期刊
IEEE SENSORS JOURNAL
卷 22, 期 9, 页码 8418-8427出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSEN.2022.3161794
关键词
Adjustable range; force sensor; neural network; soft sensor; variable stiffness
资金
- U.K. Engineering and Physical Sciences Research Council (EPSRC) [EP/S014039/1]
- Department of Mechanical Engineering, University College London (UCL)
This paper proposes a novel optical-based soft-tipped force sensor capable of adjusting its range and sensitivity through pneumatic modulation. The sensor can reliably measure normal forces and also measure the angle and magnitude of non-normal forces.
Force sensors are essential for measuring and controlling robot-object interactions. However, current force sensors have limited usability in applications such as grasping and palpation, where the range of angled forces changes between tasks. To address this limitation this paper proposes a novel optical-based soft-tipped force sensor capable of adjusting its range and sensitivity through pneumatic modulation. This research describes the sensor's design and examines the relationship between the internal pressure of the sensor and its sensing range, sensitivity, single-axis force-sensing accuracy, and capability of measuring the angle and magnitude of non-normal forces. Results indicate that by increasing the pressure in the sensor, the sensing range can be increased and the sensitivity decreased. These results demonstrate that the sensor can measure normal forces reliably at each pressure using 4th order fits with root-mean-square error (RMSE)is an element of [0.032 N 0.110 N]. Finally, it is also demonstrated that by using a neural network, the sensor can measure the angle and magnitude of non-normal forces with RMSEs on trained variables of 0.0120 Rad for Y-angle (theta(Y)) measurements, 0.0109 Rad for X-angle (theta(X)) measurements, and 0.102 N for force measurements.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据