Search

In-hand manipulation and grasp adjustment with dexterous robotic hands is a complex problem that not only requires highly coordinated finger movements but also deals with interaction variability. The control problem becomes even more complex when introducing tactile information into the feedback loop. Traditional approaches do not consider tactile feedback and attempt to solve the problem either by relying on complex models that are not always readily available or by constraining the problem in order to make it more tractable. In this paper, we propose a hierarchical control approach where a higher level policy is learned through reinforcement learning, while low level controllers ensure grip stability throughout the manipulation action. The low level controllers are independent grip stabilization controllers based on tactile feedback. The independent controllers allow reinforcement learning approaches to explore the manipulation tasks state-action space in a more structured manner. We show that this structure allows learning the unconstrained task with RL methods that cannot learn it in a non-hierarchical setting. The low level controllers also provide an abstraction to the tactile sensors input, allowing transfer to real robot platforms. We show preliminary results of the transfer of policies trained in simulation to the real robot hand.

Stochastic Resonance (SR) is a significant concern for design of robotic tactile sensors because of its considerable enhancement of sensitivity of human tactile sensation. In the last few decades, researchers have found that noise previously considered detrimental in fact provides the benefit of SR phenomena. However, there are no experimental results for dependence on vibrotactile frequency, which issued as basic data for robotic tactile sensors. In this paper, we investigated the influence of the SR phenomena on human tactile sensation in the context of frequency-dependence of absolute thresholds (AT). We developed a system composed of an experimental apparatus and a computer program based on psychophysics in order to measure the AT of human vibrotactile sensation. We performed three series of psychophysical experiments using normal vibration, 8-mm-sized stimulus point and the continuous sinusoidal signals of 32, 50 and 75 Hz to obtain the AT variation in frequency. The results show that since normal ATs decrease with appropriate noise, SR is observed in normal stimuli tests in the frequency range from 32 to 70 Hz. Since the inclination of the curve of AT in a double logarithmic graph is around -1 in both with-noise and without-noise conditions, SR is caused by the FA-I mechanoreceptive unit in the range of frequency. [ABSTRACT FROM AUTHOR]