![]() The length of the sliding window is controlled by a parameter on the EyeLink tab. The calculations are independent for X and Y. If recording binocularly, the left eye data is in HGPOS, VEPOS and the right eye in HGPOS2, VEPOS2.Įye velocity is computed “on the fly” from EyeLink data by using a central-point difference method, followed by a sliding-window average to smooth the result. When recording monocularly, the EyeLink position data is placed in Maestro analog input channels HGPOS, VEPOS as you would expect. Using gain and offset parameters set by the user on Maestro's EyeLink tab page, raw pupil coordinates (x,y) from the tracker are converted to Maestro’s eye position coordinates - as 12-bit digitized voltages proportional to visual degrees subtended at the eye. No EDF file is recorded.ĮyeLink data is converted to Maestro format. No calibration is needed on the EyeLink side, since only raw pupil data is used. It ASSUMES that the EyeLink is sampling eye data at 1KHz. Maestro does little more than connect to the tracker, configure it to send only raw pupil samples over the Ethernet link, and tell it when to start and stop recording. ![]() The GUI remains responsive while recording (and saving) trials using RMVideo for trial presentation and servicing the 1KHz eye position sample stream from the EyeLink. This approach was obviously preferred over adding a “relay” computer between the Eyelink Host PC and the Maestro PC, and testing with the EyeLink in “Mouse Simulation Mode” uncovered no performance issues with the design. Eye position data is received over the EyeLink Ethernet connection and, in turn, streamed to Maestro’s RTX-based driver ( cxdriver) over the existing IPC (interprocess communications) structure. ![]() Instead, taking advantage of the fact that Windows has sole ownership of 2 CPU cores in the lab’s current quad-core Maestro workstations, the service routine is implemented as a Win32 thread running in the Maestro GUI process. However, the EyeLink SDK is not compatible with RTX. Given that a new sample is arriving over the link every millisecond, it would seem best to implement an EyeLink service routine in Maestro's RTX driver. ![]() Of course, to take advantage of this Ethernet link, Maestro must be able to communicate with the EyeLink's Host PC over the TCP/IP connection and "keep up" with the 1KHz data stream when eye position is being recorded. The position trajectories generated by the tracker thus remain in digitized form - not subject to the system noise to which the analog signals are susceptible (the manufacturer actually includes an analog output option with the EyeLink, but discourages its use because of noise). The EyeLink 1000+ eye tracker offers another option: eye position data can be streamed at 1KHz over a fast Ethernet connection. Unfortunately, the analog signals from the lab's old eye tracker were rather noisy. These tracker systems typically have analog outputs representing eye position and/or velocity, and in the past we have connected such outputs to Maestro's HGPOS, VEPOS, HEVEL and VEVEL inputs to record eye trajectories of a human subject. In eye movements studies in humans, a high-speed camera-based tracker system is typically used to monitor the subject's eye position during an experiment.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |