Towards a Billion Eyeballs with the First MEMS Sensor for Brain Health Wearables
ABSTRACT
It has long been said that the eyes are the windows into the soul. Decades of research in eye tracking and neuro-ophthalmology suggest that the eyes further serve as a “debug port” into the brain. A human eye makes 100,000s of saccades (rapid, ballistic movements) each day, and even when we fixate on a target, tiny eye movements (e.g. microsaccades and intersaccadic drifts) occur incessantly. The subtle dynamics of eye movements are tightly correlated to neurological health, as 6 of the 12 cranial nerves are directly connected to the eyes. Measurements of gaze and pupillometry also reveal our interests, desires, and state of mind. Furthermore, rapid and accurate eye tracking serves to optimize the visual quality and HCI (human computer interaction) experience in VR and AR products.
Unfortunately, the most widely used eye trackers to date are video-based, power-hungry and computationally demanding, and have therefore been relegated to laboratory environments in which short studies are typically performed across many people with widely varying baselines. MEMS technology has recently enabled an all-day, frictionless wearable that captures research-grade eye tracking data on a single user for extended periods to robustly establish one’s baseline states. The system offers continuous reporting of key measurements (e.g. cognitive load, attention span, fatigue, anxiety) and biomarkers (e.g. concussion, seizures, stroke, intoxication), which promise to vastly improve personalized monitoring of neurological and mental health.
The MEMS approach to eye tracking, developed by AdHawk Microsystems, offers unprecedented speed, low power, and ease-of-use in a simple architecture. A beam of light is swept across the eye thousands of times per second, reflecting from the cornea surface and iris onto detectors. The pulses produced by the detectors are used to calculate gaze, vergence and pupillometry with <3ms latency at 500Hz, with <1 degree of accuracy and <0.25 degrees of resolution.
The proliferation of all-day wearables for health tracking has resulted in vast quantities of data on individuals and the human population. Extended measurements of heart rate, for example, have revealed that HRV (heart-rate variability) is a more meaningful biomarker for cardiovascular health than traditional metrics that could be captured in a comparatively brief clinical visit. Lightweight spectacles with integrated MEMS-based eye tracking offer the ability to capture all-day biometrics for brain health for the first time, in a non-stigmatizing form factor that offers a frictionless experience.
Existing and emerging MEMS devices will play an important role in the growth of the wearables market in the years to come. With 266M smart wearables sold in 2020, this $22.8B market (in 2019) is projected to grow to $90B by 2026. The proposed talk will include a high-level overview of the full-stack eye tracking solution, a discussion of MEMS-related challenges, and selected examples of real-world pain points that will be addressed by wearables for brain health.
BIOGRAPHY
Neil Sarkar is the CEO and co-founder of AdHawk Microsystems, a start-up that has produced the first MEMS-based all-day wearable for brain health. He has over 15 years of experience as a microsystem design engineer with a focus on ultra-precise mechatronic systems that reside on CMOS chips. This technology is used at AdHawk to create the world's first camera-free eye tracking microsystem that achieves order-of-magnitude improvements across all key metrics (power consumption, bandwidth, size, latency, ease-of-integration) when compared to conventional eye trackers.
Before AdHawk, Neil and his team commercialized the world's first single-chip Atomic Force Microscope at ICSPI Corp, a spinoff from the University of Waterloo that is commoditizing nanometer scale metrology. Dr. Sarkar is a PhD fellow of the Waterloo Institute for Nanotechnology and the winner of the Douglas Colton Award for research excellence in microsystems.