Executive Summary : | Gaze tracking has the potential to empower persons with special needs or subjects living with motor neuron disorders to enable them control devices and effectively communicate with others. However, estimating a person’s gaze is difficult task due to the unconstrained variables such as head pose, head position, eye rotation, working distance, variable illuminations, background noise, sometimes presence of the eye wear, etc. Commercially available gaze trackers use specialized sensors and components cost upto millions of dollars. On the other hand, software-based eye trackers are inaccurate and often prone to lighting interferences. This challenge inspired investigators to design and develop a hardware-independent gaze tracker that can generate cueing commands to assist users without using any of the limbs. There are many difficulties that must be overcome before eye gaze-based cueing in interactive applications can be widely used. For instance, there is a need for reliable and precise eye tracking algorithms that can function in various lighting scenarios and with people who have varying eye shapes and sizes. Additionally, successful communication utilizing eye gaze-based indicators requires efficient and simple user interfaces. These issues will be addressed in the project by creating sophisticated algorithms based upon AI & ML for eye gaze-based cueing, creating user-friendly interfaces, and putting the system to the test in real-world situations. |