François-Maël ROBERT Thesis defense

The eye is an organ that can be used to measure various biological and cognitive parameters. In particular, the study of eye movements and gaze position, or oculometry, has various applications in fields such as the design of humanmachine interfaces, visual restoration and rehabilitation therapies or the measurement of cognitive load. Most commonly, an oculometer consists of a device comprising a camera filming the eye, combined with a pupil detection algorithm, the position of which is linked to a gaze position in the environment. This technique requires image processing that can be complex and prone to poor pupil detection. This thesis proposes a new type of video oculometer, based on a scleral contact lens encapsulating two VCSELs, lasers emitting in the near infrared (850 nm). These lasers are powered inductively by a coil embedded in a mount. The mount also carries the camera filming the lens on the eye and a second camera filming the environment. The position of the infrared spots on the lens is associated with a gaze position by calibration. This approach has been tested and validated experimentally, on bench and on human subjects, on the Streetlab platform at the Institut de la Vision. The device has been certified and CE marked. The use of an optical position sensor, a PSD (Position Sensing Detector), to replace the camera filming the eye and gain in acquisition frequency, was also tested. The materialization of the gaze can also be achieved by direct imaging of infrared spots projected into the environment. A diffractive optics element was therefore incorporated into the lens, to collimate or shape the beam from the VCSELs in the form of a cross, and thus designate a target without disturbing the human version. The viewed position is seen by a camera filming in the infrared.