What decisions do drivers make when their lane is ending and they are traveling at highway speed? Gorski Consulting has conducted testing with traffic cameras on Highway 401 to find out.

At highway speeds, with high traffic densities and a sudden emergency, drivers need to make quick decisions that could cost them, or drivers around them, their lives. Decades of research has helped to create the roadway environments that reduce the need to make those sudden decisions. Yet, unexpected events can occur.

Highway 401 in southern Ontario has received its share of complaints. From Windsor to Montreal this super-highway carries the highest numbers of Canadian drivers, by far, than any other. Its character changes greatly whether it contains multiple express and collector lanes in Toronto, or whether it contains only two lanes without a medianĀ  in areas east of Kingston or close to Chatham.

Construction, maintenance and policing activities become a problem when traffic volumes increase. Often lanes become closed for these activities and drivers must make adjustments in their speed and lane travel. Numerous collisions occur in construction zones when closed lanes cause drivers to change lanes or bring traffic to a halt. Many collisions occur when unsuspecting drivers approach the stopped traffic at highway speed but are too late in their detection that traffic is stopped. This is more problematic when heavy trucks are loaded with cargo but their braking systems make if difficult to stop as quickly as passenger cars and light trucks. Numerous problems like these require objective data in order to develop counter-measures that provide realistic solutions to the problem.

Recently Gorski Consulting has been conducting observations along Highway 401 to gather the objective data that may form the grounds for improvements in the future. The use of synchronized, multiple, video cameras, placed in a short-range environment, allow for data to be collected about driving patterns and drivers’ actions.

The most recent observations were conducted on October 30, 2018 on Highway 401 near the Westminster Driver overpass on the outskirts of London, Ontario. This location has moderate traffic volumes that allow a relatively free flow. Considerable construction activities are currently taking place between Chatham and London however, at the time of the observations, no such construction activity existed in the vicinity.

The typical procedure is to paint markers at 100 metre intervals along the road edge. A video camera is pointed at the marker and it documents when vehicles pass that reference point. At the next 100 metre location another video camera captures the same vehicle. By knowing the time interval that it takes a vehicle to pass between the two markers an average speed is obtained. Other information such as the following distance of one vehicle behind another is also obtained.

Video cameras placed along the side of the highway document vehicles as they pass by paint markers and traffic cones placed at 100 metre intervals. This method provides information about the average speed of the vehicles.

The Westminster Drive site is particularly useful because the westbound lanes of Highway 401 become reduced from 3 lanes to just 2. This produces observations of drivers who must change lanes. Information about how long drivers wait to change lanes before reaching the end of the lane can be useful as failures in this decision making can be the source of collisions. Information is also obtained about how other drivers react to a vehicle attempting to cross into their lane and if they are accommodating and helpful or otherwise.

Panoramic views along the highway are also taken from the Westminster Drive overpass with cameras pointing east and west. This provides an additional perspective of the traffic.

View of two video cameras positioned on top of the Westminster Drive overpass and pointing at traffic east and west of the location.

Decisions regarding when to make a lane change are still far away from being completed by self-driving vehicles. While an algorithm developed through substantial research may provide the correct response in high percentages of instances, there are still situations where a human driver may be able to foresee something that the computer cannot, thus requiring human control to change the vehicle’s motion.

In this view the driver of the red car has succeeded crossing out of the ending lane at the very last instance. Decisions such as these can be problematic when traffic volumes are high and opportunities to make that lane change are limited.

The study of human decisions and how they drive can be beneficial to the creators of self-driving vehicles because, even if self-driving cars fully populate the road system there will still be a long ramp up process when self-driving cars will need to live among human decision makers.