Public thesis defense of Arthur Sluÿters

Mid-air gesture recognition by ultra-wide band radar echoes for manipulating multimedia objects towards cross-context transfer

This thesis focuses on the recognition of 3D gestures (gestures performed with some part(s) of the body) captured with radar-based sensors, with the end-goal of providing better and more intuitive ways to interact with systems such as TVs and computers in challenging environments (for instance, bad lighting conditions). 3D gesture recognition has the potential to impact many facets of our lives, with potential applications such as looking-up patient data in the operating room without any physical contact, thus limiting the risks of contamination, or flipping through the pages of a recipe e-book while cooking without having to talk or touch a screen. However, there are still many challenges left before the generalization of 3D (radar) gesture recognition, which I try to address in my thesis. These challenges include identifying gestures among parasitic movements, dealing with sensor noise, accurately recognizing gestures in as little time as possible, and finding the most appropriate gestures for specific actions.