Author: Andy Sigler

Our project has two paths that are about to merge, one being the WREX exoskeleton, and the other being the muscle stimulation.

The WREX has taken up most of our time, and we we now have a design to start building a working model off of. Last week, we sanded down our thoughts on how the levers and joint would function. Now we have a 3D model of the WREX arm thanks to Will, based off of measurements taken from out arms, and off some geometry Carl ran up.

This coming week will consist of use putting together the initial materials for manufacture, and once compiled, figuring out:

1. How to attach it to a chair.
2. How to make the shoulder joint strong enough to support our arm.
3. Where and how to place the joints and leverage points (rubber bands).
4. How to attach to arm itself (molded cast / velcro).

The second path is the electrical stimulation. We received the device in the mail last week, and I have taken the liberty of trying it out, on myself and others.  I have found that different users have different responses, and that to get the widest range of motion, the user’s muscles must already have some relative strength to them. Users who did not have “strong” arms were not able to get their elbows close to 90 degrees, while others were easily able to reach that point.

While the first and most obvious muscle to jump after was the bicep, we also looked at the forearm and shoulder muscles. The shoulder muscles were difficult to control, as they are very close to each other, and require a high level of current to get movement. Also, they’re a little too close to the chest (heart) for my comfort. The forearms, however, were easy to manipulate, and required less current. Namely, the wrist flexors, on both the top and bottom of the forearm, were easy to control, and created movement with half as much current as needed for the bicep and shoulder. We were able to reach 90 degrees from most testers. We tried both a steady pulse width, at 40hz most of the time, but occasionally tried much lower frequencies to achieve a jumping effect in the muscle.

Once we are able to combine these two paths, we will then start working on hacking our stimulator, so that we can control the pulses without having to interact with it’s built in buttons. This is a scary step, because we don’t want to break it, plus it’s electricity…going into our bodies. The most viable approach would seem to be opening up the device, finding the contact points that the buttons close, and manipulating those through first our own switch, and then perhaps a transistor so that we can control it with software.

 

For my part of the lit review, I found this project called Possessed Hand, which came out of the University of Tokyo in 2010. The project used an arm strap, worn around the foreman, which contained 24 electrodes for stimulating the muscles of the hand and wrist. The goal of the project was to control fine movements of the hand, so as to eventually play the koto, a Japanese instrument, remotely. Their research gave an answer to my initial question when starting out with this idea, and that was whether fine control is feasible for controlling the hand to do meaningful movements. They found that, through electrical stimulation, the hand was not able to move accurately, nor with enough force to play the koto, or even lift an object. After failed experiments in these areas, they concluded that their device could be used to help teach people how to play the koto, by giving them stimuli as feedback for when/how to play, instead of simply playing FOR them. This leads me to think that our project for this class should need to lean away from fine movements, and more towards interesting applications. Perhaps the small muscles of the hand are out of the question, while the body’s larger muscles may be too big to get their corresponding limbs moving. Also, relating back to my first blog post in this class, I would be interested in stimulated these muscles for the goal of impacting some external object. But of course, this impact would have to be less refined than a koto, while still interesting.

Their research also revealed a bit of the user experience when being stimulated by such a device. They recorded comments by testers, saying that they felt ‘hacked’, they wanted to be controlled remotely, or it was ‘scary…just scary’. I feel this is a fascinating aspect of controlling movement with EMS, as the concept of controlling another person’s body is a sensation that few, if any of us, have had, and also in reverse, being controlled.

Another source I found for starting to think about applications was this project by the artist Choy Ka Fai. He used audio signals, generated from Max/MSP, to serve as the pulse-source for his muscle stimulators. His goal was to stimulate muscles so as to mimic the movements of a deceased dancer. He has videos and images documenting his progress, and he seems to have come across the same limitations as the Possessed Hand folks, in that he wasn’t able to achieve a level of control he seemed to be happy with, and so instead used the stimulation as a sort of haptic feedback for learning muscle movements and performance.

Fai ended up applying his electrodes to a group of dances, allowing them to feel the pulses, and essentially dance like this deceased dancer. This is something I find very interesting, using the stimuli to control a group of people in the same way. The video we watched in class of the faces all synchronized is similar to this, but they were recorded at different times. Perhaps it would be interesting to have a group of people all controlled, so they they could be ‘played’ by a traditional input device, or even a myo-electric input device we make :).

I’m drawn towards using the conductivity of our bodies to either stimulate muscles or senses, or muscle and body activity as an input. A few of the students here this past semester have been working on a sensor called the Touche. It measures the capacitance of any conductive body or object along a frequency sweep, so that a sort of spectrum analysis is done. This gives the sensor reading much more than a linear reading, so that machine learning algorithms can map it’s different positions, or the readings can be mapped onto more complex forms of data. I’ve been dying to try out this sensor, and so I’ve started brainstorming on different projects applicable to this class. The paper can be found at this link, and they seem to feel tied to using machine learning to memorize the different positions for a given body. This would be the ideal way to map the sensor to a range of values or few different states, but it needs engineering skills I don’t possess, and more importantly to me it costs time, somewhere around a few hundred milliseconds at times.

Another idea I’ve looked into is using haptic feedback to augment, replace, or accent our current senses. Sanniti has the idea of using vibration and motors to help deaf dancers stay on beat, and I’ve found this paper on a group of people who made a haptic feedback suit. Their results showed that users showed signs if increased sensitivity to sounds and touch when added to by some form of haptic feedback. This seems promising to be a project with strong results, plus it would be fun to work with music, and wireless wearables. The big questions with this idea would be the design of the wearable, so that it would impede the dancer’s abilities, and also the design of where and what the feedback is. I’m leaning towards vibrations near the head, but this isn’t based on any research just yet.

My third idea is from the topic of using the electrical activity if our muscles as a computer input. This paper I found focuses on the sounds our muscles make while being in use, but this is really just amplified currents in our muscles. These changes in amplitude could be measured and mapped to any input. I recently saw a product being advertised called the Myo, and it seems to me that, like the Touche, it requires heavy coding and also costs time, however simple amplitudes are less complicated than frequency responses, so perhaps it wouldn’t be as big of a problem. This could then be mapped to some sort of game, instrument, or whatever form of media.