Issue 1 - How to measure conflicts?
Generally, two dimensions can be used to measure traffic conflicts: the closeness (proximity) to a crash and the intensity of evasive actions taken by road users to avoid a crash1. Many conflict indicators have been proposed to measure conflicts, the majority of which measure time/space proximity (the degree of closeness to a crash in time and space). However, most conflict studies use a single measure, most often the time to collision (TTC) or the post encroachment time (PET). Relying on one conflict indicator to measure traffic conflicts can represent a serious pitfall. In many real-world circumstances, it is difficult to differentiate between true traffic conflicts and normal changes in the position and velocity of one road user while navigating a traffic stream1 This uncertainty causes conflict indicators to only represent a partial image of the true severity of traffic events . It may also explain why time proximity indicators such as TTC sometimes may fail to capture true severity aspects of traffic conflicts.
In view of the limitation of using one single conflict indicator, studies have proposed to combine different indicators to gain better understanding on the underlying severity of an event. In general, the conflict indicators used should cover three aspects, including temporal and spatial proximity, intensity of evasive action, and the consequence of potential collision[2]. An example of using the strength of pedestrian evasive action (running, sudden stopping, etc.) to measure conflicts is shown in the Figure 1. Research by AMAG founders has shown that understanding crash risk requires the measurement of several unique conflict indicators.
[2] Ismail, K., Sayed, T., and Saunier, N. (2011) “Methodologies for Aggregating Indicators of Traffic Conflict”, Transportation Research Record: Journal of the Transportation Research Board, Vol. 2237, pp. 10–19.
Issue 2 - Which conflict threshold to use?
Issue 3 - The accuracy of conflict extraction
The foundation of using conflicts to predict crash risk is built on the accuracy of the process to extract conflicts from the video, Lidar or connected vehicle data. Research has shown that only extreme conflicts are the best predictors of crash risk, and the identification of true extremes is highly sensitive to the quality of the conflict extraction process. Many conflicts act simply as a measure of traffic volumes and do not really reflect crash risk.
The accuracy of conflict extraction is very sensitive to several factors usually not considered by many. Ignoring these factors can completely bias the analysis, leading to incorrect conclusions regarding underlying safety. These factors include:
- Road users are assumed to be represented by points (e.g., centroid) and the road user size (e.g. vehicle size) is usually ignored. This can lead to significant underestimation and considerable error in conflict indicator values.
- When using video data, to accurately recover the positions of various road users, geometric elements need to be mapped back from camera image space to the world space. This is a necessary step, since accurate measurements are only possible in terms of world space coordinates. This mapping process is called “camera calibration” and is ignored in many systems that rely instead on tracking in the image pixel space. Conducting road user tracking in real-world coordinates significantly improves the accuracy of the tracking performance by correcting for perspective effect and other distortions due to projection on the image plane. Ignoring this step leads to significant errors in measuring quantities such as distances, speeds, deceleration and conflict indicators [4],[5] Even when camera calibration is undertaken, a poor camera calibration can also cause significant errors in conflict indicator values.
- The identification of conflict measures involves modeling the expected path of a moving road user. Naïve extrapolation techniques should be avoided in favour of more reliable but computationally complex probabilistic approaches.
Summary
Not all systems for extracting traffic conflicts are the same and the reliability of the results needs to be established before relying on them to evaluate safety and predict crash risk. It is important to recognize that not only are several conflict indicators required to fully understand crash risk, but different crash types are related to different conflict types and with different threshold values. Furthermore, these relationships may change between jurisdictions with different traffic environments. Improperly selected conflicts and thresholds may merely be measuring the amount of traffic instead of a true underlying crash risk that can be mitigated. The conflict extraction process must also consider vehicle size, calibration of the camera (for video data), and sophisticated methods for road user path prediction. Not doing so may result in seemingly small errors that render the results meaningless.
As an emerging technology in road safety management, proactive methods will require ongoing research and development to continually refine, improve, and validate the techniques in the real world. Thus, rapid translation of this rapidly developing technology from research to practice (i.e. commercially available tools) is key for these technologies to have maximum impact, and subsequently reduce injuries and fatalities on the global transport infrastructure.
About the Author
Dr. Tarek Sayed is a Co-Founder of Advanced Mobility Analytics Group (AMAG). Professor Sayed has won numerous awards in recognition of his contributions to road safety, including the UBC Distinguished University Scholar, ITE Wilbur Smith Distinguished Transportation Educator, Canadian Society of Civil Engineering Sandford Fleming, and Transportation Association of Canada Centennial Road Safety Award, Prince Michael International Road Safety Award and several best paper and teaching awards. He is also a Fellow of the Engineering Institute of Canada, the Canadian Academy of Engineering, and the Canadian Society for Civil Engineering.. Professor Sayed is a world-renowned leading authority in the development of video analytic methods for road safety and pioneered much of the early work in this field.
