Magic in a Box: My First Day Using Myo
TL;DR — I’ve got a Myo. It is comfortable, has a reasonable battery life, seems durable, and it works for long stretches without adjustment. Hurray!
UPDATE: For a data heavy analysis of the Myo checkout my follow up.
My Myo developer kit arrived Monday. If you’re not sure what that is, checkout the video buried at the bottom of this page. Delivered something like five days prior to the delivery date provided by Canada Post, I was pleasantly surprised. The only problem? I’m heads down, writing the first chapter of a book on Docker. The book is titled, Docker in Action and it is currently available through the MEAP program. Priorities are nothing if not frequently challenged. So, after tinkering in the evening, I set it aside.
The plan? I’m going to wear this thing all day.
But not for long. This morning I grabbed it along with my other gear on the way out the door. The plan? I’m going to wear this thing all day. I wanted to really put it to the test as a first generation consumer product. I know its not really intended to be a wearable. But I was going to wear it none-the-less. I wanted to measure how accurate it was. I wanted to see if I could stand wearing it for long periods of time. I wanted to see if it would get in the way, or when I’d bump it on things. I wanted to see how often I’d need to recalibrate it. I wanted to see how well it would hold up.
I haven’t seen much in the way of detailed use reviews. And while I doubt this is why Thalmic Labs sent out developer kits early, I also doubt that a customer review could do much to help or hurt the product. I do know that quite a few people have already asked me about it, and I’d rather not repeat myself. ☺
So, I set the terms of the experiment. I’d wear it for as long as I could for the day. I wouldn’t take it off unless its obvious that the activity I was doing would put the device at risk. I decided that I would measure the accuracy of the device by attempting to make each gesture and validating my success in the pose window. The pose window is a small window that shows you the current gesture that is being detected. Every hour on the hour I would perform each gesture 10 times with my arm at rest on a desk or table and 10 times elevated or hanging at my side. When it failed, I’d note what gesture was detected. I’d note any time that I had to recalibrate the device, either because I had to briefly take it off or I bumped it out of position. On battery life, I’d note at what point the battery was low and how the device expressed that fact. I’d also note the time when it dies. At the end of the day, I’ll look for dents, scratches and scuffs.
Another thing I decided that I’d keep track of are the moments that I wanted to use it to do something, but can’t yet. I think this will be of particular use when I’m sitting around wondering what I should build for it, or really, when any of my more tech savvy readers are doing the same.
I wore my Myo a total of six continuous hours without adjustment.
The following are my results.
This is a beta product.
I know this, and I’m not sure I can state it any more clearly.
Even if wearing it higher on my arm improved accuracy I would hesitate to do so. At around a quarter of an inch thick, you’d notice if your arm was resting on it.
The first thing I’d like to talk about is comfort. Thalmic Labs advises the wearer to place the Myo as far up the arm as comfortable and ideally a few inches below the elbow. As you can see in the photo above, I wore the Myo at about mid arm. The day prior I had experimented with the fitment and accuracy at different points on my arm. I found that my ridiculous forearms were most comfortable and the device was most accurate in this position. This was actually quite lucky. As I sat at a table resting my forearms and typing, the Myo was far enough down my arm, that the biggest part of my arm could rest without discomfort.
Resting my arm on a desk was distracting at first but comfortable.
“But why would I wear a Myo whilst sitting at a computer? Why wouldn’t I control my computer with the keyboard and mouse? You should be walking through the woods, guiding a drone or something.” Those are totally fair questions. I’m wearing one at a desk because I’m developing software for it. I’m writing about wearing it. I’d like to use gestures to augment other functionality that you get with things like Mac trackpad gestures. I’d like to fire off a script with a flick of the wrist, without jumping to another window and typing a command. But really, given the process for putting one on and calibrating it, this is the sort of gadget that you put on and leave on.
At first I felt shortsighted for not seeing this coming. The device is thick, maybe a third again as thick as my iPhone 5s. So of course I’m going to notice that its on my arm. But one saving grace in my opinion is that the feeling is uniform around my arm. The band is wide enough that any pressure from resting on it is distributed on a pretty wide surface area. This is good because its never painful.
In fact, the odd feeling of wearing it while typing or resting just might be offset if you were wearing one on each arm. I think it was the asymmetry more than anything else that felt odd.
The pressure from the band itself is firm but not too tight. At no point did I fear that this would slip off or giggle around, but I never felt like it was cutting off my circulation.
Thalmic Labs executed perfectly here.
Obviously if they found a way to reduce the thickness (outer less the inner diameter) of the sensors and thus the whole band, it would be better. I’m not so concerned with the width of the band itself. Because of the way the band stretches to meet the contoured diameter of your arm, it fits snugly and comfortably wherever it is placed.
Unlike every watch I’ve ever worn, the Myo never caused any itching at any point during the six hours that I wore it.
True to my goal, I didn’t move the device all day. This is evidenced by the pressure marks that remained on my skin after I removed the Myo. The battery died after hour six and so there was little point in leaving it on.
First, holy crap. Pretty much every other wearable maker around needs to get on the horn with Thalmic Labs. I cannot believe that I was not jumping out of my skin to get this thing off after an hour. I went six hours and it never itched. Not once. Whatever magic they used works.
“[Myo’s battery] is designed to last a full day of continued use”
Thalmic Labs’ claim that the battery should last a full day of continued use holds up. I charged the device the night before and left it unplugged before starting my experiment today. Between that and true continued use, six hours seems totally reasonable. If you were using this for your job, you’d likely have a few periods of downtime and a lunch in there somewhere. Any time you take the device off, it will put itself into sleep mode. But if you never take it off it will never sleep.
One potentially important caveat to this is that I made very little use of the haptic feedback. This means that in using the device today, I did not account for battery drain from vibration. Whenever you’re talking about mechanical energy, you’re talking about faster battery drain. So, be warned that if you plan on using a Myo with applications that are constantly activating the haptic feedback, the battery might not last as long.
Something to keep in mind when designing applications for the Myo, is that it is not designed to be worn while charging. This means that, you get about a full day of use. That is it. Some of you might be planning integration for sensitive and time consuming tasks, like controlling surgical devices. Please make sure that users don’t plan on one Myo lasting for the duration of a 16 hour surgery. Maybe the guidance that Thalmic Labs provided here is just a comfort thing. I certainly wouldn’t want to put the plug in while I’m wearing it. If that’s the case, then maybe you could feasibly extend the battery life with a portable battery charger.
This thing is supposed to detect hand gestures correct? How does it do?
The current software supports five gestures (not counting the calibration gesture). Those are:
- Touching your thumb to the tip of your small finger
- Making a fist
- Opening your hand wide
- Swiping your hand outward at the wrist
- Swiping your hand inward at the wrist
I tested each gesture ten times from rest on a desk and ten time while my arm was elevated or hanging by my side. I repeated this test at calibration and at the end of each hour. It is important to note that there are all sorts of factors that go into gesture detection accuracy. Thalmic Labs notes that position on the arm and rotation of the Myo may impact its accuracy. Their advice is to play around with it and find a spot that works well. So, that is exactly what I did. Once you find that spot it should be consistent and calibrated.
First thing to note is that the last two gestures, swiping your and inward and outward were always highly accurate. With my arm resting on the desk, or elevated and regardless of how long it had been since I calibrated the Myo, it never failed to detect either of these gestures. Awesome.
The second thing to note is that touching your thumb to the tip of your small finger takes just a bit of practice, but once you get it, Myo gets it consistently.
Third, the device really doesn’t move around at all. You have to really bump it against something to impact the accuracy. You can see this in the data where for the first few hours of the day Fist and Open Palm were both highly accurate. But after I went to lunch, when I was swinging my arms around walking, driving, bumping into things like tables and pews (went to the Cornish Pasty Co. in Tempe, AZ), Myo tested far less accurate for the Open Palm gesture. At that point I recalibrated the device so I could get data showing it drift again.
Fourth, resting your forearm on a table might impact accuracy for some gestures that use your small finger. But resting your forearm on the edge of that table has a measurable impact. I suppose this makes sense. If you’re resting your arm on a table, or edge of a table it will be applying pressure to the muscle or tendons that control your hands. Some interference should be expected. That being said, the cause and impact were obvious because I could actually feel that interference, and I knew when the Myo failed to detect the gesture.
Fifth, when Open Palm fails, it usually registered at Thumb to Small Finger. This might be an issue. Currently Thumb to Small Finger is used as the unlock and lock gesture because it is so difficult to do accidentally. If I’m always accidentally sending that gesture because I opened my palm while resting my arm on a table, I might trigger something I wasn’t expecting.
I’m not terribly worried about the minor issues I saw here. Thalmic Labs has stated several places that the gestures and the patterns used to detect them are all in flux and they are actively working on improving them. I believe it. But honestly, I think they are probably fine as is. I have opted into their data collection program. I hope that helps. ☺
All of the data I collected is appended at the end of this article.
The device is not quite rugged, but not exactly fragile.
I wore this thing all day. I spent the morning hours working with my arms at rest on top of a steel table top. I threw caution to the wind. I banged my arm against things and pulled no punches. Stopping just short of intentionally trying to scratch, dent or scuff it.
I’m happy to report that the device went nearly unscathed.
The only mark on the device was a very small scratch near the outward facing edge of one of the sensors. I have no idea how this actually happened since this particular sensor was positioned on the top of my forearm.
What I didn’t find really surprised me. I had fully expected to find wear or scuff marks on the outside surfaces of the sensors beneath my arm. I mean, they were rubbing against a steel surface for hours. Nothing. Not a single mark. Great work people!
I believe these things will hold up well. You won’t even need to baby them. Just treat them with a moderate level of respect and they’ll likely go the distance.
Occasions when I wanted to use it, but that use-case had not been implemented.
I took a few notes today, but I’m not ready to share them. Instead I’ve decided to continue the experiment and complete a week of wearing it every day. At the end of the week I’ll write another article and include all of the things that came to mind.
So, what is the verdict?
Okay, so, I would be really shocked if hundreds of millions of people were wandering the streets with these things like their iPhones. I doubt I’d ever overhear a conversation where someone was telling someone else about how rude it was to use a Myo at the dinner table. I really doubt anyone will be creeped out if you wear one on a date, though they may think you need to step away from tech for a while. All that being said, this could be a great tool and there is nothing from a user perspective to stand in its way. It is high quality, easy to use, and comfortable to wear. It will happen, but I’m having a hard time imagining someone complaining about needing to use one for work.
In the near future we may lose the liberation of our wrists, but we will gain the liberation of our hands. Thanks for reading!
If you’d like to chat about this stuff or be notified when I publish my One Week with Myo article, follow me on Twitter. In the meantime, there is a bit of a conversation going on over on Hacker News. So let us know what you think!
Raw Data (worn on right arm)
- Calibration (arm elevated):
Thumb to Pinky: 9/10 miss was open palm
Fist: 10/10
Open Palm: 10/10
Swipe Out: 10/10
Swipe In: 10/10 - After Hour 1 (arm elevated):
Thumb to Pinky: 8/10 (misses: open palm, swipe out)
Fist: 10/10
Open Palm: 10/10
Swipe Out: 10/10
Swipe In: 10/10 - After Hour 2 (arm elevated):
Thumb to Pinky: 8/10 (misses: open palm, swipe out)
Fist: 10/10
Open Palm: 10/10
Swipe Out: 10/10
Swipe In: 10/10 - After Hour 3 (arm at rest on desk — arm elevated):
Thumb to Pinky: 10/10 — 9/10 (swipe out)
Fist: 9/10 (just missed it) — 10/10
Open Palm: 4/10 (fist, swipe out, thumb to pinky) — 10/10
Swipe Out: 10/10 — 10/10
Swipe In: 10/10 — 10/10 - After Hour 4 (arm at rest on desk — arm elevated):
Thumb to Pinky: 10/10 — 10/10
Fist: 10/10 — 10/10
Open Palm: 0/10 — 3/10 (thumb to pinky)
Swipe Out: 10/10 — 10/10
Swipe In: 10/10 — 10/10 - After fail at Open Palm, I recalibrated with the same arm position until that gesture was repaired. My guess is that I bumped the device on the table during lunch. It was not moved enough to detect that it needed recalibration.
- After Hour 5 (arm at rest on desk — arm elevated):
Thumb to Pinky: 10/10 — 10/10
Fist: 8/10 (thumb to pinky) — 10/10
Open Palm: 10/10 — 10/10
Swipe Out: 10/10 — 10/10
Swipe In: 10/10 — 10/10 - At 1:40pm the flat light started periodically flashing orange (10 second interval). I think this is the low battery indicator. I’m going to keep wearing it until it dies completely.
- After Hour 6 (arm at rest on desk — arm elevated):
Thumb to Pinky: 10/10 — 10/10
Fist: 8/10 (thumb to pinky) — 10/10
Open Palm: 1/10 (thumb to pinky) — 10/10
Swipe Out: 10/10 — 10/10
Swipe In: 10/10 — 10/10 - Around 3:46 the device’s battery died. ☹ This is well within the range that I’d expect this to last.