Effect of Muscle Fatigue and Joint Angular Velocity on EMG signal and EMG to Torque Model Khalil Ullah * , Ali Raza ** , Jung-Hoon Kim *** , Kang Park **** * Myongji univ.,dept. Electronics and comn. Eng. (TEL: 010-8691-8402;E-mail:khalil_dirvi@yahoo.com ) **Myongji univ.,dept. Software. Eng. (E-mail:aliraza.ac.kr ) *** Prof. Yonsei univ.,dept. Civil and environmental Eng. (E-mail: junghoon@yonsei.ac.kr ) ****Prof. Myongji univ.,dept. Mechanical. Eng(TEL: 010-4641-6344;E-mail:kang@mju.ac.kr) Abstract When we pick up some loads again and again, the performance capacity of our arm muscle decreases. This decrease in performance is called muscle fatigue. This fatigue can be identified by a decline in the frequency spectra of the EMG signal. In addition this muscle fatigue has an effect on EMG to torque model. The parameters of the EMG to torque model changes as fatigue of the muscle approaches. Similarly if force is applied with different joint velocities, activation of the same number of motor units produces a different level of force. So it means that both muscle fatigue and joint velocity have effect on muscle force, which ultimately effect joint torque. In order to validate these assumptions we first analyze and monitor the muscle fatigue by analyzing EMG signal median frequency, mean frequency and power spectrum. Then we introduce our EMG to torque model and study the effect of muscle fatigue and different joint angular velocities on the model. The EMG to torque model is derived previously using non-linear regression. In order to record EMG signal from bicep muscle, joint angle and joint torque data, ten subjects took part in the experiments. They were asked to perform maximal voluntary contractions (MVC) repeatedly until fatigue of muscle is reached. They were also asked to do MVC with different joint velocities. The experimental analysis and the results of EMG to torque model show that our model successfully map EMG signal to joint torque in fatigue, non fatigue MVCs and with different joint velocities. Keywords EMG, Muscle fatigue, Non-linear regression, Power spectrum 1. Introduction In human body muscle is a source of producing force. This muscle is consisted of motor units. When some force is applied by the muscle an electrical signal is generated on the muscle surface which is a summation of all the action potential generated in motor units. This signal is called electromyogram (EMG) signal. There are some mechanical and physiological factors such as muscle length, muscle fatigue and joint velocity which can have an effect on EMG signal and thus effecting EMG to torque model [1, 2]. Over the years, numerous attempts are being made by researchers to find a relation between EMG signal and joint torque and the effect of muscle fatigue on the relation between EMG and torque. Fatigue occurs when a prolonged voluntary muscle contraction is sustained; due to this some motor units on the muscle are activated. After being activated for a period of time, the activated motor units start to develop fatigue due to factors such as insufficient supplies of oxygen and glycogen, increased lactic acid level in blood and muscle, etc [3]. This muscle fatigue can be identified by a decline in the power spectrum of the EMG signal. The effect of this fatigue on EMG to torque model is investigated by various researchers. Fatigue is expected to change the amount of force produced for a given number of motor units being activated. Some researchers say that torque of maximal voluntary contraction (MVC) declined and the contraction and relaxation prolonged after fatigue [4]. Muscle fatigue also results in slowing of contractile speed and increase recruitment during submaximal isometric contractions [5], the error of movement, especially fast movement, should increase. This means that EMG signal characteristics are changed as fatigue is developed. Similarly joint angle and joint angular velocity also affect the EMG to torque model. The parameters of the model changes when same force is applied with different velocities. Previous studies have examined the effect of walking speed on muscle activity using EMG measurements, but these studies either examined a limited set of muscles or walking speeds [6]. These studies only concentrate on effect of different joint velocities on EMG signal. Therefore, the goal of the our study is to examine EMG signal of the bicep brachii muscle groups across a wide range of speeds to assess whether all muscle activity systematically increases in response to increasing contractions speed and if the EMG