，为极端环境下飞行器的自主决策与复杂任务执行提供有力支撑。 |
| 关键词: 飞行器视觉导航 地形匹配导航 景象匹配导航 空间计算导航 空间智能认知导航 |
| DOI： |
|
| 基金项目: |
|
| New era of aircraft visual navigation: from terrain matching to spatial intelligence（Cover Article） |
| SHANG Kejun,HUANG Chenglin,XU Ce,MING Li,ZHAO Liang,ZHANG Weijian |
| (Beijing Institute of Automation Control Equipment, Beijing 100074, China;School of Automation, Beijing Institute of Technology, Beijing 100089, China) |
| Abstract: |
| Visual navigation is highly significant for the autonomous and intelligent development of aircraft. With the rapid development of various technologies, especially artificial intelligence technology, aircraft visual navigation has entered a new era. By analyzing the evolution of aircraft visual navigation, as well as its technological characteristics and applications at different stages of development, the development trends and priorities of aircraft visual navigation are revealed. Based on the characteristics of the input information and the output navigation parameters, the development of aircraft visual navigation technology can be divided into four stages: terrain matching navigation, scene matching navigation, spatial computing navigation and spatial intelligent cognitive navigation. Representative navigation methods, information processing characteristics and application situations in each stage are summarized, and the research conducted by the team on aircraft spatial computing navigation is briefly introduced. The current integration of visual navigation and artificial intelligence is discussed, and the future development is prospected. With the continuous accumulation of data, the continuous iteration of algorithms, and the continuous improvement of terminal-side computing power, “spatially intelligent cognitive navigation” based on large spatial models may, in the future, “enable aircraft to navigate like humans”, inaugurating an entirely novel paradigm of autonomous aircraft navigation and providing robust support for autonomous aircraft decision making and complex task execution in extreme environments. |
| Key words: Aircraft visual navigation Terrain matching navigation Scene matching navigation Spatial computing navigation Spatial intelligent cognitive navigation |
