MicroTraffic is a leading provider of microscopic traffic data for road safety engineers. Our technology is based on computer vision, video analytics, machine learning, predictive modelling, and the safe systems philosophy. We assist engineers to apply surrogate safety methods for proactive decision making. We produce conflict reports for individual traffic safety studies and also offer a proactive road safety network screening service.

© 2019 by MicroTraffic Inc.

Measuring pedestrian left turn risks in the GTA

MicroTraffic has measured pedestrian left turn risks at 140 locations in the Greater Toronto Area (GTA). This article focuses on risk measurements at a subset of 24 crossings with permissive signal phasing at arterial-arterial intersections in suburban areas of the GTA.

Collisions between left turning vehicles and pedestrians are a big problem in urban areas. Effective mitigation strategies for this collision type are essential to any vision zero strategy for road safety. The financial or operational costs of mitigation strategies can be high. Therefore, risk measurement should be used to focus the application of mitigation strategies to places where they are needed the most.

The driver of left turning white car focuses on gaps in oncoming traffic, placing the downstream pedestrian at risk.

Measuring the Severity of Left Hook Pedestrian Risk Events

The safe systems conflict measurement technique used by MicroTraffic considers the time by which a collision was avoided and the speed of the left turning vehicle at the potential point of impact to (a) determine if a conflict occurred and (b) assign it a risk category. The figure below shows an example for northbound left vehicles in conflict with pedestrians on the west crosswalk at an intersection in the GTA.

Risk scatterplot for NBL vs pedestrian on west crosswalk. N = 898 peds. n_critical = 0 (red). n_high_risk = 8 (orange). n_medium_risk = 63 (yellow).

Conflict Rate Data

At the 24 locations analyzed, a total of 14,280 pedestrians were observed. The mean observed pedestrian count per location was 618. In this case, a location refers to a single crosswalk at an arterial-arterial intersection in a suburban area of the GTA. All locations were controlled by protected-permissive or permissive signal phasing.

No critical risk conflicts were observed.

The mean involvement rate in high risk conflicts was 0.16%, ranging across sites from 0.000% to 0.900% with a standard deviation of 0.26%, all expressed as a percentage of crossing pedestrians.

The mean involvement rate in medium risk conflicts was 8.75%, ranging across sites from 0.00% to 33.33%, with a standard deviation of 7.52%, all expressed as a percentage of crossing pedestrians.

The mean involvement rate in low risk conflicts was 8.86%, ranging from 0.00% to 33.33% with a standard deviation of 7.50%, all expressed as a percentage of crossing pedestrians.

82.23% of pedestrians were not involved in any form of conflict with left turning vehicles (i.e. a time separation of more than 3 seconds was present).

Detection of Risk Anomalies

The relatively high variance in the data suggests that measurement of risks can isolate specific locations and specific turning movements where risks are abnormally high.

4 of the 24 crossings presented high risk involvement rates for pedestrians vs left turns that were between 2 and 6 times higher than the average high risk involvement rate. These 4 crossings are obvious candidates to be further analyzed for prioritized risk mitigation measures.

Reducing Risk of Pedestrian vs Left Turn Crashes

Measures to reduce risk of pedestrian vs left turn crashes generally fall into the categories of visibility enhancements, signal operations, and speed management.

1. Visibility and conspicuity enhancements to manage left turn pedestrian crash risk

Negative or neutral offsets of opposing left turn lanes make it difficult for drivers to see around an opposing left turning vehicle to opposing oncoming traffic. This can make the gap search and acceptance task very difficult, consuming all of the driver's attention. Because they are not paying adequate attention to the downstream crosswalk, when they do choose a gap a pedestrian may be there. An effective mitigation strategy in this scenario is to introduce positive offset opposing left turn lanes.

Effective lighting of crosswalks is not present at all intersections. Lighting improvements can target contrast, luminance, glare, and uniformity to enhance the night conspicuity of pedestrians.

Crossing areas at intersections typically have pavement markings to help with driver's general awareness of the conflict area. If these pavement markings are worn, or are not very visible at night or in wet conditions, the conspicuity benefit of the pavement markings will not be achieved. Improvements to consider include using durable paints with reflective bead content that functions under wet weather conditions.

2. Signal operations measures to reduce pedestrian left turn crash risk

Converting a signal to operate in protected only left turn mode can eliminate a large majority of left turn pedestrian risk. In this mode, left turn vehicles have a dedicated phase during which pedestrians are not allowed to cross, and left turn vehicles are not allowed to proceed during the pedestrian phasing. This technique is also extremely effective in reducing left turn vs oncoming vehicle crashes. This treatment increases motorist delay. In some countries such as the UAE, permissive left turn phasing has been banned completely.

Depending on the time of day profile of the pedestrian vs left turn risks, a signal timing plan with protected permissive phasing could be modified to change the length of protected phasings. It can also operate under protected only control for certain portions of the day.

A leading pedestrian interval (LPI) can give pedestrians a chance to establish themselves in an intersection before conflicting turning vehicles are released. This can also be considered a visibility enhancement. Montreal has systematically applied LPIs at all signalized intersections while New York has recently installed hundreds of LPIs.

Emerging ITS technologies can also be deployed to operate signals on a risk adaptive basis. For example, a signal with a lagging protected left turn can be programmed to delay the start of the protected left if the crosswalk is still occupied by pedestrians.

3. Speed management measures to reduce pedestrian left turn crash risk

The speed at which a typical left turning vehicle exits an intersection is a major determinant of the risk posed to pedestrians on the exiting crosswalk. Recently, a lot of attention has focused on right turn curb radii reduction to manage speeds of right turning vehicles as they conflict with pedestrians. Equivalent strategies and innovative variants can and should be considered for left turn speed management. A wide intersection and high radius left turn will allow high exit speeds.

Exit Speed Profiles for Southbound Left Vehicles at Downstream Crosswalk (point of potential impact with pedestrian)

Strategies to manage left turn intersection exit speeds include applications of slightly raised crosswalks, positive offset left turns (slot lefts) which reduce the left turn distance, road width reductions, and modifications to turning radius through pavement markings or bullnose extensions.

There are some strategies to mitigate left turn pedestrian risks that fall outside of the above three categories. For example complete restriction of a left turn, or closure of a pedestrian crossing to pedestrians are highly effective measures in terms of safety, but could be considered difficult to accept in most cases.

Our vision for vision zero - getting there through risk measurement and management.

If an agency wants to get to vision zero, addressing left turn pedestrian crash risk is a must. Responding to locations with past fatalities is a not a productive approach because crash counts at individual locations are so low that they cannot reliably measure risk. Systematic application of improvement measures has potential, but it could be infeasible due to budget constraints and it could also alienate stakeholders who are angered by delay implications at locations where the measures are not needed.

A better approach, in our view, is to take the the major intersections in a network - maybe a few hundred or a few thousand, and measure the near misses systematically in a network risk screening program. In doing so, the agency would know exactly where the highest risks are and exactly which fixes would drive the greatest ROI.