A first order transfer function to balance the workload in brake-motor hybrid actuators

A first order transfer function to balance the workload in brake-motor hybrid actuators
复制标题

用于平衡制动电机混合执行器工作负载的一阶传递函数

DOI:
10.1109/haptics.2014.6775508
复制
发表时间:
2014
期刊:
2014 IEEE Haptics Symposium (HAPTICS)
影响因子:
--
通讯作者:
H. Gurocak
H. Gurocak
中科院分区:
--
文献类型:
--
作者:
M. Antolini;Orhun Kose;H. Gurocak

文献摘要

被引文献

相似文献

本文提出了一种控制算法,用于在适合阻抗控制触觉设备的制动器和电机(混合驱动)之间分担扭矩负载。与其他几种方法不同,它不需要力或扭矩传感器,从而可以设计更紧凑、更便宜的触觉设备。所提出的算法解决了所谓的“粘墙”效应的问题。当虚拟墙碰撞期间,当用户想要离开虚拟墙时制动器仍然接合时,就会出现此问题。当这种情况发生时,用户可以感觉到这样的虚拟墙是“粘性的”。当使用多自由度制动器(例如我们研究实验室构建的球形磁流变制动器)时,这个问题会更加突出。事实上,由于制动器同时锁定所有自由度,如果没有力/扭矩传感器,就无法执行在墙上滑动等操作。所提出的算法背后的想法是在需要被动(制动)扭矩时使用制动器,并通过根据一阶传递函数使用电机进行补偿来缓慢衰减制动器的激活。传递函数允许制动激活呈指数衰减,与线性转换相比具有更快的响应。我们开发的算法比在虚拟墙碰撞过程中仅使用电机更加节能。此外,对于高达 5000 Nm/rad 的刚度值,对虚拟墙的冲击提供了更少的振动。
The paper proposes a control algorithm to share the torque load between a brake and a motor (hybrid actuation) suitable for an impedance controlled haptic device. Unlike several other methods, it does not include the need for a force or torque sensor, allowing to design a more compact and cheaper haptic device. The proposed algorithm solves the problem of the so called “sticky wall” effect. This problem occurs when, during a virtual wall collision, the brake is still engaged when the user wants to get off from the virtual wall. When this happens, the user can sense such virtual wall as “sticky”. This problem is emphasized when using multiple DOF brakes such as the spherical MR brake built in our research laboratory. In fact, since the brake locks all the DOF simultaneously, operations like sliding on a wall could not be performed without a force/torque sensor. The idea behind the proposed algorithm is to use the brake when passive (braking) torque is necessary, and slowly decaying brake activation by compensating it using motors according to a first order transfer function. The transfer function allows an exponential decay of the brake activation, with a faster response compared to linear transition. The algorithm we developed resulted to be more energy efficient than using just motors during virtual wall collision. Furthermore, the impact against a virtual wall provided less vibrations for stiffness values up to 5000 Nm/rad.