Head-Orientation-Prediction Based on Deep Learning on sEMG for Low-Latency Virtual Reality Application

Tommy Sugiarto, Chun Lung Hsu, Chi Tien Sun, Shu Hao Ye, Kuan Ting Lu, Wei Chun Hsu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Reducing end-to-end latency on virtual reality system is important since it can remove several negative effects like motion-sickness and head orientation prediction is one of the solution to do that. On this study, signal from surface Electromyography (sEMG) was utilized to predict future head orientation with model trained from various deep learning algorithms. Total 20 subjects were participated with 6 muscles on neck were recorded for training purpose. The result showed that for both intra-subject and inter-subject method pre-processed sEMG signal + IMU input outperformed model with input from sEMG features + IMU. The result of inter-subject testing method on this study extended opportunity for real-world application in which the user data has never been include in training database.

Original languageEnglish
Title of host publicationProceedings - 4th IEEE International Conference on Robotic Computing, IRC 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages183-186
Number of pages4
ISBN (Electronic)9781728152370
DOIs
StatePublished - Nov 2020
Event4th IEEE International Conference on Robotic Computing, IRC 2020 - Virtual, Taichung, Taiwan
Duration: 9 Nov 202011 Nov 2020

Publication series

NameProceedings - 4th IEEE International Conference on Robotic Computing, IRC 2020

Conference

Conference4th IEEE International Conference on Robotic Computing, IRC 2020
Country/TerritoryTaiwan
CityVirtual, Taichung
Period9/11/2011/11/20

Keywords

  • convolutional neural network
  • deep learning
  • electromechanical delay
  • electromyography
  • low-latency
  • motion sickness
  • virtual reality

Fingerprint

Dive into the research topics of 'Head-Orientation-Prediction Based on Deep Learning on sEMG for Low-Latency Virtual Reality Application'. Together they form a unique fingerprint.

Cite this