Driving Performance Data Verification Process at the NADS

Driving Performance Data Verification Process at the NADS Judith Wightman Ginger S. Watson I/UCRC Program Review XVII

Data Verification at NADS • • • Overview Raw Data Reduction Verification Tools Issues, Concerns, Future Research I/UCRC Program Review XVII

Data Verification at NADS • The goal is to ensure that collected data are – Reliable • Consistent or reproducible – Valid • Meaningful and relevant – Complete, yet concise I/UCRC Program Review XVII

Raw Data • Master list of up to 400 internal variables available on NADS • Subset chosen based on goals of experiment – Binary collected in the data acquisition system (DAQ) – Various frequencies available per variable (up to 240 Hz) – Data can be converted to other formats and used in various ways • • • MATLAB ASCII Excel ISAT (Interactive Scenario Authoring Tool; for playback) Reduced by coded algorithms I/UCRC Program Review XVII

Data Reduction • The process of transforming raw data into meaningful summary variables (called “reduced data”) that address experimental hypotheses • Components – Identifying and defining summary variables of interest – Filtering the data – Applying various algorithms (automated or human processed) to obtain the desired measures – Verifying that reduced data are valid (meaningful) and reliable (consistent) I/UCRC Program Review XVII

Data Reduction Methods • Automated code – Used to transform many data points into a concise measure such as reaction time – Examples • Reaction time to 10% release in accelerator pedal from point of incurring vehicle creation • Minimum time to collision with incurring vehicle from point of incurring vehicle creation I/UCRC Program Review XVII

Data Reduction Methods • Human coding – Used when video analysis or decomposition of nondriving behavior is needed, or when more efficient than automated coding – Examples • Determination of number of seconds taken to dial a wireless phone • Loss of control in ESC studies (90 degrees nonzero yaw rate) I/UCRC Program Review XVII

Verification Tools • ISAT – Displays top-down view of logical database and scenario saved in raw data stream – Allows for review of participant drives and examination of various reduced measures – Examples • Verification of reaction time • Verification of accelerator release I/UCRC Program Review XVII

Video Example I/UCRC Program Review XVII

ISAT Playback: Incurring Vehicle Creation I/UCRC Program Review XVII

ISAT Playback: Accelerator Release I/UCRC Program Review XVII

ISAT Playback: Brake Onset I/UCRC Program Review XVII

ISAT Playback: End of Event I/UCRC Program Review XVII

Verification Tools • SAS/Excel – Used to calculate measures of central tendency for group or treatment, plot, or statistically analyze measures – Examples • Velocity variance as calculated in SAS/Excel equivalent to the reduced variable velocity instability in certain situations • Excel used for calculating length of various parameters of wireless call data (e. g. , dial/answer, connect, converse, disconnect) I/UCRC Program Review XVII

Verification Tools • MATLAB – Used to plot performance over time – Example • Sanity checks on phase (continuous measure of the delay in the participant’s reaction time to changes in lead vehicle’s speed) I/UCRC Program Review XVII

MATLAB Examples I/UCRC Program Review XVII

Issues, Concerns, Future Research • Missing Data • Generalizability • Questions I/UCRC Program Review XVII