面向ERT大面积触觉传感的自适应优化成像方法  

Adaptive Optimal Imaging Method for Large-Area Tactile Sensing in ERT

作　　者：郑文栋 刘华平[1] 孙富春[1] ZHENG Wendong;LIU Huaping;SUN Fuchun(Department of Computer Science and Technology,Tsinghua University,Beijing 100084,China)

机构地区：[1]清华大学计算机科学与技术系,北京100084

出　　处：《测控技术》2024年第4期21-28,共8页Measurement & Control Technology

基　　金：国家自然科学基金国际合作重点项目(62120106005);国家自然科学基金青年基金项目(62303259);中国博士后科学基金项目(2022M711825)。

摘　　要：在动态的非结构化环境中,有效地感知物理接触对于智能机器人安全交互至关重要。为了能够检测各种潜在的物理交互,需要在机器人表面部署大面积触觉传感器。目前,现有的大面积触觉传感器主要是通过传感阵列方式实现的,但是大规模部署传感元件在实际应用中存在巨大挑战。电阻层析成像(Electrical Resistance Tomography,ERT)技术作为一种连续传感方式,有望克服传统触觉传感阵列的一些限制。为此,利用ERT设计了一款新型的大面积触觉传感器。在此基础上,提出了一种基于自适应感兴趣区(Region of Interest,ROI)的图像重构算法,将图像重构限制在交互区域内,以提高传感器的空间分辨率。为了验证提出成像算法的有效性,通过仿真与物理实验对其进行了全面评估。实验验证了该算法可以有效提高触觉传感器在交互区域的空间分辨率,使其具有较高的测量精度。实验结果表明,该传感器的平均定位误差为0.823 cm,能够准确地识别8种不同交互模式,其精度高达98.6%。这一研究工作表明,该传感器为机器人具身触觉传感的实现提供了一个新的解决方案。In the dynamic unstructured environments,being able to sense physical contact is critical for safe interaction of intelligent robots.To be able to detect a variety of potential physical interactions,large-area tactile sensors need to be deployed on the robot surface.At present,existing large-area tactile sensors are mainly implemented through sensing arrays,but large-scale deployment of sensing elements poses huge challenges in practical applications.As a continuous sensing method,electrical resistance tomogaphy(ERT)is expected to overcome some of the limitations of traditional tactile sensing arrays.Therefore,a new large-area tactile sensor is designed using ERT.On this basis,an image reconstruction algorithm based on adaptive region of interest(ROI)is proposed to limit image reconstruction to the interaction area to enhance the spatial resolution of the sensor.In order to verify the effectiveness of the proposed imaging algorithm,it is comprehensively evaluated through simulation and physical experiments.The experiment verifies that this algorithm can effectively improve the spatial resolution of the tactile sensor in the interaction area,and the sensor has high measurement accuracy.Experimental results show that the sensor has an average positioning error of 0.823 cm and can accurately identify 8 different interaction modes with an accuracy as high as 98.6%.This study shows that this sensor provides a new solution for the realization of robot embodied tactile sensing.

关 键 词：大面积触觉传感 电阻层析成像 机器人触觉感知 人机交互 图像重建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]