基于空地协同的山地冰川连续探测系统  

Mountain Glaciers Continuous Detection Systems Based on Air-ground Robots’Co-operation

作　　者：钟继康 李鹏[1,2,3] 赵品辉[1,2,3,4] 杨丽英 何玉庆[1,2,3] ZHONG Jikang;LI Peng;ZHAO Pinhui;YANG Liying;HE Yuqing(State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016,China;Key Laboratory of Networked Control Systems,Chinese Academy of Sciences,Shenyang 110016,China;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang 110169,China;College of Computer Science and Technology,University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院沈阳自动化研究所机器人国家重点实验室,沈阳110016 [2]中国科学院网络控制系统重点实验室,沈阳110016 [3]中国科学院机器人与智能制造研究所,沈阳110016 [4]中国科学院大学计算机科学与技术学院,北京100049

出　　处：《无人系统技术》2023年第6期21-32,共12页Unmanned Systems Technology

基　　金：国家自然科学基金联合基金重点项目(U22B2041);国家自然科学基金(61991413,91948303);中国科学院青年创新促社会(Y2022065)。

摘　　要：针对无人机、无人车在复杂环境下运行不稳定、协作性差的问题,提出一种可以在冰雪表面、地形起伏等山地冰川环境下运行的空地机器人冰川探测系统。首先,该系统中无人机可适应高原山地低压、大风环境,无人车可在低温环境、冰雪表面长时间行驶,相应的轻量级地形建模和预测框架可将冰川表面环境点云模型转换成数据量更小的模型,且不造成大幅精度损失。其次,所提系统不仅可通过对驾驶数据、车辆状态数据与地形数据之间的相关分析建立驾驶技能模型,并利用增量式学习方法使驾驶技能模型快速适应冰川环境和复杂地表结构,其基于无人机机动性、无人机能耗、水平距离约束和视距约束的空地协同策略还可以保证系统的通信稳定和平稳运行。验证实验表明,提出的驾驶技能学习方案可以保证无人车在驾驶建议下平稳通过冰川危险地带,提出的空地协同策略可以保持较低水平的跟踪误差与通讯延迟。Aiming at the problems of unstable operation and poor cooperation of UAV and unmanned vehicle in complex environments,an air-ground robot glacier detection system is proposed which can operate in mountain glacier environments such as snow and ice surface and terrain fluctuation.First of all,the unmanned aerial vehicle(UAV)in the system can adapt to the low pressure and high wind environment in the plateau mountain,the unmanned ground vehicle(UGV)can travel for a long time in the low temperature environment with the snow and ice surface.The corresponding lightweight terrain modeling and prediction framework can convert the environmental point cloud model of the glacier surface into a model with smaller data volume without causing significant accuracy loss.Secondly,the proposed system can not only build a driving skill model through the correlation analysis of driving data,vehicle state data and terrain data,but also make the driving skill model adapt to the glacial environment and complex surface structure quickly by using incremental learning methods.The air-ground coordination strategy based on UAV mobility,UAV energy consumption,horizontal distance constraint and visual distance constraint can also ensure the stable and smooth operation of the system communication.Verification experiments show that the proposed driving skill learning scheme can ensure that the UGV can smoothly pass through the glacier danger zone under the driving suggestion,and the proposed open-ground coordination strategy can maintain a low level of tracking error and communication delay.

关 键 词：人机协同 多机协同 驾驶技能模型 增量式学习 轨迹规划 移动机器人 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]