New Tech: 3DWearable Motion Tracking

Triathlon Forum
1

Originally published at: New Tech: 3DWearable Motion Tracking - Slowtwitch News

There is no need to convince Slowtwitch readers of the importance of a bike fit and motion efficiency. So, we can skip the preamble in this profile of 3DWearable, a start-up that offers a system that combines motion tracking and analysis with real-time audio feedback.

Your bike fitter likely uses a range of metrics to dial in your fit with the goal of making you as efficient (and comfortable) as possible on the bike. Your bike fitter also likely teaches you how to pay attention to your body so that you can learn to feel what it is like to be moving optimally on the bike. This is not unlike what physical therapists refer to this as neuro re-education. The process takes time, as you can likely attest to the adaptation phase after a new bike fit being correlated with how much your position changed.

The only problem is that as soon as the fitting session ends, you are left alone without the motion metrics and the bike fitter’s teaching to reinforce what you just learned about your body moving optimally with the new bike fit. Further, you become dependent on your bike fitter to set up each new bicycle you introduce to your stable.

Motion Tracking System

The team at 3D Wearable sets out to supplement that scenario by creating a system consisting of two motion sensors that communicate via Bluetooth to an app on your phone. The two motion sensors can map the baseline of any motion (i.e., pitch, yaw, roll, acceleration) on a landmark on the body and provide real-time audio feedback of that same motion within an adjustable tolerance.

The simplest use case is saddle height. One of the motion sensors can be placed on your sacrum to locate your ideal saddle height and hip angle while pedaling. This motion can then be recorded as the ideal baseline. Then, while you are out on a ride you can receive audio feedback – a chime coming from your mobile phone – when you are outside a set degree of variance from the ideal motion.

As we fatigue, our ideal motion form is susceptible to deterioration, especially when it is a motion we are still training our nervous system to enact unconsciously. We get fatigued on a long ride, and we tend to hunch our back and roll our hips back. This then leads to compensation and potential for injury. The audio feedback from the 3DWearable system can help us correct our posture on the bike and prevent injury. It can also be an indicator that it is time to stop.

Training the Nervous System

As co-founder and long-time bike fitter, Eddie O’Dea, explained, we are constantly training our nervous system through repetition. Ideally, our training is reinforcing our bodies and nervous system to maintain an ideal motion for longer and longer durations of efficiency. As soon as we are falling outside a set range of an ideal motion, we are not only no longer reinforcing the ideal motion but also reinforcing a less efficient motion. We get tired and our bike posture (and running form) gets sloppy.

There is a very good chance we are all reinforcing less than ideal motion when stuck indoors logging hours on the trainer. I bet you can relate to moving your hands from the hoods to the bar tops, hunching your back, and rolling your hips backward late into an indoor endurance ride. Pay attention to it next long indoor ride, then imagine having audio feedback to remind you of when your posture on the bike is getting lazy. And then, consider the over-use injuries from compensation you would be avoiding.

Eddie went on to explain that the neuro re-education phase is accelerated with the audio feedback from the sensors. The chimes from the 3DWearable app allow athletes to practice the motion outside the lab or fitting studio where many other variables are at play. Newcomers to a sport or activity benefit from a wider tolerance variation as they are making far greater adjustments to their motion toward the ideal. Highly specialized, veterans of a sport of activity benefit from a tighter tolerance variation as they are seeking out small margins of improvement on already very efficient motions.

Eddie O’Dea has been in the bike fitting game for a long time, and I trust him when he tells me that the 3DWearable system has changed the way he approaches bike fits. Eddie first fit me and my team in 2006 alongside 3DWearable co-founder Tom Coleman. And, Eddie was my go to bike fitter years later when I lived in Atlanta. Eddie and Tom are driven by the desire to get more people moving more efficiently and with less injury. They invented the 3DWearable system to both extend the value of the bike fit out on to the road in real-time training and to equip the end user with tools and knowledge to refine their own motion.

Increased Variables

Eddie went on to explain that the more sensors that are introduced to the system, the complexity of the number of motions and the relationships among them grows exponentially. Eddie foresees being able to map baseline motions for additional variables experienced outside on open roads. This means that the sensors on the body are not only tracking their relative position (i.e., pitch, yaw, roll) for hip angle and cadence (as examples), but also doing so in relationship to the speed of the bike and the angle of the bike on the road’s grade.

Eddie can imagine having an aero position motion mapped for when the bike is up to speed and the rider benefits from being in the aero bars, as well as having a climbing motion mapped when the bike is at a lower speed going uphill and the rider benefits from sitting upright in the bull horns. The 3DWearable system would know based on speed and grade which set of motions was ideal for each scenario and would cue the athlete to adjust to the ideal baseline motion for each scenario.

The possibilities are endless, and 3DWearable has identified several user groups that would benefit from their system, including golf; cycling; walking/running/hiking; strength training; target sports like archery and shooting; racket sports; and skiing.

Early Adopters

3DWearable is offering limited quantities of the system to early adopters at a reduced price. The 3DWearable team will be in close communication with the early adopters to coach them through how to apply the sensors and use the system to map and track motion. As the company scales the system and number of users, they intend on offering more instructional video content through their application.

The sensors have an 8-hour battery life and are rechargeable. The application is available for both Apple iOS and Android devices.

The introductory reduced price is $799 for two sensors, or $399 for a single sensor, and the app with no subscription fee for cloud storage for the first year. If you are interested, visit them at 3DWearable.org or e-mail them at info@3DWearable.org.

Eddie O’Dea will also answer your questions in the forum.

2

How does this differ from the Leomo system? Seems pretty similar.

3

I appreciate that it took effort to identify and describe this product, whose objective - pose tracking in the wild - is one of considerable interest to many triathletes. I mean this constructively, that I’m surprised to see hardware-based (i.e. joint-sensor) ‘new’ technology for tracking human body pose for general AG athletes. In fact, after the aero article and mention of the difficulty in getting the right static image, I almost suggested just getting some video - it doesn’t even need to be perfect side view, especially if you have rigid parts to reference (like the bike!).

There are a variety of single-view video pose estimators already available, and there are already human skeleton 3D models (which you technically don’t need, but can offer faster generalization to new, untrained videos). Many of these come from the research community and so are freely available.

There are some aspects of kinematics that direct sensors will be superior for, and it’s still more straightforward to have direct IMUs /position transmitters on board and recording, for use cases of changing posture that require continuous monitoring. But for a triathlete’s movements, we’re mostly about pose so that we can check bike fit. And having videos set up during rides (possibly even on-board) could get a lot - not all - but a lot of what’s being described here, and without fiddly hardware=more futureproof.

I do understand the benefit of having a realtime feedback system continuously monitoring, which would apply to video in an indoor setting only. So I’m not saying there’s no use case, only that the technology is somewhat self-limiting already. The good news is, if they develop strong software for the feedback and UI, they could grow into video based and other modalities as inputs on the back end. I look forward to seeing where this goes.

I should probably say, I have nothing to disclose - I do not own any part of any company, IP, etc. related to this. I have used IMUS and video for pose estimation and keep somewhat up with the field, is all.

One specific request: please for the love of all that is sacred, I’m begging you, drop the “neuro re-education” mumbo jumbo. It’s training. You’re training posture, pose, and coordination. There are no CME credits for neurons. Thank you. And good luck!

4

Thank you for the question. Leomo paved the way, but like the BlackBerry smart phone they got a little stuck in the weeds. We sought to overcome some of the issues we recognized:

Smaller sensors: if the form factor is too big the muscles near the landmark you are trying to track moves the sensor and skews the data. Garbage in is garbage out.

Ferromagnetic interference: our sensors are immune to the electromagnetic interference that comes from things like smart trainers, steel bikes, or mountains, full of iron. Having worked with many sensors over the years we learned the importance eliminating this interference. Again, garbage in is garbage out.

Simplicity of use: Our goal with 3DWearable is to simplify the user interface. We don’t want you to have to have a PhD to understand the feedback you’re getting. You shouldn’t have to read a book to understand the device you just purchased and want to use to improve yourself. I remember getting my first power meter and being so excited about it until I realized I had no idea how to interpret it. A month or so later I was on my way. We are trying to be easier to use, but we can also be better in this area. It’s our ethos with every iteration and functional change.

5

Thank you for the feedback. It sure did take some effort. We came into this as a bike fitters, coaches, and athletes and that formed our take on why this tech was needed.

  1. Video is useful, but not enough. Video cannot track the minutia happening on the inside. Even the best set up with a video only system cannot track the small effects that say 2 mm in saddle height can change. 3DWearable can do that and I have been doing that for the last year or so with our system. You can do this at home with our system. If your saddle is too high, you will rock. If your saddle is too low you will rock. 3DWearable track this to one thousandth of a degree in real time and guide you to your smoothest saddle height. You can even set the saddle height based on a particular effort level. The harder you are going the tighter your muscles will get. There’s a small but important saddle height difference between someone trying to maximize their set up for a sprint versus a full Iron Man type effort. Video cannot measure this.
  2. Video does not work well in the real world. Small changes in the throw or the angle in relationship to the subject can skew the angles one is viewing. I’m not saying video isn’t a useful tool, but optical illusions are just that…illusions. 3DWearable is a like a power meter vs a heart rate monitor. Both are useful, but heart rate is susceptible to interference that power meter is not. We agree, combining both together is useful, but video is limited in the real world and on the budget most of us are training with.
  3. I think I’ve already addressed the “self limiting” part of our tech. Video is very limited. Having a tool that can track your movements and go with you everywhere is the challenge 3DWearable is trying to overcome. We have bettered what video can do already and we see a path to even more improvement with both our product and your athletic endeavors in the not so distant future.
  4. Having the feedback in real time all of the time is a game changer. Changing the way we move is a huge challenge. Try learning to run backwards or writing with your opposing hand. Habits are incredibly useful, but also hard to change. That’s what coaching is all about. Most AG athletes don’t have a coach with them on every training session. 3DWearable can be there every time for every session giving the feedback for 8 hrs at a time. And if you train longer than that then a second set would be useful.
  5. Our apologizes for the big words. We have been bridging between the complicated world of understanding human bodies and how they move through the world (kinematics) and explaining it on many levels for so long that eventually the big words slip out. It’s a long and complicated path to learning new motor skills and changing habits. Neuro reeducation is not just mumbo jumbo to us. It’s what we do every day. It’s what every athlete does every second they are training and in some cases even when they are not “training.” Rather than use the term less we want to see it used every day. It’s a term that should be in every athletes’ and coaches’ daily lives.
6

Who determines the ‘ideal’ position and what proof exists that this really is ideal?

7

[quote=“eddie3D, post:5, topic:1283704, full:true”]
Thank you for the feedback. It sure did take some effort. We came into this as a bike fitters, coaches, and athletes and that formed our take on why this tech was needed.
-That was what I figured. So, I’m hoping to keep this constructive. I think your heart is in the right place.

1. Video is useful, but not enough. Video cannot track the minutia happening on the inside. Even the best set up with a video only system cannot track the small effects that say 2 mm in saddle height can change. …If your saddle is too high, you will rock. If your saddle is too low you will rock… Video cannot measure this.

  • video can’t get what’s inside, true, nor can the IMUs, but both can get the rocking/pose adjustments. That’s what I’m saying. Image analysis has actually come that far. It can track tiny changes in eye position in a moving head to track gaze point of regard. It can track subtle changes in posture and movement economy with learning.

3. Video does not work well in the real world. Small changes in the throw or the angle in relationship to the subject can skew the angles one is viewing. I’m not saying video isn’t a useful tool, but optical illusions are just that…illusions.

  • I’d hesitate to make such a blanket statement. In fact what’s getting so good with modern ML/AI is that you can account for those variances: you don’t need a particular angle to get accurate modeling. Especially in case of rigid structures like you have with people cycling, there are reference points to allow 3-D models to fit different 2-D views. That said, point 6 is where I really think you have something special.

6. I think I’ve already addressed the “self limiting” part of our tech. … Having a tool that can track your movements and go with you everywhere is the challenge 3DWearable is trying to overcome. We have bettered what video can do already and we see a path to even more improvement with both our product and your athletic endeavors in the not so distant future.
-Although the wearable solution is arguably klunky, we don’t have video monitoring everywhere outdoors yet, AFAIK! I can very much see how your wearables could be beneficial outside, for continuous monitoring. And I would completely buy into the argument that these real-world positions don’t look like the indoors positions, and may only happen in specific situations that your continuous monitoring can capture (e.g. after 3 hours outside on the bike, or after getting out of the saddle on a climb, etc.) I know my position changes outdoors unlike indoors, but I don’t know e.g. if it’s less aero or powerful, or what I would need to do to my bike fit to better accommodate the posture my body is trying to adopt. Love the tech for this! :raised_hands:

8. Having the feedback in real time all of the time is a game changer. Changing the way we move is a huge challenge. Try learning to run backwards or writing with your opposing hand. Habits are incredibly useful, but also hard to change. That’s what coaching is all about. Most AG athletes don’t have a coach with them on every training session.
AGREED!

10. Our apologizes for the big words. We have been bridging between the complicated world of understanding human bodies and how they move through the world (kinematics) and explaining it on many levels for so long that eventually the big words slip out. It’s a long and complicated path to learning new motor skills and changing habits. Neuro reeducation is not just mumbo jumbo to us. It’s what we do every day. It’s what every athlete does every second they are training and in some cases even when they are not “training.” Rather than use the term less we want to see it used every day. It’s a term that should be in every athletes’ and coaches’ daily lives.

  • Hold my water bottle. It wasn’t the “big words”, big guy. It was the euphemism, and the pretense that you were invoking a neuroscience term. In my opinion, you don’t need to do this and if you care about your brand image appearing legitimate, you’d do better to avoid it. I recognize that some fringe PT and clinical rehab speak does tread along this path, and I don’t care for that either. But even then, it usually at least places the “education” on the individual, their learning, or some “system” and not on neurons. Neurons don’t go to school, and accordingly, don’t get education; yet no learning occurs without neuronal modification, making the term vacuous. It doesn’t add anything to the process of (re)training to invoke it. The processes we’re talking about here include motor planning, motor adaptation and proprioceptive awareness. This includes consolidation effects after learning. I’m guessing we’d agree that motor learning is a term that should be in every coaches or athletes life, daily or otherwise. Adding ‘neuro reeducation’ -when you referring to learning how to hold a position, adds nothing to the process you refer to, but for me at least, it does undermine the legitimacy of what you’re pitching. Look, I’ll even write you a blurb you can use on the importance of real-time feedback on changing motor habits, b/c that’s so clear from the research, if you decide to ditch the poser ‘neuro-terminology’. Just one point of view, for your consideration.
2 Likes
8

Brave of you to open yourself up to public scrutiny on a forum like this. A couple questions if you don’t mind:

What IMU manufacturer are you using for the device and which validation studies would you point to representing its tracking accuracy?

What parameters are you using to calculate joint centre locations based on boney landmarks? I’m assuming it’s De lava but those parameters are struggling with single digit count IMU systems rather than full boney landmark motion capture.

How involved with practitioners are you planning on being to make sure they are able properly palpate/place a sensor on the correct landmarks?

What angles are you reporting to practitioners? Are they 3D? 2D Sagittal? Sagittal with a transverse offset?

As others have mentioned, where do your guidelines for fits come from?

Cost? Per unit? Subscription based? Software costs?

1 Like
9

I’ll focus on the sacrum to keep this from being a dissertation. There two main points of interest at this location:

  1. The goal is to have the rider working from the middle of their range of motion. When the hip (most any joint tbf) is at the middle of ROM you get the best muscle recruitment. You’ll get the most cross bridging of the actin and myosin and thus get more muscle cells firing. You can google this and find endless articles about it. You also get the least stress on the connective tissues (fascia, tendons, ligaments). You can feel this on a bike or in a chair. Roll your hip forward and the hamstrings get more stretch. Too much stretch and you get micro tears in the connective tissue. Repeat that tear 5000-6000 times an hour and you’ll eventually have a bigger issue.
  2. Less rocking at the sacrum means a more stable hip. We can measure this to the .0001 degree. You can then raise or lower the saddle a few millimeters. If the rocking increased that increases instability. If the rocking decreases, then we have made the hip more stable. Repeat this process in the direction of less rocking until it then begins to increase. Go back to the previous saddle height and there you have the best saddle height for that person, on that bike, with those components and clothing combo. A more stable hip means less wasted motion and more comfort on the saddle. I noted the components and clothing: Let’s take tri shorts vs cycling shorts. Theres usually a few millimeters difference. If you dialed in the saddle height for the tri shorts, you’ll likely be rocking slightly more with cycling shorts. 3Dwearable can measure that difference.

I hope this helps.