Driver behavior indices from large-scale fleet telematics data as surrogate safety measures

Area of Science:

• Transportation Engineering
• Traffic Safety
• Urban Planning

Background:

• Large-scale telematics data offer high-resolution insights into road safety and driver behavior.
• Existing research has explored safety surrogates and crash predictors from telematics, but a comprehensive analysis linking driver behavior to crashes and road networks in cities is lacking.

Purpose of the Study:

• To analyze driver behavior indices derived from telematics data.
• To correlate these behaviors with crash data and road networks in New York City.
• To propose novel safety metrics for data-driven network screening.

Main Methods:

• Extracted driver behavior indices (speed, speed variation, hard braking/acceleration rates) from telematics data of 4000 vehicles in NYC.
• Compared these indices with street-level collision frequencies and rates.
• Utilized a traffic AADT model for normalizing crash frequencies with traffic volume.

Main Results:

• Moderate correlations were found between driver behavior indices and collision rates.
• Hard braking is linked to higher collisions on highways; hard acceleration on urban roads.
• Speeding on highways indicates collision risks, while higher travel times correlate with crashes on non-highway roads.

Conclusions:

• Telematics data can effectively identify safety-critical regions and driver behavior patterns related to crashes.
• Proposed safety metrics like speed corridor maps and hot-spots can advance road safety assessment.
• Data-driven network screening using telematics holds significant potential for improving urban road safety.