Prototype device for experiment (Tsukamoto-Terada Lab.)

 Load cell, Projector, openFrameWorks(C++), Visual Basic(C#)

A recognition system of activities conducted on a table with load cells only. 

The proposed system uses four load cells installed in the four corners of the desk. Since only the data on actions performed by touching the desk through the road cells, user privacy is not invaded. Load cells are installed to the bottom of the legs of the table, thus it is easy to instal our system to the general desk with four legs. There are no devices on the desk, so user can use the desk without regard to sensors. The proposed system obtains weight and position of an object placed on the desk, recognizes activities such as writing and typing, and identifies the user. Through the experiment using a desk I manufactured, I have confirmed that error of weight is 38g, error of position is 6.8cm, average recognition accuracy of four activities is 95.5%, and user identification rates for four users is 88.8%.