【作者】 郭兵；

【导师】 秦岚；

【作者基本信息】 重庆大学 ， 仪器科学与技术， 2010， 博士

【摘要】 触觉是机器人与外界环境交互的感知,机器人通过触觉能感知外部环境信息。触觉传感系统已成为高级智能机器人标准特征之一,是研究可覆盖机器人全身触觉传感服装的关键技术。机器人借助具有类似人皮肤功能的传感服装可以判断是否接触到外界物体,并测量与接触物体之间的压力以及压力的空间分布特性,同时判断接触物的外形,以便使其完成更复杂的任务。因此研究机器人大面积传感服装与大规模传感信号处理是机器人触觉研究的热点和难点。本文针对触觉传感阵列未解决的兼具柔韧性和测量压力分布信息等难题,完成了机器人传感服装制备工艺、大规模阵列传感器信号处理等方面的创新性研究。研究工作取得了以下创新性成果。①研制出了一种以导电橡胶为基料兼具柔韧性的触觉传感阵列新导电单元。借助机理实验,获得了在不同参杂浓度条件下导电橡胶单元的压阻关系特性及压力、温度与电阻的三维关系特性,为传感服装数据处理研究提供了技术理论支撑。②研制出了一种具有接触面积大、柔韧性好、分辨率可调和可拼接组合等特点的新型机器人传感服装。针对该系统构建了一种新型的信息获取与处理系统,其硬件电路采用行列扫描方式获取传感信号,采用改进型电压镜扫描电路可避免伪信号产生;对于机器人触觉服装大规模传感信号的处理,提出了触觉数据处理构架以此提高系统的稳定性和可靠性。③提出了在不确定传感数据模型基础上的一种基于统计信息融合的机器人传感服装数据处理算法,实验研究表明该信息处理算法实现了对压力位置和分布的精确识别,有效保证了传感数据处理结果的精确性和可靠性,增强数据可信度与系统的生存能力,扩展时间空间覆盖率。④提出了一种基于BP神经网络大规模传感阵列的信息融合方法,通过该方法可有效消除如环境温度、电压扰动等非目标参量及大规模传感信号间串扰对传感服装压力传感阵列输出特性的影响,以此改善机器人触觉传感服装系统的稳定性和可靠性,同时可以提高机器人系统的智能判断力与信息利用率。⑤提出了一种基于重构机器人接触物形体的归一化转动惯量特征识别方法,建立了机器人服装触觉传感数据处理的模糊认知图模型,给出相应的推理运算步骤。可借助模糊认知图运算确定机器人对接触物刺激的反应行为,找出传感服装感知过程中各要素间存在的明显因果关系。最后验证了模糊认知图可作为一种计算平台,适于机器人服装感知全过程数据处理。综上,本文针对机器人传感服装与数据处理,研制了机器人触觉传感服装,提出了数据处理架构及相关信号处理方法。理论分析和实验结果表明,该系统设计是合理可行的,能够实现对触觉传感信息的有效和准确获取,提高了机器人的感知能力。更多还原

【Abstract】 Tactile is a perception of interaction between robot and its external environment, the robot can sense the external environment information through the tactile function. Tactile sensing system has become one of standard features of high-level intelligent robot. It is the key technology of tactile sensing suit for covered all the robot body. By means of sensing suit owned similar human skin, it can judge that whether the robot touches to the external object, and measure the pressure between robot and touched object as well as space distribution characteristic of pressure, and judge the outline of touched object so as to let robot accomplish more complex task. Therefore the study on suit of robot sensing with large area and processing with large-scale sensing signal is the puzzle and hot point in research on robot tactile.In this paper, aimed at the puzzle that are not solved in the tactile sensing array with both outstanding flexility of material and measure information of pressure distribution, it has been completed for innovative research to the preparation technology of robot sensing suit, and signal processing of sensor with large-scale array and so on. The research work has obtained the following innovative results.①Developed a new conductive material of tactile sensing array based on both basic material of conductive rubber and owned outstanding flexility. By means of mechanism experiment, it obtained the characteristic of voltage and resistance as well as 3D relation characteristic among temperature and pressure and resistance under the condition of different mixed consistence for new conductive material. It provided the foundation of technical theoretical support for data processing of sensing clothing.②Developed a sort of novel robot sensing suit owned the property with being large in contact area, better in flexibility, adjustable in resolution and being able in sewing and pruning, And also constructed a sort of novel acquisition and procession system for information. In which, the hardware adopts scanning manner of row and column to capture sensing signal, and its voltage mirror scan circuit can avoid to produce false signal.③Presented a sort of novel fusion algorithm of multi-sensor information processing, in which the algorithm is based on stat characteristic of condition probability and has the data compensation function. It can enhance the intelligent judge ability and information utilization efficiency, strength the data reliability and system viability, and extend time and space coverage.④Presented a sort of novel fusion method of large-scale sensing signal based on BP neural network. It can effectively eliminate the impact, such as environmental temperature and voltage disturbance and other non-target parameters and cross interfering among large-scale sensor signals, of output characteristic for sensing clothing pressure sensor, and enhance stability and reliability of pressure sensing.⑤Presented a sort of recognition method of normalized rotation inertia feature based on outline of touch object of reconstruction robot, and established the fuzzy cognitive map model of tactile sensing data processing for robot clothing, and gave the steps of corresponding reasoning operation. By means of cognitive map operation, the robot can determine the reaction behavior for touch object activating, find the obvious causality among important factors in the perceptive process of sensing clothing. Finally it validated that the fuzzy cognitive map can be as a computing platform, and it is suitable for the whole process data processing of robot perception clothing.In summary, aimed at the robot sensing clothing and data processing, it developed robot tactile sensing suit, and presented the architecture of data processing and processing method of correlated signal in the paper. Theoretical analysis and experimental results show that the system is reasonable and feasible, it can realize to achieve tactile sensing information effectively and accurately, and enhance the robot’s perception.更多还原