Engineering Design Centre Project Updates –August 2014
Engineering Design Centre Aim of the study To compare –performance and preferences of users for different input modalities –a standard computer mouse and HOTAS Joystick with different eye-gaze, head and hand movement tracking based pointing systems
Engineering Design Centre Vision parameters
Engineering Design Centre Cognitive parameters Trail Making TestDigit Symbol Test
Engineering Design Centre Motor parameters Range of Motion of wrist (Palm facing down) Measuring Radial Deviation Measuring Ulnar Deviation
Engineering Design Centre Participants AgeSexNationalitiesGSROMWDSTTMTVACB 126FIndian N 223FBritish N 353MBritish N 434MIndian N 530FPolish N 646MBritish N 728MSouth African N 823MBritish N 919MBritish N 1030FItalian N 1122MAmerican N 1230MBritish N 1326MSpanish N
Engineering Design Centre Task
Engineering Design Centre Design Modalities 1.Eye Tracking with Z-Axis selection (ZET) 2.Eye Tracking with Voice-based selection (VoiceET) 3.Adapted Eye Tracking (AdaptedET) 4.Multimodal Eye Tracking (MmET) 5.Head Tracking with Z-Axis selection (ZHT) 6.Head Tracking with Voice-based selection (VoiceHT) 7.Adaptive Head Tracking (AdaptedHT) 8.Hand Tracking with Z-Axis selection (ZGS) 9.Hand Tracking with Voice-based selection (VoiceGS) 10.Adaptive Hand Tracking (AdaptedGS) 11.HOTAS Joystick 12.Mouse Targets –Size 45, 55, 65, 75 pixels –Distances 80, 160, 240, 325 pixels [ 1 pixel ≈ 0.25 mm ]
Engineering Design Centre Results All participants can undertake trials in all conditions We measured –Selection time –Cursor trace #Wrong Selections Main movement + Homing Time Extra Distance Travelled over Target Axis Length –Pupil diameter –TLX scores –BRS scores
Engineering Design Centre Selection Times
Engineering Design Centre Fitts’ Law
Engineering Design Centre Selection Times - ANOVA SourcedfFSig. Eta Squared DEVICE Error(DEVICE)96.36 WIDTH Error(WIDTH)52.70 DIST Error(DIST)72.00 DEVICE * WIDTH Error(DEVICE*WIDTH) DEVICE * DIST Error(DEVICE*DIST) WIDTH * DIST Error(WIDTH*DIST) DEVICE * WIDTH * DIST Error(DEVICE*WIDTH*DIST) Device × Width × Distance
Engineering Design Centre Selection Times - ANOVA Point × Select × Width × Distance SourcedfFSig. Eta Squared POINT Error(POINT)31.58 SELECT Error(SELECT)48.00 WIDTH Error(WIDTH)72.00 DIST Error(DIST)49.56 POINT * WIDTH Error(POINT*WIDTH)82.77 SELECT * WIDTH Error(SELECT*WIDTH)88.56
Engineering Design Centre Interaction Diagrams - Pointing
Engineering Design Centre Interaction Diagrams
Engineering Design Centre Analysing Trajectory Source Target Reached Target Click: End of Task
Engineering Design Centre Main Movement Time
Engineering Design Centre Time Spent near Target Homing Time = Selection Time – Cursor reached Target
Engineering Design Centre Selection Times – Device Comparison
Engineering Design Centre Extra Distance Travelled Extra Distance = Total Distance – Target Axis Length
Engineering Design Centre Wrong Selections
Engineering Design Centre Cognitive Load
Engineering Design Centre Spare Mental Capacity
Engineering Design Centre Synopsis Pointing –Mouse << Head Movement << Hand Movement << Eye Gaze << HOTAS Joystick Selection –Adaptive << Z-Axis << MmET << Voice Cognitive Load –Mouse << AdaptedHT << …<< Joystick << VoiceET –Not enough spare mental capacity for VoiceGS, VoiceET and HOTAS Joystick
Engineering Design Centre Pupil Data Analysis
Engineering Design Centre Maximum Pupil Diameter
Engineering Design Centre Uncertainty Principle
Engineering Design Centre Gabor Limit
Engineering Design Centre Pupil as a wave signal
Engineering Design Centre Existing work & Patent Sudden change in pupil diameter Methods –Wavelet Transform –Linear Discriminate Analysis –Neural Network Applications –Driving simulation –Aviation –Map reading –ET as passive not active
Engineering Design Centre Power Spectrum Participants ρ P10.94 P20.93 P30.67 P40.24 P50.73 P60.64 P70.19 P80.51 P90.72 P P P P131 Average0.83
Engineering Design Centre Comparing Correlations with TLX Scores
Engineering Design Centre Power Spectrum vs TLX Scores
Engineering Design Centre Power Spectrum vs Selection Times
Engineering Design Centre Power Spectrum for different IDs
Engineering Design Centre Power Spectrum vs Selection Times
Engineering Design Centre Comparing Effect Sizes ( η² )
Engineering Design Centre Advantages over Previous work FFT over FWT –No need to choose basis wavelets Application agnostic evaluation Works with low frequency (and cheaper) eye gaze tracker Validated while ET is used in both active and passive modes However distance to target can also affect pupil diameter which may not indicate a change in cognitive load