CASIA OpenIR  > 毕业生  > 博士学位论文
基于电刺激的动物运动行为控制方法研究
其他题名Research on the Control Methods of Animal Robots
王永玲
学位类型工学博士
导师原魁
2008-03-21
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业控制理论与控制工程
关键词动物机器人 脑机接口 电刺激 目标跟踪 位姿检测 Animal Robot Brain Machine Interface Electrical Stimulation Feature Tracking Poses Monitoring
摘要随着信息科学、电子技术和控制理论的发展,出现了各种不同类型、不同功能的智能机器人。但由于其稳定性、灵活性、智能性和适应性的不足,此类机器人大多处于实验室研究阶段,远未达到矿难救援、战争搜救等现实场合的需求。于是科学家开始独辟蹊径,提出了动物机器人的研究思路。 如今动物机器人的研究已经逐渐成为热点。动物机器人是指利用动物作为运动本体,背负控制电路、传感器等装置,或者利用动物本身的感觉器官作为传感器,经过一定的训练,通过遥控方式操纵动物的运动行为来得到所需的信息或完成某种任务。此外还可以把动物机器人平台作为生物传感器,或制作不同的动物疾病模型来研究疾病的发生机理及康复治疗方法。因此,本研究不仅对于基础研究、临床应用具有重要的意义,而且可以用作搜救、探索、排爆等具有重大价值的工程和军事领域。 然而如何有效地训练动物,使其能够听从外部指令,成为真正实用的动物机器人,国内还没有行之有效的训练方法和训练工具。为此,本文研究设计了用于训练动物机器人的软、硬件实验平台,在实验的基础上探索了多种训练方法,总结出了一套较为实用的动物机器人的训练方法,并初步训练出了基本可控的动物机器人原型。本文的主要研究内容及成果如下: (1) 提出了一种基于操作性条件反射与直接电刺激相结合的动物机器人的混和行为训练方法,并且通过对大鼠和家兔的行为训练证实了该方法的有效性。 (2) 设计了用于动物机器人训练的无线遥控电刺激器,并对其性能进行了实验验证。该刺激器体积小重量轻,能方便的设置不同的刺激参数模式,不仅可以用于动物机器人的训练实验,还可以用于神经科学和动物行为学的相关研究。 (3) 为了能对实验动物在刺激作用下的运动行为特性进行实时、定量的观测,本章设计开发了用于动物机器人训练实验的视觉监控追踪软件系统。该系统采用一种基于颜色模型的动物机器人视觉跟踪方法,利用普通的商用视频采集设备,构建了一套廉价高性能的动物行为跟踪监控系统,为动物机器人的训练和其他相关实验提供了一个便捷的观测平台。多次动物训练实验验证了该系统的实时性和准确性。 (4) 为满足室外动物行为监控的需要,设计了一种基于头部动作判别的动物机器人遥控训练系统。该系统由远程控制端和本地训练端两部分组成。远程控制端通过无线接收模块接收本地训练端的监控信息,并通过无线发送模块向训练端发送相应的控制指令,刺激器模块根据远程控制端的控制指令对受训动物进行刺激。头部行为监控模块对安装在动物颈部的弯曲传感器的状态进行监控,并采用RBF神经网络对动物的头部行为进行分析。该遥控训练系统具有体积小、重量轻、可靠性好等特点
其他摘要With the development of intelligent robotics and information technology, multifarious intelligent machines have come forth. Unfortunately, because their stability and flexibility are not satisfied, these machines are still under tentative schedules. For example, these machines can not be applied in complicated situations, such as searching for survivals from earthquake or bombing area. So some scientists and researchers present the concept of animal robot or hybrid robot in order to overcome the limitations of traditional robotics. Research on animal robotics is becoming a hot topic recently. That is, by carrying electronic devices or sensors, animals are remotely controlled to perform specific tasks and serve as robots. Thus, this will not only provide insights into how animal learns, which may eventually lead to better prosthetics, but also open new possibilities for biomedical, artificial intelligence and cognitive science researches. It also has a practical value in searching for victims from earthquake or bombing area, and the military value can never be overestimated. It is a big problem to make animals perform specific tasks and serve as robots, and there is not an effective method to train animal robots yet. So we designed a platform for animal robots training, which includes both the hardware and the software. After testing many training methods and experimenting different kinds of animals, a valid training method was found and animal robot prototypes were designed. The main contents and achievements of this dissertation are as follows: (1) After analyzing the established animal training methods, we presented a new animal robot training method, which integrated with both sound cue and electrical stimulation. The training results showed that our methods were valid for training animal robots,and rat-robot and rabbit-robot prototypes were trained successfully. (2) A new remote-controlled stimulator was developed, which had the characteristic of compactness and light weight. By setting the stimulation parameters conveniently, it can also be used to cognitive and other researches, and its performance and effectiveness were tested by experiments. (3) An animal tracking system was designed to monitor the animals’ performance after being stimulated. With a digital CCD camera, an image capture card and a personal computer, we developed a fast and robust tracking system, which used a color segmentation technique. The training results showed that the platform worked accurately even under interference conditions. (4) A remote-controlled animal robot training system based on head movements was designed to satisfy the requirements of animal behavior monitoring in outdoor.
馆藏号XWLW1174
其他标识符200418014628097
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/6053
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
王永玲. 基于电刺激的动物运动行为控制方法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2008.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20041801462809(16933KB) 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[王永玲]的文章
百度学术
百度学术中相似的文章
[王永玲]的文章
必应学术
必应学术中相似的文章
[王永玲]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。