机器人抓取对象硬度触觉感知研究

物体硬度感知对于机器人进行物体灵巧操作具有重要意义。针对物体硬度感知中传感信号复杂、物体压缩量大而导致的系统鲁棒性差以及容易损伤物体的问题，提出了一种基于触觉传感的机器人抓取对象硬度感知方法。该方法使用两指夹持器轻微挤压物体，通过安装在两指指尖的柔性触觉传感器阵列采集压力序列信号。将压力序列信号进行多项式处理得到非线性特征序列，使用基于决策树的Adaboost算法处理非线性特征序列，实现抓取物体在线硬度等级分类。将基于决策树的Adaboost算法和其他各种算法进行比较，并进行实际物体硬度识别实验。实验结果表明所提方法能够准确实时识别不同抓取对象的硬度。

Research on Robotic Grasping Object Hardness Perception Based on Tactile Sensing

The perception of object hardness is of great significance for robots to perform fine manipulation task. Aiming at the problems of poor system robustness and easy damage to objects caused by complex sensing signals and large object compression in object hardness perception, a method of robotic grasping object hardness perception based on tactile sensing is proposed. Pressure sequences are obtained by the flexible tactile sensors when the two fingers gripper slightly squeezes the object. The pressure sequence signals are then polynomial processed to obtain nonlinear characteristics and then input into the Adaboost algorithm which is based on the decision tree to realize online grasped object hardness perception. The Adaboost algorithm is compared with other algorithms, and the hardness perception experiment for novel objects is carried out. Experimental results show that the proposed method can accurately identify the hardness of different grasped objects.