微型仿生爬虫机器人类脑环境感知方法  

Brain-inspired environmental perception method for micro-biomimetic crawling robots

作　　者：王运梦 李涛[1,4] 徐剑君 唐凤珍 崔龙 刘钊铭[1,2,3] 刘连庆 Yunmeng Wang;Tao Li;Jianjun Xu;Fengzhen Tang;Long Cui;Zhaoming Liu;&Lianqing Liu(State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016,China;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang 110169,China;University of Chinese Academy of Sciences,Beijing 100049,China;School of Information Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]中国科学院沈阳自动化研究所,机器人学国家重点实验室,沈阳110016 [2]中国科学院机器人与智能制造创新研究院,沈阳110169 [3]中国科学院大学,北京100049 [4]沈阳工业大学信息科学与工程学院,沈阳110870

出　　处：《科学通报》2023年第23期3095-3106,共12页Chinese Science Bulletin

基　　金：国家自然科学基金(U2013208,92067205);中国科学院稳定支持基础研究领域青年团队计划(YSBR-041)资助。

摘　　要：微型仿生机器人是一种仿照生物外形和运动形态制作设计的机器人,凭借其体型小、机动性高、环境适应性强等优点,在复杂环境探索、敌情侦察等特殊场景中具有较高的应用前景,近年来备受研究人员的关注.但是微型机器人运动稳定性弱、单目相机环境感知精度低等问题的存在一直制约着其实际应用.本文仿照昆虫“独角仙”设计了一种新的微型仿生爬虫机器人,开发了基于生物运动调控机制的六足机器人控制系统,改进了基于啮齿类动物空间导航原理的同时定位与建图系统.使用自研的微型仿生爬虫机器人在人造沙盘和真实室内走廊两个场景中进行了实验验证.在人造沙盘场景,由于微型仿生爬虫机器人提供的环境图像质量模糊,ORB-SLAM3算法无法完成回环检测,不能正确识别曾经见过的场景.类脑同步定位与地图构建(simultaneous localization and mapping,SLAM)系统回环准确率高达100%,比原RatSLAM算法提高了4.36%.在真实室内走廊场景,ORB-SLAM3算法和RatSLAM算法建图效果都比较差,而类脑SLAM不仅有着较好的建图结果,而且在准确率100%的同时召回率也高达97.87%.与ORB-SLAM3和RatSLAM算法相比,类脑SLAM系统取得了较好的建图结果.因此,自研的微型仿生爬虫机器人具有灵活的运动能力、鲁棒的导航定位能力,促使微型仿生机器人离实际应用更近了一步.Micro-biomimetic robots are designed in such a way that animals are imitated morphologically and kinematically.Such robots have many merits.Firstly,their body size is usually very small such that they can pass narrow passages.Secondly,they commonly have high maneuverability such that they can move freely in very complicated environments.Last but not least,they also have strong environmental adaptability such that they handle different environments.Therefore,microbiomimetic robots have great potential in the applications of many special scenarios.For instance,they can be applied to explore in complex and cluttered environments,or collect information from enemies secretly.Consequently,the research on micro-biomimetic robots has recently drawn many attentions.However,the existing micro-biomimetic robots still face several problems.For instance,the stability of their motion is still low and their visual perception through monocular cameras is rather unreliable.These limitations highly restrict the practical application of the micro-biomimetic robots.To alleviate these problems,this paper presents a new micro-biomimetic crawling robot system.The micro-biomimetic crawling robot is designed to mimic the insect which is referred to as‘Trypoxylus dichotomus'.The robot is equipped with a control system that targets hexapod robots,providing the robot with the ability to move.The control system is developed based on the biological motion regulation mechanism.The proposed robot system also includes a vision system based on brain-inspired simultaneous localization and mapping(SLAM)to provide environmental navigation ability for the microbiomimetic crawling robot.The included brain-inspired SLAM is an improved version of Rat SLAM,which is developed mainly inspired by the spatial navigation mechanism in rat's brains.The proposed micro-biomimetic crawling robot system has been verified in both an artificial environment and a real-world indoor environment.The size of the artificial environment is relatively small.Moreover,the en

关 键 词：微型仿生机器人 生物运动调控机制 空间导航神经机制 类脑SLAM 环境感知 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]