Researchers at MIT have created a new tactile sensing carpet, able to estimate human poses without the need for cameras. Researchers about this project said it was a step towards improving personal health services, smart homes, and personalized games. The researchers noted that most daily activities people involve physical contact with the soil, whether it runs, exercise, or rest.
These interactions have information that can help understand the movement of people, and previous research into the subject has used a single RGB camera, an omnidirectional camera that can be used, and standard off-the-shelf webcam. Challenges with systems that use cameras are blocked cameras or privacy issues. MIT researchers only need cameras to make data set by their system trained and only capture the movements of people who do activities.
After training, the system can identify someone’s 3D poses by simply requiring them to ride the carpet and take action. Neural networks in teams made using tactile information to determine whether the person does sit-up, stretch, or do other actions.
Carpets at low costs and can be scaled and consist of sensitive films against commercial pressure and conductive yarn. Carpets measuring 36 times two feet and have more than 9000 sensors in it. Each sensor changes the pressure from humans into an electrical signal through physical contact between legs, limbs, bodies, and carpets.
The application for the system includes the ability for a computerized exercise regime like showing someone’s video that does push-up if the user rises the carpet and starts doing push-ups. This can help people train yourself to understand the right shapes and techniques to prevent injury during exercise. The researchers say if used solely to exercise, the carpet can count the number of reps and calculate the number of calories burned.