MIT researchers have created a new tactile detection mat, capable of estimating human poses without the need for a camera. The researchers of the project say it is a step towards improving personalized health, smart homes and custom personalized games. Researchers note that most of the daily activities of people involve physical contact with the soil, whether walking, exercise or rest.
These interactions have information that can help understand people’s movements and previous research on the subject have used simple RGB cameras, portable omnidirectional cameras and off-towering webcams. Challenges with systems using cameras are blocked cameras or confidentiality problems. MIT researchers only need cameras to create the dataset that their system has been formed and capturing only the movement of the person performing the activity.
After training, the system could identify a person’s 3D installation by necessitating them only to turn on the carpet and proceed with the action. The deep neuronal network The created team uses tactile information to determine if the person makes sit-ups, stretching or another action.
The carpet is low and scalable and consists of a pressure sensitive commercial film and the conductive wire. The carpet measured 36 to two feet and has more than 9,000 sensors inside. Each sensor converts the pressure of the man into an electrical signal by physical contact between the feet, the members, the torso and the carpet.
Applications for the system include the ability of computerized exercise schemes such as showing a video of someone making pumps if the user is on the carpet and begins to make pumps. This could help people to train alone to understand a suitable form and technique to prevent injury during exercise. Researchers say whether they are used only for exercise, the carpet could count the number of representatives and calculate the number of calories burned.