Real-time classification of teleoperation data with a neural network
Fiorini, P. ; Losito, S. ; Giancaspro, A. ; et al.
Jan - 1992
DOI: 10.1109/CDC.1992.371334
ISBN : 0-7803-0872-7

Event Title :
Published in:
Publisher:
type: Conference Proceedings

Abstract
The development of a monitoring program that can be used in the future to evaluate operator performance and provide symbolic feedback about task progress is described. A classifier has been designed to recognize teleoperation task phases, independently of variations due to differences in working conditions and in phase features. Neural networks have been used to recognize task phases by using force data. Two network architectures have been tested in simulation on real teleoperation data, and the one with the best performance has been implemented on a teleoperation system. During tests on actual telemanipulation tasks, the classifier had a lower recognition percentage than the simulated tests, but it showed an unexpected generalization capability. It was able to correctly segment tasks whose phase sequence was significantly different from those in the training data

keywords : Actuators,Control systems,Employee welfare,Error correction,Feedback,Laboratories,Neural networks,Neurofeedback,Propulsion,System testing,Testing,Training data,User interfaces,computerised monitoring,feedback,generalisation (artificial intelligence),generalization,monitoring program,neural nets,neural network,phase sequence,real-time classification,symbolic feedback,task segmentation,telecontrol,teleoperation data,teleoperation task phase recognition