9. PLAN DE MANEJO AMBIENTAL DE LA PLANTA DE TRATAMIENTO DE AGUAS
9.8. Plan de contingencias
2.6.1 Electrooculography (EOG)
EOG data was recorded using a BlueGain EOG Biosignal Amplifier at 1000Hz (Cambridge Research Systems Ltd., UK). Horizontal eye movements were recorded by placing two surface electrodes adjacent to either eye’s lateral canthi. To record vertical eye movements, electrodes were placed centrally on the upper and lower orbital rim of the left eye socket. An earthing electrode was placed in the centre of the forehead (Figure 2-6).
Figure 2-6 A diagram showing the position of EOG electrodes on the face.
Electrodes fed in to an arm mounted biomedical amplifier, which wirelessly transferred the data in real-time to a computer. A manually operated infrared trigger sent pulses of infrared light to a receptor channel on the BlueGain amplifier, and fed in to the Vicon system at 1000Hz allowing for accurate temporal alignment between the EOG and kinematic data. The electrodes detected the potential difference between the positively
charged cornea and negatively charged retina of each eyeball. When the eyes rotated in their socket, the corneas moved closer to an electrode altering the potential difference between the electrode pairs. This was detected and recorded on an electrooculogram as vertical and horizontal eye position within the head.
Prior to testing, each participant was asked to fixate gaze on a central point straight ahead and rotate their head from side-to-side and then up and down. Head rotation in any direction was accompanied by eye counter-rotation driven by the vestibuloocular reflex. The eye movement signal generated was then correlated with the angle of the head segment to produce a scale-factor by which their EOG data could be translated into degrees. Raw EOG data was filtered using a zero-phase second-order Butterworth filter with a cut-off frequency of 30Hz, and then converted to degrees (see Appendix F.2 for Matlab script). Saccade initiation was identified as when the velocity of the eye movement was greater than 100°s-1 (Young et al., 2011). The time of these eye
movement onsets were marked on the EOG trace with a vertical line depicting foot contact time and the saccade closest to the foot contact time being highlighted by default. In some trials, small deviations in the EOG trace could result in incorrect identification of a saccade; therefore each trial was visually examined to ensure the largest, closest saccade to foot contact time was identified (see Appendix F.3 for Matlab script). The time difference between the saccade initiation and foot contact time was then calculated (Figure 2-7). Due to some technical errors with the synchronisation pulses it was not possible to temporally align the EOG and kinematic data in the study presented in Chapter 3, therefore saccade timings were excluded from this study.
Figure 2-7 An example of trial EOG data showing how gaze transfer time was calculated. The x-axis represents time and the y-axis represents vertical eye rotation within the head. The vertical red line indicates foot contact in the stepping target (F.C.). Participants generally looked at the target box during the final part of their approach as can be seen as the trace drops. Their gaze transfixion changes around foot contact time to focus further obstacles or steps (the saccade). Saccade initiation (S.I.) was highlighted when velocity surpassed 100°s-1. The time
difference between foot contact time and saccade initiation was used to measure gaze transfer time. In this example gaze transfer occurred following foot contact, so the value would be positive.
2.6.2 Video Eye Tracking
In the final study presented in this thesis (Chapter 6), a 25Hz Dikablis wireless head- mounted monocular eye tracker was used (Ergoneers GmbH, Manching, Germany). This system used a small infrared camera pointing back in to the subjects left eye to correlate relative pupil position to locations from a scene-view forward facing camera. The luminance of the infrared view of the eye was adjusted such that the pupil was the
darkest area of the image (Figure 2-8). The built in Dikablis algorithm then isolated the pupil and found the centre point, meaning that the pupil could be tracked within the image. Participants were then asked to stand facing the obstacle course while four small reflective vicon markers were placed near the corners of the scene view capture area to identify a calibration area. To calibrate the pupil position to the scene view, participants were asked to stand with their head still and to only move their eyes as they looked at each of the 4 calibration points. When focused on these points, the respective points on the scene view image were selected, allowing calibration of the area between them using the Dikablis software. Once the calibration was complete, a crosshair could be recorded on the scene view to identify where each participant was looking.
Throughout the trials in this study, 20cm square black and white markers were placed throughout the trial area (Figure 2-9a); two parallel to the near and far left edge of the target box, and two on either side of each obstacle. These markers were automatically detected in each trial by the D-Lab software (Ergoneers GmbH, Manching, Germany). This means that areas specific to the target and obstacles could be marked out within the video recordings (Figure 2-9b), and gaze fixations and timings within these areas throughout each trial could be measured. Sectioned events (i.e. preview time start and
Figure 2-8
An example of the image used by Dikablis to isolate the pupil.
finish, and walk time start and finish) were marked on the video recording in D-Lab. The output variables of D-Lab were total fixation time, mean fixation time, number of fixations, percentage fixation time of the section (preview or walk), and fixation frequency (Hz).
Figure 2-9 a) An example of a marker image automatically detected in the D-Lab software, and b) a screenshot of the marker images being identified (outlined in red), and the areas which were anchored to them (blue) while the scene view moved. The green crosshair represents current fixation.