Transcript using the learning method

GRRC Int. Workshop 2008
Development of Surface EMG Sensor
Network and its Application System
Youngjin Choi
Hanyang University
Contents
•
•
•
•
•
•
Introduction
Development of EMG Sensor
Arm Motion Tracking Algorithm
Experimental Results
Conclusion
Future Works
What is EMG
• EMG(ElectroMyoGram)
- is one of various bioelectrical signals generated from the human body such as
ECG, EEG, EOG, ENG.
- has been actively studied for the human muscle analysis, motion imitation, etc.
- is one of the best basis signals to develop the bio-mechanical system by
transferring the robotics technologies to the rehabilitation engineering
• Application research filed
- diagnosis in the medical science
- sports science
- rehabilitation engineering
Literature Surveys
• Existing studies about the EMG
- using the learning method
(D. Nishikawa, “EMG Prosthetic Hand Controller Using Real-Time Learning Method”)
- using the AR model
(J. Zhao, “ Levenberg-MarQuardt based neural network control for a five-fingered prosthetic hand”)
- using the Hill model
(E. Cavallaro, “Hill-based model as a myoprocessor for a neural controlled powered exoskeleton
arm-parameters optimization”)
- using the pattern recognition for the motion generation of real hand
(Y. Su,“Towards and EMG-Controlled Prosthetic Hand Using a 3-D Electromagnetic Positioning System”)
- using the ARMAX model for the tele-operation
(P. K. Artemiadis, “EMG-based Teleoperation of a Robot Arm in Planar Catching Movements using
ARMAX Model and Trajectory Monitoring Techniques”)
Development of EMG Sensor I
- Circuit design for signal acquisition
High Pass Filter
for DC voltage rejection
Low Pass Filter
for noise rejection
1000 times
Amplifier
Signal Input
Voltage Adder
for AD conversion
Analog Switch
for op-amp drift prevention
Development of EMG Sensor II
• Characteristics
- 4 Channel EMG Measurement
- Fast Microprocessor
- 150M[Hz]
- High Resolution
- 12bit ADC
- EMG Sensor Network
Surface-EMG Waveforms
•The electrodes are for the acquisition of biceps and deltoideus EMG muscle signals.
•The biceps muscle signal is used for the elbow joint angle extraction.
•The deltoideus muscle signal is used for the shoulder forward directional angle extraction
•The maximum amplitude level is about 300[uV].
•The EMG signals are amplified by about 1000 times for the signal processing.
Range of Motion (ROM)
• Biceps brachii
• ROM of an elbow joint
• 0~145 degrees
• Deltoideus
• ROM of a shoulder joint
• 0~180 degrees
Motion Tracking Algorithm I
• Taking RMS values
- take the RMS value for EMG[k] grouped by 64 samples
Motion Tracking Algorithm II
• Taking LPF
- filter out the signal by using the 1[Hz] low-pass filter
Motion Tracking Algorithm III
• Scaling function
- LPF signals are not proportional to the flexion and extension angles of elbow and
shoulder joints
- So, we make use of the curve-fitting method
- For this, we assume the 3rd order scaling function for 4-point curve-fitting
Motion Tracking Algorithm IV
 Pre-angle Process
- Taking RMS
- Taking LPF
- Applying scaling function
Motion Tracking Algorithm V
• Optimization Process
• Finally, the recursive-least-squares method is applied to the
pre-angle for self-adaptation
• In this case, the tap-weights are adjusted by itself during the
optimization procedures.
Coupling Effect b/w Biceps and Deltoideus
- Though the signal measured at a deltoideus muscle must be dominant for the
shoulder joint motion, the sEMG signal measured from the biceps brachii has the
coupling effect with the shoulder motion
- So, we remedy the equation by subtracting the Deltoideus from Biceps brachii
Biceps brachii sEMG
Deltoideus sEMG
Block Diagram of Entire Algorithm
Experimental Results
Experimental Video
• The initialization procedure is firstly performed.
• At the specific joint angles measured from inclinometer, we measure the EMG signal
values for the biceps and deltoideus muscles, respectively.
• Then, we can get the scaling function for elbow joint and shoulder joint angle calculations.
• And then, we perform the real-time motion tracking simulation.
•As we can see the video, the experimental results show the real-time good tracking performance
for human arm motion.
Concluding Remarks
• We have developed EMG measurement sensor
• We have suggested the real-time motion tracking
algorithm based on the surface EMG signal
processing
• We have showed the validity of the suggested
algorithm through the experiments
Future Works
• EMG sensor networking
• Applications
▫ Master (human) arm device for tele-operation
▫ Prosthetic arm for an amputee
Thank you
Question : [email protected]