INTRODUCTION
Exoskeleton arm is an outer framework that can be worn on a
biological arm. It is powered by actuators and can provide assistance or
increase the strength of the biological arm, depending on the power of the
actuator. EMG is the suitable approach for human machine interface with the
help of exoskeleton.
When working with EMG we actually measure the motor unit
action potential [MUAP] generated in the muscle fibers. This potential builds
up in the muscles when it receives a signal from the brain to contract.
The Nerve Potential
•
MOTOR UNIT ACTION POTENTIAL (MUAP) is
generated on the surface of our arms whenever we contract or relax our arm.
•
Amplitude is in order of 0-10 millivolts
.
•
The frequency in between 0-500Hz.
•
This MUAP is the core of this project
and the basis of EMG processing.
THE
EXOSKELETON ARM
•
It is an outer framework that
can be worn on a biological arm.
•
It is uses a Non-invasive method to
acquire MUAP from muscles to control the framework, that can be worn on a
biological arm.
•
Powered by a high torque servo motor.
•
Can provide assistance or increase the
strength of the biological arm, depending on the torque of the servo motor.
•
Electromyography (EMG) is the suitable
approach for human machine interface (HMI) with the help of exoskeleton (EXO) .
Software Used:
1).KEIL uVision for compiling the code and
monitoring the signal.
2). Multisim for circuit simulation.
Hardware Tools:
1) Microcontroller
board: EVAL-ADuCM360 PRECISION
ANALOG MICROCONTROLLER (Analog Devices Inc.)This
microcontroller board is used in our project as the brain to control the
exoskeleton arm. This process will be used for interfacing our EMG sensors with
the arm (servo motors).
2) Instrumentation
Amplifier: AD620AN (Analog
Devices
Inc.)This receives signal from EMG electrodes and
give the differential gain as the output.
3) OP-AMP:
ADTL082/84(Analog Devices Inc.)The output from the
DIFFERENTIAL AMPLIFIER is rectified and this output is fed to the LOW PASS
FILTER and then to the GAIN AMPLIFIER.
4) SERVO
MOTORS: 180kg*cm torque. It is used
for the movement of the arm.
5) EMG
Probes: For the acquisition of signal.
6) Battery:
Two 11.2V, 5Ah Li Po battery,
it will be used to power the servo. Two 9V battery to power the
EMG circuit
7)
Metal
Frame
METHODOLOGY
The exoskeleton arm
works in two modes .First mode is automated mode in which EMG
signals after the signal
processing will command the servo and second manual mode ,a potentiometer will command servo motor .
ADI COMPONENTS USED:
1)AD620 Low cost low
power instrumentation amplifier.
2)ADTL084 Low cost
JFET input operational amplifier.
3)EVAL-ADuCM360
Precision analog microcontroller.
Various stages in EMG signal processing:
1) Signal Acquisition: The Motor Unit Action
Potential (MUAP) signal is acquired from the bicep and triceps of the patient’s
arm. Three EMG electrodes are used in the process. Two EMG electrodes are
placed on the bicep and triceps, one on the elbow for ground reference. The
acquired signal is fed into the AD620 high quality instrumentation
amplifier. Which will amplify (gain=500) the potential
difference between the active
electrodes.
Gain of
instrumentation amplifier G=1+49.9KOhms/R
1 2) Filtering
and Amplification: This amplified signal is then fed to a dc coupling
capacitor and a full wave rectifier which eliminates DC error offset
and negative half cycles to make the signal compatible with the
microcontroller. This rectified signal then goes through a low pass filter
to eliminate high frequencies and make an envelope of the signal. The signal is
sent into an amplifier with variable gain for further amplification. All the
stages are designed using ADTL084 op-amp
Gain of op-amp Vout/Vin=-Rf/Rin
Fig: signal
after full wave rectification.
Data
Acquisition:
The amplified signal is fed to a microcontroller
EVAL-ADuCM360 PRECISION ANALOG. The analog voltage is read by ultrahigh
precision 24-bit ADC present
in the microcontroller. The data is sampled at a rate of 2.450 kHz. ADC Chopping scheme is
used. This chopping scheme results in excellent dc offset and offset drift
specifications and is extremely beneficial in applications where drift and
noise rejection is required. The offset obtained when the muscle is relaxed is
subtracted from the ADC output
Control
Logic:
Since noise rejection is required in the final
stage, linear mapping of the ADC
output to the DAC is avoided. We have created a lookup table that writes discrete
values to the DAC. Don’t care conditions are created for low voltage analog
signals so that the servo is not activated unnecessarily. The threshold for
maximum voltage is set manually, after testing, as it is different for every
test subject.
DAC:-
The microcontroller comes with a 12-bit DAC. The DAC has two selectable ranges: 0 to 1.2 V & 0 to 1.8 V. Coincidently 1.8V input
to servo motor gives the optimum turning angle for the servo motor. There this
range is used as it requires no further amplification.We
have used DAC Interpolation Mode .The interpolation mode uses 16-bits.
12-bits are used for the writing the data and 4 bits for interpolation.
Servo
Motor:
The servo motor has a torque of 180kgcm. It runs on
two modes Pulse Width Modulation and Potentiometer mode (analog signal).We
have used the analog mode because it is easier to monitor and analyze as
compared to PWM. When given an input of 5V the servo turns 270 degrees .It runs
on 14V to 30 V. 30 V for maximum torque .
Exoskeleton Arm on a test subject
Challenges:-
While working on
EMG sensor we found that the ADTL084
op-amp are only available in SOIC
and TSSOP Packaging so we came up an
alternative SOIC to DIP adapter board.
High torque
servo motor (380 Kg. Cm) that we had bought from China gone malfunction with no
other options we replace it by new one.
We discussed
about which material to choose for the frame and its design eventually we
decided to go with aluminium because it is the lighter and stronger metal. A
backpack is designed to hold other parts such as EMG circuit, microcontroller
and batteries.
There was an offset in the final stage of the EMG
circuit, it could be reduced by decreasing the gain of the final amplifier. But
reducing the gain also reduce the amplitude of the signal. So extra efforts had
to applied to activate the servo.
But for patients undergoing rehabilitation such high efforts was not possible.
Therefore we had to increase the gain.
To handle the problem of drift, we subtracted the
offset value from the input.
First linear
mapping was done between ADC and DAC output. But when we increased the
sensitivity of the setup for physiotherapy patients, the servo started rotating
randomly due to the high sensitivity. To tackle this problem we used Boolean
logic to give discrete analog values to the servo and rejected signals with low
amplitude.
Future
prospect:
With further research on this we can improve the
design of the needs to meet the needs of the patient in rehabilitation centers .
Instead of using servos, alternate methods can be used to power the arm like
NANO MOTORS(Smart Memory Alloys) . When extreme power is required hydraulic systems can be used.
This design provides support for the arm to lift the objects only. We have been
working on acquiring signals from the triceps as well, so that objects can be
pushed as well. Mechanical suits have been Sci-Fi films from a long time We
think with further research we can arrive to a point where IRON MAN suit is
becomes a reality.
The team consists of three students namely Surya Pratap S Deopa ,Naval
Kishore Mehta, and Abhishek Saxena .
No comments:
Post a Comment