摘要
为了使机器人在光线不足、黑暗、沙尘等环境下感知目标物体的运动,提出了基于电子皮肤的机器人运动感知算法。算法首先使用电子皮肤作为触觉传感器采集触觉图像,接着利用柔性基底和图像形态学技术改善触觉图像;然后通过识别触觉图像形心和配准最优旋转来计算目标的平移和旋转;最后使用线性回归模型拟合平移距离和旋转角度以计算运动速度,完成运动感知。此外,使用高精度并联定位平台验证了算法精度,算法计算平移距离及速度的最大误差不超过2.7%和6.0%,计算旋转角度及速度的最大误差不超过9.7%和4.0%。该结果表明机器人触觉能在极端的工作环境下代替视觉感知物体运动,进而改善与目标的交互,扩展机器人应用场合。
Aiming at sensing the motion of target in dark or dusty environment for robots,a motion sensing algorithm of robots based on electronic skin(e-skin)was proposed.Firstly,an e-skin was employed to sample tactile images,and a soft substrate and image morphology were used to improve tactile images.Then,the translation and rotation of target was calculated by centroid recognition and best rotation registration of tactile images.Finally,a linear regression model was used to fit the translation distance and rotation angle to calculate the motion speed.Moreover,a high precision parallel position platform was used to experimentally investigate the accuracy of the algorithm,which showed that the maximum error won t exceed 2.7%and 6.0%when the algorithm calculated the distance and speed of translation,besides,the maximum error won t exceed 9.7%and 4.0%when calculating the angle and speed of rotation.The accuracy of the algorithm shows tactile sensing can perceive the motion instead of vision in complicated environment for robots,furthermore,improve the interaction with targets,extend applications of robots.
作者
陈楚浩
谢瑜
代稷珅
CHEN Chu-hao;XIE Yu;DAI Ji-shen(School of Aerospace Engineering,Xiamen University,Xiamen 361102,China)
出处
《仪表技术与传感器》
CSCD
北大核心
2020年第6期100-105,109,共7页
Instrument Technique and Sensor
基金
国家自然科学基金项目(61403320)
广东省自然科学基金项目(2017A030313352)
空间智能控制技术国防科技重点实验室2018年度基金项目(KGJZDSYS-2018-07)。
关键词
机器人
电子皮肤
触觉传感
旋转配准
图像处理
运动感知
robot
e-skin
tactile sensing
rotation registration
image processing
motion sensing