【作者】 吕晓玲；

【导师】 张明路；

【作者基本信息】 河北工业大学 ， 机械制造及其自动化， 2010， 博士

【摘要】 随着世界各国对国家安全、社会治安等公共事业的高度重视,研究在危险环境中代替人类工作的移动机器人,有着重要的理论和实际意义。听觉系统,作为人类感官的重要组成部分,已经成为机器人领域的重要研究对象。由于声音可以绕过障碍物,听觉可以和机器人视觉、嗅觉等感官相配合,弥补其他传感器视场有限且不能穿过非透光障碍物的局限。目前,对于能够实现移动机器人对可疑声源的准确定位、搜索,是应用移动机器人执行危险作业亟待解决的关键问题。本论文在国家863计划项目“极限环境下面向危险品检测的多感官机器人系统”(项目编号:2006AA04Z221)的支持下,利用所研制的多感官拟人机器人系统,对危险声源定位、搜索进行了深入研究。本文的创新性工作包括:1、针对声源定位的实际环境,从信号滤波、利用声音的优先效应、针对环境噪声进行改进、限制互相关函数峰值的搜索范围、多帧加权、根据信噪比动态的调整权重函数的大小等六个方面提出了改进的广义互相关时延估计法,并给出了实际不同环境中的权重函数的最优取值,有效地改善了低信噪比、混响环境下的时延估计算法的性能,很大程度的提高了基于时延定位算法的精度。2、在研究传声器阵列的基础上,提出了机器人听觉系统模型,构造了拟人听觉系统,该系统有五个传声器组成,其中四个传声器组成的平面阵确定声源空间位置,另外一个传声器辅助完成声源位于机器人前后方的判断。实验证明,所设计的声源定位系统,运算量小,实时性好,具有良好的实际应用的价值。3、提出并设计了基于行为控制的声源搜索实现方案。融合机器人听觉和超声避障的自主声源搜索策略按优先级分成三个模块:声源确认、超声避障、声源定位搜索,通过优先级的判断确定当前执行模块。系统在室内环境下测试,实验结果证明机器人可以定位声源并且可以绕过障碍物接近并确认声源,结果证明该方法具有良好的实时性和有效性。4、借鉴人类综合利用听觉和视觉感官功能进行声源搜索的方法,提出了一种融合听觉、视觉信息的机器人声源定位搜索策略。设计了基于多层黑板模型的机器人声源定位搜索系统,通过声源定位搜寻黑板和声源确认黑板完成机器人声源搜索。通过实验验证了该策略具有最高的效率和成功率,具有较强的环境适应能力。更多还原

【Abstract】 As greater and greater importance is attached to the public utilities such as national security and public order, the research on the mobile robot that can replace human beings in the dangerous environments is of great theoretical and practical significance. Auditory system, as an important sense component of human, has become an important research subject. As the sound can bypass the barrier, auditory system can co-operate with the other sense, such as robot vision、olfaction and so on. It can make up for the other sense’s limited field that can not pass through non-transparent barrier. At present, in order to make mobile robot do the dangerous work for human beings, the key problem to be urgently solved is how to realize the mobile robot’s accurate localize and search the suspicious sound source.This paper is supported by the Hi-tech Research and Development Program of China -- the Multi-Sensory Robot System of Detecting Dangerous Goods in Extreme Environments (No. 2006AA04Z221). It combines the multi-sensory humanoid robot system, and makes a deep research into the sound source localization and searching.This paper covers the following original points:1. According the real environment of sound source localization, modified generalized cross correlation time delay estimation method is proposed from six parts: sound signal filtering; using sound priority effect; improving according environment noise; reducing the search region according to the distances of microphones; more frames weighting; adjusting the weight factor dynamically according to signal noise ratio. Optimalizing weight factor values under different environments are given. This algorithm has better performance in reverberate and low signal-to-noise ratio (SNR) condition, and time delay localization precision is improved largely.2. On this basis of research on microphone array, robot auditory system model is proposed, humanoid auditory system is presented, the system is composed of a five microphone array, sound source position in space is calculated by the planar array of four microphones, and another microphone is used as an auxiliary unit to judge the source in front or rear of the robot. Through experiments, the robot system is proved to have better practical value for its low computation cost and good real-time property.3. Sound source searching proposal is proposed and designed based on behavior control. The autonomous sound source search strategy composed of robot hearing and ultrasonic obstacle avoidance is divided into three modules: source confirm, ultrasonic obstacle avoidance, sound source search, current executing module is determined through the priority judgment. The system was tested in room environment, experiment results show that the robot is capable of localizing sound source, approaching and confirming it without collision. The method is proved to have good real-time property and validity.4. Considering the experience of human beings who make use of hearing and vision sense to search sound source, a strategy to locate and search the sound source by fusing the auditory and visual information is proposed. A control system of the robot sound source searching on the basis of multi-blackboard model is designed. Through sound source searching and confirming blackboard, the control system can realize the sound source searching. Through experiments, this strategy is proved to be efficient, successful and suitable to adjust to the environment.更多还原