Built by Wolves for Wolves
EMG Controller Prototype:
The EMG (Electromyographic) Controller was a project that I did for my BME595 class, it is now the basis for the bionic tail
The following is the theory of operation taken from my original proposal:
External Device Control
Via Electromyographic Signals
The purpose of this research project it to
explore the controllability of external devices with signals picked up from the
human body. The signals to be
picked up will be electromyographic (EMG).
The EMG signal will be picked up via silver/silver chloride (Ag/AgCl)
adhesive pads connected to a differential amplifier.
The EMG signal out of the amplifier will
then be sent thru a band pass filter to remove all unwanted signals and then
processed in a microcontroller. A signal from the microcontroller will then be
produced that is proportional to the amount of muscle activity.
This signal will then be used to control an external device.
External devices that could be controlled
via this unit range from an artificial limb to a mouse pointer on a computer, in
actuality any device requiring a proportional input could be controlled, since
the correct signal needed for control of that device, could be generated via the
The specific aims of this research project are to test the feasibility of proportional control of an external device, via an EMG signal. This will be accomplished using different hardware filter designs and software algorithms to quantify the EMG signal, in an attempt to produce higher resolution (finer control), then what is currently available. This device will then be used to determine which muscle groups give the best results and trainability and repeatability, when controlling an external device.
EMG signals will be
picked up by surface electrodes (Ag/AgCl), see Figure 1, they are found to have
amplitude of 100mV
to 1 mV, in the frequency range of 25Hz to 5Khz. The differential amplifier will thus have a total gain of
around 5,000 to 50,000 to get the pickup signals to around usable logic levels.
The output stage of the differential amplifier will have a band pass
filter to remove background radiation, such as the 60 Hz signal found in most
inside environments. The band pass filter will have a frequency response of 100 Hz
to 4Khz. Once the EMG signal has
been recovered and amplified it will have to be quantized.
Quantitaztion will be preformed by rectifying the signal then sampling by the analog to digital converter in the microcontroller. Software algorithms will be developed and used to retrieve a usable control signal from the sampled EMG signal. Resulting in any device to be proportionally controlled by the human body, via the microcontroller’s output.
following is the Prototype for the EMG
Controller, currently a single channel pick up with filters, and microcontroller
capable of controlling a single servo proportionally, this is what the simple
block diagram turns into on a breadboard:
EMG Controller Final Prototype:
This is the final prototype, after about 6 months of research and development...
Below is a Silver/Silver Chloride (Ag/AgCl) pickup, they are commonly used for monitoring signals at a hospital, they come in packs of 7
These pickups are placed 3 on each leg, so that the tail can move in time with your steps
Note the Snap connector on the Ag/AgCl pickups they plug into 6 leads going into the EMG Controller
The 6 leads, 3 per channel, go into the EMG controller
The black wire is ground and the red and yellow wires are the differential signal input, they enter from the right on the pic below
The "Altoids" box is used for its excellent shielding properties
Each channel is kept separate until the microcontroller, so the top and bottom halves are identical
Moving from the right to the left, the first quad op-amp chip is configured as a differential operational amplifier
The second quad op-amp chip (to the left of the first) is configured as an active 2nd order band pass filter
Below the potentiometers are the output stage conditioners, they perform two tasks, they take the signal and rectify it so only the positive part is seen by the microcontroller, and then it is clamped to make sure it never goes above 5 volts
The two potentiometers are used to adjust the threshold triggering point for the left and right channel, this will allow for adjustment of how much muscle activity will cause the tail to move
The microcontroller then takes in all 4 signals (2 from the potentiometers and two from the output stage conditioners) on its ADC (analog to digital converter) and performs a software filtering on the signals to remove the "jitters" from the EMG signals
The microcontroller then decides if the person is standing, sitting, walking, running, etc... and outputs commands to the tail to move it accordingly
The commands from the microcontroller are sent via Serial UART to the tail, cable pictured on the right in the above pic
The EMG controller is powered by two 9 volt batteries, cable pictured on the left in the above pic
During the test phase I configured one of my LCD con-badges as a RS232 Snooper and had the EMG microcontroller output the values seen on its ADC to the UART, where my Snooper could display the values
The 'I' and 'J' values are the left and the right channels respectively and the 'K' and 'L' values are the two potentiometers channels respectively
The left pic is when I am standing on my left leg, and the right pic is when I am standing on my right leg
Noted Values of leg muscles
at rest, below 50
standing, around 200
standing on one leg, around 400
going up a stair, around 700
Bio Controlled Tail
Tail actuator module currently under construction, now using 180.53 oz-in of torque servos
Tail with out covering, note that snaps are now used to hold the sheath and batten, this is so the user can remove the covering and easily make any repairs to the tail spine, if needed
Finished Tail with Tiger looking covering