A new intelligent sensing carpet from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) can track human poses without using cameras, which could spur breakthrough innovation in healthcare and gaming. 

• The research team spent just $100 to build a large-scale carpet system, using commercial, pressure-sensitive film and conductive thread, with over 9,000 sensors spanning 36-by-2 feet. Each of the sensors converts the human’s pressure into an electrical signal, through the physical contact between the feet, limbs, torso, and the device, according to MIT.  The system was specifically trained on synchronized tactile and visual data, recreating the user’s 3D pose.

• The carpet can predict which exercise the person is doing with an error margin by less than 10 centimeters. For now the device understands the lower body and torso movements better than the upper-body data and can’t predict gestures without direct floor contact, like free-floating legs during sit ups, or a twisted torso while standing up, MIT News reports. 
• The team is planning to train the model to track multiple movements, creating a 3D model of people dancing or doing martial arts. 
• The carpet, which could span the entire floor plan, can be also used in healthcare to help detect falls. “You can imagine leveraging this model to enable a seamless health-monitoring system for high-risk individuals, for fall detection, rehab monitoring, mobility, and more,” Yiyue Luo, a lead author on a paper about the carpet, told MIT News.