Dual-Purkinje Eye Tracker

Fourward Technologies Manual:

• http://twiki.cis.rit.edu/bin/view/VPL/TempTest

Applied Optics Paper: Technical Description

• Crane_Steele_Generation-V_dual-Purkinje-image_eyetracker.pdf: 1985 Applied Optics paper describing Gen V tracker

Instructions

• From March 2, 2007 Warren visit:
  • etguide_3-20-2000.doc: The Idiot's Guide to Eyetracking
  • eyealign.doc: Basic Alignment Procedures
  • eyeopsandcontrol.doc: Basic Operation and Control

Automatic Data Logging:

We discussed using the voltage on one of the Aux lines to trigger data recording. This is possible, using the Software triggers and Trigger-Conditions, but it doesn't look like we can keep on displaying data while it is not being displayed, so instead, I'm going to log all the data to disk during a 'record' session, then do the data reduction before saving the text files. This also avoids any data loss due to noise on the image & status Aux lines.

We should also consider doing a D/A A/D auto-calibration periodically or even at the start of each run. We could run through a sweep of a range of voltages, or maybe even just a few, then recalculate the slope/intercept for the calculation to recover the image and status codes. Right now it is robust, but we should expect drift over time. This would also allow us to use more than 8 bits with some confidence...

-- JeffPelz - 08 Aug 2007

Calibration issues:

• compare_4_calibrations_July_30_07.doc: compare_4_calibrations_July_30_07.doc
• compare_4_calibrations_July_30_07.pdf: Repeated calibration on one observer:

Stage setup values

L/R U/D I/O Slide Post Observer - date

---

-7.55 +4.47 +2.18 19mm 82mm KY 7/26/07

-9.52 +4.03 +1.09 26mm 75mm JP 7/26/07

A/D USB-2527 vs. USB-1616FS systems

The USB-2527 is capable of high-speed collection, but is actually a single very A/D with a very high-speed mux. There is significant 'cross-talk' between the channels unless you allow significant settling time (~1ms) between samples. The output impedance of the DPI is evidently relatively high, so a voltage on one channel was affecting the other channels; for example, a large horizontal signal on Ch0 was influencing the vertical signal on Ch2.

The only thing MCC could suggest was longer settling times or intermediate buffers between the DPI and the USB-2527, so we are going to switch the 2527 for a USB-1616FS A/D. It is (on paper) a slower device, but it is a true 16 channel A/D with 16 independent A/D subsystems, so it has the added advantage that the sampling is truly synchronous between channels (the USB-2527 was actually offset by a few microseconds). Initial testing of the 1616FS (Sept '07) shows very high immunity between channels; 5V into channel 0 has no measurable influence on ch 1.

-- JeffPelz - 09 Sep 2007

A/D General Issues

It turns out that the 1408-FS runs at 48kHz when converting from analog to digital, but reading raw digital values is much slower - 8 msec to poll the DIO pins. This means going back to plan A: reading the Track-noBlink lines with analog lines instead of digital. This also means it's not practical to use the digital lines to pass trial info to the lab computer logging data. Options include:

1. Upgrade to a faster box. Measurement Computing has a USB-2500 Series A/D system. The USB-2523 has 16 single-ended/8 differential 16-bit analog inputs, 1 MHz throughput, 24 digital I/O, and four 32-bit counter input channels for ~ $720 (acad). The USB-2527 adds four 16-bit analog out channels and costs ~$900 (acad).

2. Use only the analog inputs on the 1408FS. It is easy to use the analog inputs for the Track-noBlink - it's just TTL (0/5VDC). Signaling trial info to the lab computer is less obvious, but could also be done with analog signals. Putting another 1408FS (or any D/A device) on the stimulus computer, you could set the output voltage at any time coded based on stimulus display, etc. Reading the analog voltage on the lab computer would allow trial information to be stored with the eye position data. The problem with this solution is that we no longer have enough ports available to read (H1,V1) and (ThetaH^?,ThetaV) for right and left trackers concurrently (which would greatly simplify calibration by eliminating the need to flip the switches all the time and read from a single meter).

The USB-1208FS (<$150 acad., and can be controlled from Psychophysics Toolbox) has two 12-bit analog outputs (0.000 - 4.095 VDC in 1mV steps; 2^12 = 4096). In single-ended mode, the USB-1408FS has 13-bit precision, but if we give 'headroom' for noise and signal loss between the systems and limit the read to 10 bits/channel (1024 steps, 4 mV/step), that gives 1024 x 1024 = 1M distinct codes using both channels. This gives up a 'button box' so if you wanted that you would have to use one analog channel for trial info and the other for a button box.

I think that you can run two USB-1408FS boxes at the same time, so if you wanted to, you could run one for each tracker, maintaining the ability to monitor (H1,V1) even in binocular mode.

Another option is to use the audio output of the stimulus computer to generate a signal. It could be a fixed frequency with varying volume, which would create a varying voltage at the input, or varying frequency which could be read. It would be simplest if the trial info were continuous during the trial so it could be read at any time, but it could be a short code given at the start of the trial too.

We could use one analog input channel (13-bit precision in single-ended mode) to read an analog signal from a 1208FS from the stimulus computer (with somewhere between 8 and 12 bits output), and use the second for a button box. We would have to do some testing to see how precise the signaling would be, but I suspect it would be OK at 12-bits, and it would certainly be fine at 10 bits (1024 levels; 4mv/step). A button box could be built that has a number of resistors set as voltage dividers with a regulated DC power supply. The output would be the sum of the voltages from the buttons pressed: For example, if there were 4 buttons, they could be tied to 0.25, 0.5, 1.0, and 2.0 VDC. A reading of 3.25VDC would indicate a chord of the 1st, 3rd, & 4th keys, etc.

-- JeffPelz - 26 Mar 2007

A/D Breakout Box Documentation

• A-D_breakout_box_documentation_jbp_6.pdf: Breakout box documentation

• A-D_breakout_box_documentation_jbp_6.doc: A-D_breakout_box_documentation_jbp_6.doc

• Configuring_the_USB-2527_with_InstaCal_5.pdf: Configuring_the_USB-2527_with_InstaCal_5.pdf
• Configuring_the_USB-2527_with_InstaCal_5.doc: Configuring_the_USB-2527_with_InstaCal_5.doc

-- JeffPelz - 16 Apr 2007

A/D Breakout Box Low-pass Filter

Passive low-pass (RC) filters are built in to the breakout box in channels 0-7 (R&L ThetaH^?&V, H1 & V1) and channels 10&11 (Aux A&B for 'XDAT'). The following document has information about the filter and sample data to show the effect of the filters. Note that the remaining channels are not filtered, so if you ever want to collect 'raw' data, just use channels [8,9,12,13,14,15]. Because the USB-1616FS is has 16 independent A/Ds, you could even collect both by just connecting copies of R&L ThetaH^?&V to ch 12-15.

• Raw_vs_filtered_5.doc: Raw_vs_filtered_5.doc
• Raw_vs_filtered_5.pdf: Raw_vs_filtered_5.pdf

-- JeffPelz - 09 Sep 2007 (updated with photo of interior of bo box)

• Raw_vs_filtered_4.pdf: Raw_vs_filtered_4.pdf
• Raw_vs_filtered_4.doc: Raw_vs_filtered_4.doc

-- JeffPelz - 24 Aug 2007

A/D Board Manuals

[ See http://twiki.cis.rit.edu/bin/view/VPL/InstructionManuals#MCC_USB_Analog_Digital_Converter ]

• Matlab_Data_Acquisition_Toolbox_2_Users_Guide_daqug.pdf: User's Guide
• Matlab_Data_Acquisition_Toolbox_2_Quick_Reference_Guide_daqug.pdf: Quick_Reference_Guide

Pseudo XDAT calibration:

• calibrate_XDAT_AD.pdf: calibrate_XDAT_AD.pdf

-- JeffPelz - 19 Apr 2007

MATLAB programs

• first_AD.m: A MATLAB program (cells) that reads 2 channels from the A/D, logs the data to disk, reads it back, and plots it.
• preview_and_extract_data_mcc.m: A MATLAB program (cells) that reads 1 channel of data from a mcc box, previews the data in real time, then extracts the data and displays it. Good example to use for monitoring EMS during collection with peekdata() while logging to disk.

-- JeffPelz - 27 Mar 2007

• DPI_binoc_75.m: DPI_binoc_75.m - Capture signals from R&L trackers, preview to display, and stream data to disk.
• DPI_setup_43.m: DPI_setup_43.m - Setup trackers by showing H1, V1, Theta H&V from both trackers in real time
• DPI_explore_data_arbitrary_70.m: DPI_explore_data_arbitrary_70.m - look at data file from binoc

-- JeffPelz - 12 Apr 2007

• DPI_setup_50.m: DPI_setup_50.m
• DPI_binoc_80.m: DPI_binoc_80.m
• DPI_explore_data_arbitrary_84.m: DPI_explore_data_arbitrary_84.m
• DPI_read_daq_3.m: DPI_read_daq_3.m

-- JeffPelz - 15 Apr 2007

• DPI_binoc_82.m: DPI_binoc_82.m
• DPI_setup_53.m: DPI_setup_53.m

-- JeffPelz - 16 Apr 2007

• calibrate.mat: calibrate.mat
• USB-1208_6_USB-2527_calibrate_43.m: USB-1208_6_USB-2527_calibrate_43.m
• DPI_binoc_83.m: DPI_binoc_83.m
• DPI_setup_54.m: DPI_setup_54.m

-- JeffPelz - 19 Apr 2007

• DPI_check_calibration_11.m: DPI_check_calibration_11.m
• DPI_binoc_96.m: DPI_binoc_96.m
• DPI_calibration_39.m: DPI_calibration_39.m
• DPI_setup_graphical_96.m: DPI_setup_graphical_96.m
• DPI_setup_71.m: DPI_setup_71.m
• DPI_explore_data_arbitrary_85.m: DPI_explore_data_arbitrary_85.m

(Adapted most for 1024x768 display) -- JeffPelz - 26 May 2007

• DPI_calibration_48.m: DPI_calibration_48.m
• DPI_setup_graphical_102.m: DPI_setup_graphical_102.m

Same center point for both programs now, and saves all calibration data.

-- JeffPelz - 10 Jul 2007

• DPI_25pt_calibration_109.m: Calibration routine with 25pt target
• DPI_25pt_calibration_122.m: Calibration routine with 25pt target - with visual check at end

-- JeffPelz - 31 Jul 2007

• DPI_binoc_251.m: Data collection/viewer with new USB-1616FS A/D
• DPI_get_HV_from_daq_file_20.m: Read in data from .daq file, prompt for left, right, or binoc, and file name
• DPI_test_routines_23.m: wrapper for several analysis routines

-- JeffPelz - 09 Sep 2007

Papers:

• Crane_Steele_Generation-V_dual-Purkinje-image_eyetracker.pdf: 1985 Applied Optics paper describing Gen V tracker
• deubel_bridgeman_fourth_purkinje_image_signals_reveal_eye-lens_deviations_and_retinal_image_distortions.pdf: Eye was tracked with Gen V DPI and scleral coils, showing that the eyelens moves inside the eye during saccades.
• deubel_bridgeman_perceptual_consequences_of_ocular_lens_overshoot_during_saccadic_eye_movements_1995.pdf: Companion paper describing perceptual shift in position of flashed targets near endpoint of saccades.

-- JeffPelz - 22 Mar 2007

Movies of Warren: Tracker training

• WWWard_nov_15_07_med_res_2.mov: Warren Ward Nov 14, 2007 Instructions
• Warren_DPI_training_tape_1.mov: Warren DPI August 2007 Training Tape (part 1)
• Warren_DPI_training_tape_2.mov: Warren DPI August 2007 Training Tape (part 2)
• Warren_DPI_training_tape_3.mov: Warren DPI August 2007 Training Tape (part 3)
• Warren_DPI_training_tape_4.mov: warren DPI August 2007 Training Tape (part 4)

Movies from Warren: How to repairs

• Mirror-Motor_Repair.exe: Executable lecture on how to repair the mirror-motor
• 4th_Motor_Replacement.exe: Executable lecture on the 4th Motor Replacement

Dental Impression Compound:

Johnson & Lund Co. Rochester, NY 14623-2824 (585) 475-0160
Monday - Friday 8:00 a.m. - 4:30 p.m.
Winton Place, Bldg D, last door

Kerr Impression Compounds are available in both cakes and sticks. The different colors signify the softening temperatures of each compound and are as follows:

Soft Green 122-124°F (50-51°C)
Grey 128-130°F (53-54°C)
Red 132-133°F (55-56°C)

Directions:
Thoroughly soften the compound in clear water in the temperature range noted above. Knead the compound with warm, wet fingers to be sure that it is uniformly softened throughout.

• Adapt to a suitable impression tray and flame surface
• Dip into hot water before inserting into mouth.
• Hardening can be hastened by use of cold water or cold air.
• Working temperatures for cakes and sticks are in the same range as specified above.
• Sticks may be softened over a flame to a workable condition.

• Kerr_impression_compound_information_sheet.pdf: Kerr_impression_compound_information_sheet.pdf

-- JeffPelz - 13 May 2007

• Set ALLOWTOPICCHANGE = VplGroup

-- SueKola - 02 Mar 2007

Computer instructions

• Display_Mirroring.doc: How to enable display mirroring for Mach

• USB-1208_6_USB-2527_calibrate_43.m: USB-1208_6_USB-2527_calibrate_43.m

• DPI_binoc_ONLY_228.m: Capture, display, and log binocular DPI data

Pictures

* DSCN_1469.JPG: * DSCN_1470.JPG: * DSCN_1471.JPG: * DSCN_1473.JPG: * DSCN_1474.JPG: * DSCN_1476.JPG: * DSCN_1477.JPG: * DSCN_1479.JPG: * DSCN_1480.JPG: * DSCN_1482.JPG: * DSCN_1483.JPG: * DSCN_1484.JPG: * DSCN_1485.JPG: * DSCN_1486.JPG: * DSCN_1487.JPG: * DSCN_1488.JPG: * DSCN_1489.JPG: * DSCN_1490.JPG: * DSCN_1491.JPG: * DSCN_1492.JPG: