1. Home
2. Ideas
3. ASSESSING THE RELIABILITY OF HARD BRAKIN...

ASSESSING THE RELIABILITY OF HARD BRAKING AND OTHER VEHICLE-BASED METRICS AS SURROGATE SAFETY MEASURES

• Jun 10, 2025

Research Description

Several agencies and data providers have suggested that high frequencies of hard braking imply increased probabilities of near misses and crashes. Those agencies have been using these data collected from connected vehicles (CVs) as a reasonably reliable surrogate safety measure. In addition, several agencies have also used high acceleration and high speeds at cornering as surrogate measures for safety that could be used in addition to hard braking. However, the validity of using these surrogate measures might not be all-inclusive or fully representative for all conditions (e.g., sight distance constraints, geometric conditions, speed, traffic control devices, work zone configuration, queue warning locations). This research will use statistical analysis and modeling to identify and validate the conditions under which hard braking and other vehicle-based metrics may be used as reliable surrogate safety measures. In addition, the research will examine and contrast the effectiveness of using these surrogate measures with other potential novel and newly discovered measures and whether new combined models could be used for predicting crashes with increased certainty.

Objective The objective of this research is to evaluate the correlation between hard braking and other vehicle-based metrics with actual crashes and to assess their applicability as reliable surrogate safety measures to CV locations and apply safety countermeasures.

Potential benefits The research is expected to result in timely improvements in transportation safety due to the utilization of selected surrogate safety measures to implement corresponding effective operations strategies before the occurrence of crashes. This will lead to a reduction in the number and severity of crashes, improving mobility metrics and providing safe mobility.

Additional Supporting Information

The integration of probe data into traffic management has facilitated detection of interstate queues, crash-prone areas, and work zone risks. Studies have examined critical aspects of traffic safety, particularly hard-braking events related to hydroplaning, traffic signal operations, and risk prediction models. Hard-braking data has been useful in identifying locations susceptible to rear-end crashes, allowing for quicker mitigation strategies compared to traditional crash analysis methods. Research at urban intersections highlights how multiple data sources can be used to identify rare but high-risk driving behaviors, serving as valuable test cases for validating automated vehicle systems. Methodologies developed for using high-resolution vehicle trajectory data have improved traffic signal performance measures by leveraging CV data. Research into driver hazard perception has shown that individuals with lower hazard awareness tend to experience more hard-braking events.

Hydroplaning incidents have been analyzed in relation to hard-braking events, revealing that hard braking occurs nearly four times as often, typically at lower speeds. Geospatial analyses of regional transportation networks show that clusters of hard-braking events align with crash hotspots, indicating that these incidents can serve as early warning signs for traffic safety concerns. Research into driver behavior further supports this idea, demonstrating that high-risk drivers can be identified by analyzing time-invariant driving patterns. Surrogate safety metrics suggest that frequent, low-severity incidents such as hard braking can serve as effective indicators of crash-prone areas, providing a more immediate alternative to relying solely on historical crash data.

Predictive modeling techniques, such as hierarchical dynamic models and vehicle telematics data, have been employed to assess crash risks in real-time. These models demonstrate a strong correlation between hazardous traffic patterns and the frequency of hard-braking incidents, particularly on freeways and during peak travel times. Further research has integrated CV data into safety performance functions, improving the assessment of road risks in both urban and rural settings. This data has also been used to enhance work zone and winter operations management by providing real-time insights into emerging traffic patterns.

Advancements in machine learning have further refined hard-braking detection, with smartphone-based models significantly outperforming traditional GPS or accelerometer-based methods. These technologies have been applied at scale, enabling more precise identification of road safety risks.

These studies underscore the value of leveraging real-time driving data, machine learning, and CV technologies to enhance traffic safety. By identifying high-risk behaviors and hazardous locations more efficiently, transportation agencies can implement proactive safety and operations strategies to reduce crashes and improve operations.

This research aligns with the 2021-2026 AASHTO Strategic Plan by addressing: • Safety, Mobility, and Access for Everyone: Identifying the most effective surrogate safety measures and incorporating them in the design of transportation systems operation strategies to reduce the number and severity of predicted crashes advances a safe, multimodal transportation system. • National Transportation Policy Leadership: Evaluating and identifying effective surrogate safety measures that use advances in data collection technologies provide proactive leadership in transportation innovation. The results can support greater safety integration in the TOM.

This research would also support the Safe System Approach, which emphasizes proactive applications.

---

Submitted By

Tracy Scriba

FHWA

202-366-0855

Attachments