__________ © Centre for Advanced Research on Energy
Decoding wrist gesture with combinational logic for the development of a practical EMG electrode sleeve
Z. Fu
1,
, A.Y. Bani Hashim
1
, Z. Jamaludin
1
, Imran Syakir Mohamad
2 1
Faculty of Manufacturing Engineering, UniversitiTeknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia.
2
Faculty of Mechanical Engineering, UniversitiTeknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia.
Corresponding e-mail: zinvifuyahoo.com
Keywords: EMG, human machine interface, synthetic system ABSTRACT - One of the reasons why EMG still lags
as a machine input signal is because EMG is a physiological signal and differs between individuals. As
a result, EMG devices also require training regardless whether it is a single user or multiple users. It is
hypothesized that although the EMG signals are different among individuals, they still contain some
similar traits that have potential for cross-user reusability. The characteristics of an EMG signal are
studied across a sample of subjects. These signals are recorded and analyzed for similarities.
1. INTRODUCTION
An Electromyography EMG signal is an electrical signal that is generated when a muscle
contracts. The EMG signal is a result of an electrochemical ionic exchange that takes place at the
nerve-muscle junction during muscular motion. By using electronic pickups, these signals can be detected
and utilized in various forms. Typical applications include motor-neuron medical analysis and more
recently, gaining momentum as an input signal for computing and robotic systems.
As neural interfaces for machine input are increasingly studied, the EMG signal is an attractive
alternative because EMG signals can be acquired from the skin surface. This form of EMG is known as surface
EMG sEMG and it can be obtained via non-invasive means.
Although there are many researches on improving the acquisition and decoding of sEMG, commercial
productization is still distant. The underlying reason is that sEMG is a physiological signal and variations exist
among people. Even for a single subject, the sEMG recorded off the same muscle may deviate due to muscle
fatigue, skin condition and also sEMG electrode placement [2]. Due to these reasons, sEMG machine
input systems are not cross user applicable and even a single user requires a considerable calibration and
training time [3]. 2. METHODOLOGY
In order to make the sEMG system viable for everyday uses, it must be practical and user friendly.
Unfortunately the sEMG system setup is complicated and the user needs anatomic knowledge of the target
muscles, since precise electrode placement is critical [4- 6].
The user of the sEMG system should not be burdened with clinical setup. In this research, the
approach to the problem is to develop the sEMG system with minimal electrodes. In order to achieve this, fewer
but significant forearm muscle were targeted, while maximizing the number of classifiable gestures.
A grid coordinate system was introduced for the lower forearm Figure 1a. The target muscles were
clearly mapped. For a single user, this system considerably cuts down initial setup time.
The grid system will be implemented onto a flexible material that would be constructed in the form
of a wearable sleeve. This sleeve would stretch with the skin as the forearm rotates Figure 1b.
a b
Figure 1 Coordinate system for locating muscles in the forearm
flex, extend,
abduct, adduct,
hand_open, hand_grasp,
f e
b d
o e
g f
w FCR FCU
FCU FDS FDS FCR
w ECR ECU
ECU EDT ECR EDT
w FCR ECR
w FCU ECU
h F
FDS h
F EDT
1
Together with the coordinate system, the motion of the wrist and hand muscles was also defined as logical
functions. Six gestures were defined and their
82 algorithms are shown in Equation 1. The gestures are
wrist flexion and extension, wrist adduction and abduction and hand open and close. Only major muscles
close to the elbow were considered because they are large and less susceptible to shift during forearm
rotation.
3. RESULTS AND DISCUSSION
A simple discrete logic relationship has been established for the wrist motion decoding. Based on
Equation 1, logic circuits can be built to perform the functions. An example of the circuit to decode wrist
flexion is shown in Figure 2. This is however, a simple circuit to serve as the foundation for the ensuing
processing stage. The sEMG signals still have to be filtered and processed with feature extraction and
classification algorithms to attain a satisfactory precision.
FCU FCR
FDS Wrist flex
Figure 2 Logic circuit for decoding wrist flexion The second phase of the will involve the
development of a wearable electrode sleeve. The electrode sleeve will be tested on a sample of subjects
that represent the target users. Basing on [7], the muscles in Equation 1 will fall in the same proximity.
The larger surface size of these muscles will offset the anatomical difference between individuals. We expect
the sleeve to speed up setup time for the user whereby the electrodes are already preset, as opposed to having
to fix the electrodes individually. The third phase of the research will focus on
recording and analyzing the sEMG signals from the eight muscles listed in Equation 1. The target muscles
are listed in Table 1. Table 1: Major muscles and their functions
FCU FCR FDS ECU ECR EDT
Flex 1
1 1
Extend 1
1 1
Abduct 1
1 Adduct
1 1
Hand open
1 Hand
Close 1
The sEMG signal waveform will be acquired with the Noraxon EMG acquisition device. Waveforms from
the sample will be compared among subjects and the variations will be recorded. An example of an EMG
waveform is shown in Figure 3. Analysis will be done on the power spectrum, frequency, bandwidth and phase
to identify possible similar traits that can be used as a platform for cross-user compatibility.
4. CONCLUSIONS
The aim of this paper is to propose an sEMG input system that is practical for real-world appications. The
coordinate system proposed as a graphical recording system so users can speed up the setup process. The
proposed logic gate decoding system is a conceptual platform for the future development of feature
extraction and classification.
Figure 3 Typical waveform of the sEMG signal [8] Future research will be directed to analyzing the
sEMG signals of various individual and also to the construction of the electrode sleeve.
5. ACKNOWLEDGMENT
The Malaysian Ministry of Education supports this research
through the
research grant—
FRGS22013SG02FKP022F00176.
6. REFERENCES [1] S. Herle, S. Man, G. Lazea, and P. Raica,
“Myoelectric Control Strategies for a Human Upper Limb Prosthesis,” J. Control Eng. Appl.
Informatics, vol. 14, no. 1, pp. 58–66, 2012. [2] A. Stoica, C. Assad, M. Wolf, K. S. You, M.
Pavone, T. Huntsberger, and Y. Iwashita, “Using arm and hand gestures to command robots during
stealth operations,” in Multisensor, Multisource Information Fusion: Architectures, Algorithms,
and Applications 2012, 2012. [3] S. Bitzer, A. E. M. G. Hardware, and P. van der
Smagt, “Learning EMG control of a robotic hand : Towards Active Prostheses,” in Proceedings of the
2006 IEEE International Conference on Robotics and Automation, 2006, no. May, pp. 2819–2823.
[4] J. N. A. L. Leijnse, N. H. Campbell-Kyureghyan, D. Spektor, and P. M. Quesada, “Assessment of
individual finger muscle activity in the extensor digitorum communis by surface EMG.,” J.
Neurophysiol., vol. 100, no. 6, pp. 3225–35, Dec. 2008.
[5] A. Y. Bani Hashim, M. N. Maslan, R. Izamshah, and I. S. Mohamad, “Delivering key signals to the
machine: seeking the electric signal that muscles emanate,” J. Phys. Conf. Ser., vol. 546, pp. 1–4,
Nov. 2014. [6] G. J. Tortora and B. Derrickson, Principles of
Anatomy and Physiology, 12th ed. John Wiley Sons, 2009, p. 1174.
[7] R. . Northrop, Noninvasive Instrumentation and Measurement in Medical Diagnosis. Florida: CRC
Press, 2002, p. 113.
flex, w
f
= FCR.FCU + FCU.FDS + FDS.FCR
