The goal of this project is to conceive and develop a wearable system based on kinesiologic electromyography (EMG) that recognizes the user activity in real time. In particular, the system recognizes the following five activities: “walking”, “running”, “cycling”, “sitting” and “standing”. We conducted a study in order to select the opportune muscles and sensors’ placement. The muscles of interest were four: Gastrocnemius, Tibialis Anterior, Vastus Lateralis and Erector Spinae. Among the results, we can highlight that using the signals sensed from three opportunely selected muscles (Gastrocnemius, Tibialis Anterior and Vastus Lateralis) instead of four did not entail a sensible loss of accuracy, but reduced the computational cost by 24.1%. In particular, by sensing four and three muscles we achieved activity recognition accuracy higher than 96%, on the contrary to 99%, attained when the system was trained on a single subject.
Activity recognition, physiological signals, EMG, HCI, machine learning.