Modelization and Implementation of a Motion Control System Based on the Wrist: A Smart-Watch Function

As smartwatches and bracelets have become the norm in our daily lives, motion sensors are constantly attached to our wrists. Movement tracking paired with a variety of algorithms can enable recognizing gestures and different types of movements. One of the possible scenarios is turning the screen of the smartwatch on and off for a smooth and immersive user experience. Machine learning algorithms provide a simple, reliable, and efficient method to achieve that. This paper goes through the process of wrist movement recognition from data acquisition to classification with multiple Machine Learning (ML) algorithms, using inertial measurement unit sensors. This approach aims to enable the model’s adaptability and Edge AI, allowing the model to operate quickly, responsively, and efficiently on the watch’s computing resources.