Eye Tracking Technology

What is eye tracking?

Eye tracking is the process of tracking the movements of the human eye or referencing the absolute point of gaze (POG), which is the specific spot where the individual is able to focus. In order for an individual to comprehend the information/object that they are looking at, their gaze must remain long enough (approximately 200-600 microseconds) for the brain to process it, meaning that eye tracking can help categorize gaze into levels of attention (Majaranta & Bulling, 2014).

What is eye tracking technology?

Eye tracking technology is the measurement of various characteristics of the eye in order to determine the gaze of the person being assessed. This is done through a combination of a variety of instrumentation in order to reduce the guesswork of manual tracking. The tracking technology serves as a translation between the physical movements of the eye into a data set that can be evaluated by the administrator.

A Brief History of Eye Tracking

Eye tracking technologies originated with the discovery that human beings require a variety of eye movements in order to comprehend what they are seeing and accomplish tasks such as reading across a page. Eye movements were then categorized into four different types (Purves et al., 2001):

  1. Saccades: Quick, abrupt changes in the point of focus

  2. Smooth Pursuit: Slower, continued changes to follow a moving focus

  3. Vergence: Alignment of eyes with targets at various distances

  4. Vestibulo-ocular: Stabilize the eyes relative to the world, allowing for head movement

Eye movement was initially studied by simple observation, followed by a device that required an uncomfortable physical, and invasive, device to be placed directly on the eye.

A photography-based device for eye tracking was developed first in 1901 by Dodge and Cline. This approach paved the way for advancements in eye tracking technology as camera technologies improved, combined with increasingly complex computing processes to automate categorizing eye movements.

There are three predominantly used eye-tracking methods (Majaranta & Bulling, 2014) that are widely used in both commercial and research settings.

Videooculography

This method uses a camera and computer combination to record and analyze gaze data. It records the location of the eye and the center of the pupil to approximate the wearer's gaze. It requires a mounted headset to be utilized.

Video-Based Infrared Pupil-Corneal Reflection

This method uses infrared (IR) light to illuminate the pupil from off of the eye. This allows for more accurate tracking of the pupil, as well as measuring the IR reflection off of the cornea. These measurements are combined and extrapolated to make a more specific calculation of the wearer's gaze.

Electrooculography

This method relies on the fact that the human eye can be modelled as a dipole (2-ended pole) with a positive end located at the cornea and a negative end at the retina, creating a small and stable electrical field. Electrodes are placed around the users face and can be used to measure changes in these electrical fields to map eye movement/gaze.

Eye-Tracking as a Diagnostic Tool

Eye-tracking has evolved to be used as a screening tool for the early detection of conditions such as dyslexia, as seen in this video below.

Limitations of Current Eye Tracking Technology

The different methods of eye-tracking technologies each have their respective benefits and limitations, but some major limitations of current eye-tracking methods can be characterized as (Majaranta & Bulling, 2014):

  • Cost & customized equipment: Much of the camera and computer equipment is very powerful, meaning that it is also costly, outside of the means of many consumers. The setup necessitates technical expertise and must be fitted properly to each test subject, meaning that it can be complicated to complete outside of a laboratory testing scenario with additional funding.

  • Rigid function conditions: Much of the eye-tracking technology requires very specific conditions for optimized tracking, such as the person's head remaining perfectly still and in the same posture throughout the test. Additionally, consistent lighting is an important factor, making ambient lighting a challenge and real-world situations challenging.

  • The Midas Touch problem: Akin to the "everything you touch turns to gold" problem, the eye-tracking software needs to be able to attribute the difference between casually looking at a specific element and when there is intentional control of an element. Without a cue/trigger, all elements gazed upon will be selected as important.

  • Personalized calibration: Before any eye-tracking can be completed, the individual must complete calibration of the equipment to ensure it is functional. This may depend on the type of eye-tracking, the size/brightness of the screen, or the angle of viewing. Head-mounted hardware may also shift, altering the functionality. This process is time-consuming and necessitates a stable tracking environment.

  • Gaze vs. interaction: The nature of the human sensory system means that there is a delay between looking at something and choosing to interact with it due to the fact that gaze often precedes action. Eye-tracking for a purpose of determining what people interact with requires a second layer of interpretation or a cue to be utilized to confirm the individual's intention. Gaze is also easily distracted, meaning people often make unconscious and automatic eye movements to supplement their visual stimuli.

  • The nature of the eye: The eye is never completely still. It is constantly undergoing micro-saccades to stabilize and maintain focus. While eye-tracking technology is very precise, the constant movement means that individual pixels cannot be recognized with 100% accuracy.

  • Storage of recording data: With video-based eye-tracking software, a large amount of video storage is required for analysis to take place. Video files often take up a significant amount of storage, necessitating additional memory outside of the recording device.