@techreport{oai:ipsj.ixsq.nii.ac.jp:00229368,
 author = {Adhe, Rahmatullah Sugiharto Suwito P and Ayumi, Ohnishi and Tsutomu, Terada and Masahiko, Tsukaoto and Adhe, Rahmatullah Sugiharto Suwito P and Ayumi, Ohnishi and Tsutomu, Terada and Masahiko, Tsukaoto},
 issue = {2},
 month = {Nov},
 note = {Estimating a muscle contraction force has become an important method as it provides muscle conditions. Most studies focused on Electromyography (EMG) which requires a self-calibration procedure and its poor noise resistance. Considering the muscle-correlated modality, several studies introduced an excellent correlation of stretch sensors with EMG, thereby eliminating a complicated installation procedure. Therefore, this study proposed a method to estimate the forearm muscle contraction force using the stretch-sensor-based glove. The EMG signals of the forearm muscles and the stretch sensors' resistance values attached to each finger were recorded and trained by support vector regressor (SVR), random forest regressor (RFR), and multi-layer perceptron regressor (MLPR). The root-mean-square value of the EMG linear envelope was extracted as a muscle contraction force parameter and estimated from the stretch sensor values. The estimation performance was confirmed in the condition of grasping six cylinders varying in diameter sizes with natural and maximum grasping force. The result confirmed that the RFR model has the best performance on estimating the contraction force of forearm muscle., Estimating a muscle contraction force has become an important method as it provides muscle conditions. Most studies focused on Electromyography (EMG) which requires a self-calibration procedure and its poor noise resistance. Considering the muscle-correlated modality, several studies introduced an excellent correlation of stretch sensors with EMG, thereby eliminating a complicated installation procedure. Therefore, this study proposed a method to estimate the forearm muscle contraction force using the stretch-sensor-based glove. The EMG signals of the forearm muscles and the stretch sensors' resistance values attached to each finger were recorded and trained by support vector regressor (SVR), random forest regressor (RFR), and multi-layer perceptron regressor (MLPR). The root-mean-square value of the EMG linear envelope was extracted as a muscle contraction force parameter and estimated from the stretch sensor values. The estimation performance was confirmed in the condition of grasping six cylinders varying in diameter sizes with natural and maximum grasping force. The result confirmed that the RFR model has the best performance on estimating the contraction force of forearm muscle.},
 title = {Estimation of Contraction Force of Forearm Muscle Using Stretch-Sensor Glove},
 year = {2023}
}