期刊
IEEE TRANSACTIONS ON HAPTICS
卷 9, 期 4, 页码 499-507出版社
IEEE COMPUTER SOC
DOI: 10.1109/TOH.2016.2573298
关键词
Haptics; prosthetics neuroscience; human performance; perception and psychophysics
资金
- DARPA [N66001-15-C-4014, N66001-10-C-4056]
- NSF [533649, IOS-1150209]
- Kimberly-Clark Corporation
- Fonds special de recherche (FSR - Belgium)
Efforts are underway to restore sensorimotor function in amputees and tetraplegic patients using anthropomorphic robotic hands. For this approach to be clinically viable, sensory signals from the hand must be relayed back to the patient. To convey tactile feedback necessary for object manipulation, behaviorally relevant information must be extracted in real time from the output of sensors on the prosthesis. In the present study, we recorded the sensor output from a state-of-the-art bionic finger during the presentation of different tactile stimuli, including punctate indentations and scanned textures. Furthermore, the parameters of stimulus delivery (location, speed, direction, indentation depth, and surface texture) were systematically varied. We developed simple decoders to extract behaviorally relevant variables from the sensor output and assessed the degree to which these algorithms could reliably extract these different types of sensory information across different conditions of stimulus delivery. We then compared the performance of the decoders to that of humans in analogous psychophysical experiments. We show that straightforward decoders can extract behaviorally relevant features accurately from the sensor output and most of them outperform humans.
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