Eye-tracking calibration
View Patent ↗Aspects of the present disclosure describe an improved calibration method for systems, methods, and structures that provide eye-tracking by 1) steering a beam of light through the effect of a microelectromechanical system (MEMS) onto a surface of the eye and 2) detecting light reflected from features of the eye including corneal surface, pupil, iris—among others. Positional/geometric/feature/structural information pertaining to the eye is determined from timing information associated with the reflected light.
1. An eye-tracker calibration method comprising:
arranging a system and a user's head in a first arrangement, wherein the system includes a head tracker and the eye-tracker;
having the user look at a fixed point in space;
having the user rotating his/her head about an axis;
collecting head positional and eye-tracking data including eye angle, wherein the eye angle is determined by:
(i) scanning a light signal in a two-dimensional pattern over a scan region of the eye;
(ii) detecting reflected light from the scan region at a first discrete detector, wherein the first discrete detector provides a first electrical signal based on the reflected light, and wherein the first electrical signal includes no spatially correlated image information;
(iii) detecting a first extremum in the first electrical signal;
(iv) estimating a first location of a surface feature in the scan region based on the first extremum; and
(v) estimating the eye angle based on the first location; and
generating a map of the user's eye angle to gaze vector.
2. The method of claim 1 wherein the surface feature is a cornea.
3. The method of claim 1 further comprising:
detecting reflected light from the scan region at a plurality of discrete detectors that provides a plurality of electrical signals, wherein the plurality of discrete detectors includes the first discrete detector and the plurality of electrical signals includes the first electrical signal;
detecting a plurality of extrema that includes at least one extremum in each electrical signal of the plurality thereof, wherein the plurality of extrema includes the first extrema; and
estimating the first location based on the plurality of extrema.
4. The method of claim 3 wherein the fixed point in space is unknown.
5. The method of claim 1 wherein the first location is estimated based on timing information associated with the extremum.
6. The method of claim 1 further comprising:
constructing a profile for the user, wherein the profile includes at least one of:
a model of each eye of the user, the model including at least one of eye center, cornea offset, cornea radius, and corneal asphericity;
interpupillary distance between the user's eyes; and
eye relief;
determining a second arrangement of the system and the user's head; and
compensating for the difference in the first and second arrangements based on the profile.
7. The method of claim 1 wherein the two-dimensional pattern is a Lissajous pattern.
8. The method of claim 1 further comprising predicting a saccade endpoint by operations that include:
determining a plurality of locations of the surface feature in the scan region;
filtering the plurality of locations;
determining a velocity for the motion of the surface feature based on the filtered plurality of locations;
if the velocity is greater than a threshold value, determining a velocity profile for the motion; and
estimating the saccade endpoint based on the velocity profile.
9. The method of claim 8 wherein the saccade endpoint is determined by an operation selected from the group consisting of:
accessing a lookup table that associates peak velocity and saccade amplitude;
fitting the velocity profile to a parabolic function;
fitting the velocity profile to a cubic function; and
fitting the velocity profile to a function and integrating the area under the curve of the function.
10. An eye-tracker calibration method comprising:
operatively coupling a system relative and a user's head such that the system and user's head are in a first arrangement, wherein the system includes a head tracker and the eye-tracker;
having the user look at a fixed point in space;
collecting head positional and eye-tracking data during a vestibulo-ocular reflex (VOR), wherein the eye-tracking data includes eye angle data for a first eye of the user, wherein the eye angle is determined by:
(i) scanning a light signal in a two-dimensional pattern over a scan region of the eye;
(ii) detecting reflected light from the scan region at a first discrete detector, wherein the first discrete detector provides a first electrical signal based on the reflected light, and wherein the first electrical signal includes no spatially correlated image information;
(iii) detecting a first extremum in the first electrical signal;
(iv) estimating a first location of a surface feature in the scan region based on the first extremum; and
(v) estimating the eye angle based on the first location;
generating a plurality of data pairs, each data pair including a head position and a corresponding eye angle; and
generating a map of the user's eye angle to gaze vector based on the plurality of data pairs.
11. The method of claim 10 further comprising:
detecting reflected light from the scan region at a plurality of discrete detectors that provides a plurality of electrical signals, wherein the plurality of discrete detectors includes the first discrete detector and the plurality of electrical signals includes the first electrical signal;
detecting a plurality of extrema that includes at least one extremum in each electrical signal of the plurality thereof, wherein the plurality of extrema includes the first extrema; and
estimating the first location based on the plurality of extrema.
12. The method of claim 11 wherein the fixed point in space is unknown.
13. The method of claim 10 wherein the first location is estimated based on timing information associated with the extremum.
14. The method of claim 10 further comprising constructing a profile for the user, wherein the profile includes at least one of:
a model of each eye of the user, the model including at least one of eye center, cornea offset, cornea radius, and corneal asphericity;
interpupillary distance between the user's eyes; and
eye relief.
15. The method of claim 14 further comprising:
detecting a second arrangement of the system and the user's head; and
compensating for the difference between the first and second arrangements based on the profile of the user.
16. The method of claim 10 further comprising predicting a saccade end-point by operations that include:
detecting the occurrence of a saccade based on motion of the surface feature; and
during the duration of the saccade, determining an endpoint for the saccade.
17. The method of claim 16 wherein the saccade endpoint is determined by an operation selected from the group consisting of:
determining a velocity of the surface feature and accessing a lookup table that associates peak velocity and saccade amplitude;
estimating a velocity profile for a motion of the surface feature and fitting the velocity profile to a parabolic function;
estimating a velocity profile for a motion of the surface feature and fitting the velocity profile to a cubic function; and
estimating a velocity profile for a motion of the surface feature, fitting the velocity profile to a function, and integrating the area under the curve of the function.
18. The method of claim 16 wherein the occurrence of the saccade is detected by:
determining a velocity for the motion of the surface feature based on a filtered plurality of locations of the surface feature; and
comparing the velocity to a threshold value.