The growing adoption of wearable smart glasses and AR headsets is enabling a new generation of on-device AI systems that interact naturally with users and their environments. Eye tracking lies at the heart of this evolution, providing real-time measurements of gaze direction, pupil dynamics, and visual attention, critical for truly immersive and responsive experiences. Despite significant advances in gaze estimation algorithms, hardware and embedded systems remain a key bottleneck, preventing efficient always-on deployment on resource-constrained devices.
Recent progress in low-power and high-speed sensing, hardware acceleration, and embedded machine learning offers concrete opportunities to overcome these challenges. These include miniaturized optical modules, capacitive and event-based sensors, FPGA and NPU accelerators, and TinyML frameworks for on-device eye tracking, along with adaptive sensing and processing techniques. The SPHERA workshop will focus on these developments, bringing together researchers and practitioners at the intersection of hardware design, embedded intelligence, and eye tracking. It will explore how innovations in sensors, computing architectures, and resource-aware adaptive processing can enable energy-efficient, low-latency, eye-tracking systems that meet task demands in real-world applications.
Organizers: Marco Cannici, Simone Mentasti, Giulio Marano, Filippo Melloni, Luca Merigo, Marco Carminati, Matteo Matteucci