Motion capture technology
Sméagol ‘Gollum’ (Lord of the Rings), Neytiri (Avatar), Koba & Caesar (Dawn of the Planet of the Apes & Rise of the Planet of the Apes), Joel and Ellie (The last of us videogame) are some of the most memorable performances in cinema and video game. And what made those performances possible was motion capture, more precisely performance capture.
Remember, motion capture is a technique for recording the positions and rotations of an object or living being, to control a virtual counterpart on a computer. And, at the border between the cinema and 3D animation, performance capture is an evolution of motion capture allowing for facial expressions, movement of hands and fingers or nervous tic to be captured.
But today, motion capture technology is becoming more important than ever in sports and health. It may seem out of place to compare a man and a machine, but for example, you want the athlete (the asset) to perform to the max. You do not want him to get injured or exhausetd and in case he will, you want him or her to quickly recover. That is where the Mocap technology comes in to improve individual performance, tactics, strategy, and team composition but also figure out the risk of injury and track rehabilitation.
Today there are four main types of technologies for motion capture: optical, mechanical, magnetic, and gyroscopic.
For the optical track, there are several methods, the most popular being the one based on infrared camera with reflective passive markers. In it, markers are placed on the actor at characteristic places on the body, usually near the joints, and black and white images of these points are captured (not the images of the actor's body). For more details check Salma’s page.
Non-optical systems such as inertial systems use miniature inertial sensors, biomechanical models, and sensor fusion algorithms to track the movement. Other methods are mechanical motion - using exoskeleton motion capture systems, and magnetic motion capture - using a relative magnetic flux of three orthogonal coils on both the transmitter and each receiver.
Finally, we have the stretch sensors that are flexible parallel plate capacitors that measure either stretch, bend, shear, or pressure and are typically produced from silicone.
But what about using our clothes directly? What about using e-textile for performance capture?
We are in contact with textiles for up to 98% of our lives, and they are starting to become intelligent. Part of this revolution includes the integration of electronics and textiles. Stretching, bending, breathing and tearing apart our clothes live with us and since a few years ago they started to feel every single movement of our body, muscle dilatation and heartbeat.
Smart textile technology
A health monitoring pioneer project in E-Textile is WEALTHY1, a portable health system, the paper of which was published in 2005. It is a garment provided with several sensors of physiological signals such as for example an electrocardiogram or an electromyogram. These physiological signal sensors are textile stress sensors based on piezoresistive threads and electrodes made with metallic threads.
But textile-based performance capture is not yet quite common.
So, this project is our V1 attempt in understanding motion capture trough e-textiles but also contribute to the accessibility of embroidered textile sensors.
Aiming for a full-dress suit we started with a sleeve.
We choose to work with materials and tools that are accessible to most of the students at the DVIC. So the process of making the sleeve can be easy to follow and understandable to any coming learner.
For this project we used 3 different sensors:
Strain sensor 2
This sensor consists of an uninsulated conductive thread that is sewn in a zigzag stitch to a stretchable fabric like this:
As the thread is not insulated and the stitches are touching each other, a “short-circuit” is created. Stretching the fabric increases the electrical resistance of the circuit. This is due to the opening of the mesh and thus the breaking of the parallel contact points, forcing the current to flow in series rather than parallel. This increase in the conductive path results in greater resistance as shown in the figure:
Figure 1. Conductive path knitted fabric model (a) in the relaxed and (b) stretched position
The resistance can give us information about the percentage of stretching of the fabric (how long is the stretched fabric). We placed this sensor on the elbow so we can get the angle of flexion of the sleeve, but we can also place it on the chest to check the breathing rate.
The conductive thread is sewn over an entire surface, like this:
When the fabric undergoes creasing, this means that it will be folded on itself, creating contact points on the surface, which will allow current to short-circuit the pattern. This reduction in the conductive path will therefore result in less resistance.
This sensor can be used, for example, to record the catches that its wearer undergoes during the fights.
We placed this sensor on the shoulder to check if the arm is raised or not.
We combined these 3 sensors to get maximum information about the wearer's arm.
Those sensors combined and applied on a complete suite can be used to export the movements of a real subject to a virtual 3D model or simply just get specific data needed for a better scoring system in different sports.
Results and thoughts
The first thing that we didn’t encounter during the process is the effect of humidity and temperature on the sensor. As the sensors are using uninsulated threads, the humidity in the air, as well as the temperature, can have a noticeable impact on the lifespan of the sensor as well as his performance. We suppose that “encapsulating” the sensors in a 3d textile material may be a solution. Used in a relatively dry environment the uninsulated thread has a lifespan of several months before oxidation, but we still need to test the impact of several washing cycles on its lifespan.
Apart from the simple motion sensor, we could make the garment react to its own movements, make it recognize patterns and then react accordingly. Integrate AI to the clothing.
If in this version we used sewing and embroidery, we are currently working on a process using only embroidery and snaps. This will allow us to take a step towards a washable element. Also, we would like those sensors to create a wireless network so we are working on flexible PCB to harvest and send the data.
- Study of other sensors, in particular the pressure sensor and tactile sensor (using MuCa module)
- Linking data to make a 3D model of it
- Paradiso, R.; Loriga, G.; Taccini, N.; Gemignani, A.; Ghelarducci, B. WEALTHY-a wearable healthcare system:New frontier on e-textile.J. Telecommun. Inf. Technol.2005, 105–113.
- - Tangsirinaruenart, O.; Stylios, G. - A Novel Textile Stitch-Based Strain Sensor for Wearable End Users DOI : 10.3390/ma12091469