面向地外天体表面智能感知与操控研究的多源数据集

Multisource Dataset for Intelligent Precise Perception and Manipulation of Extraterrestrial Object Surface

未来的地外天体探测已展现出人工智能的广阔应用前景,探测器可主动感知环境、理解环境,发现高价值目标,进行可靠推理和有效决策,并通过高精度导航和操控实现任务目标.地外天体表面场景复杂,地形地貌和地质等物理化学特性均无法确知,面向地外天体表面探测的智能感知与操控研究缺少有效的学习训练和测试验证数据集.本文介绍了北京控制工程研究所智能无人系统实验室在地外复杂场景模拟场条件下建立的天衢1.0多源数据集,数据集包含多源环境感知数据、探测器自身状态数据以及场景精准外测数据,能够为地外探测智能感知与操控研究提供研究与验证数据,推动深空探测人工智能理论方法与技术应用研究.

The future extraterrestrial exploration has shown a broad application prospect of artificial intelligence.The detector can actively perceive and understand the environment,find high-value targets,make reliable reasoning and effective decisions,and realize the mission objectives through high-precision navigation and manipulation.The surface scene of extraterrestrial objects is complex and unknown,and the physical and chemical characteristics such as topography and geology are hard to be measured directly or be determined.The research on intelligent perception and manipulation for extraterrestrial object surface exploration lacks effective training and testing data set.This paper introduces the a multi-source dataset TianQu1.0 for intelligent precise perception and manipulation of extraterrestrial object surface,which is established by the Extraterrestrial Intelligent Unmanned System Laboratory of Beijing Institute of control engineeringunder the conditions of simulated extraterrestrial test yard.The data set includes multi-source environment perception data,the surface mobile detector’s own state data and accurate external measurement data,so as to provide basic data sets and experimental verification conditions for,and promotes the research on the theory,method and technology application of artificial intelligence in deep space exploration.