Explico Performs Study to Determine Pre-Impact Behaviors

When analyzing a vehicle collision, often there is testimony, roadway markings, EDR data, etc. that can provide evidence as to what took place prior to a collision. However, in other cases, there is minimal documentation or evidence available, in which case naturalistic observations of human driving can prove useful in estimating a driver’s actions before an accident. Naturalistic driver observations are made from studies that analyze drivers in their natural state when they do not know they are being observed. These types of studies can give insight into how drivers typically act on the road. To contribute to the literature on human driving behaviors, Explico conducted a naturalistic study of human driving perceptions and vehicle kinematics (vehicle movement) at stop-sign controlled intersections.

The Study

Drivers encounter stop sign-controlled intersections frequently while driving. The law is that drivers approaching a stop-sign controlled intersection must come to a complete stop at the appropriate location (M.V.C. §257.649). Despite this roadway regulation, it is well known that the laws are often ignored when there is no perceived hazard, i.e. when there is no other traffic that would conflict with a driver’s intended route. Many drivers reach a non-zero speed, often referred to as a “rolling stop,” rather than a law-abiding complete stop. 

The question arises: when drivers are not coming to a complete stop, what speeds are they slowing down to before proceeding into an intersection? Explico evaluated the naturalistic behaviors of drivers at stop-sign controlled intersections to assess the speeds at which drivers maneuvered through stop signs when no cross-traffic was present.  

Explico observed and analyzed two stop-sign controlled intersections. Intersection 1 was a 2-way stop, with north and southbound traffic governed by stop signs, and east and westbound traffic having the right-of-way with no stop signs.

Intersection 2 was a 4-way stop, with all lanes of travel governed by stop signs. At intersection 2, there was one lane for each direction of travel and designated left turn lanes. 

Driver behavior and vehicle maneuvers were documented using DJI drones positioned at a corner of each intersection, and video footage was taken of vehicles entering and exiting the intersections. Explico utilizes drones during their field work, typically to capture high quality up-to-date aerial photographs and videos, create 3D point clouds of the area, document any scene evidence or debris left from an incident, and visualize typical traffic patterns and driver behavior in specific areas. Some of the various ways drones are used for accident reconstruction have been analyzed by others in the field, including Evan Barreriro’s 2024 paper, Validating RealityCapture for Point Cloud Creation Using sUAS Imagery, and Neal Carter’s 2019 paper, An Optimization of Small Unmanned Aerial System (sUAS) Image Based Scanning Techniques for Mapping Accident Sites. 

Since the objective of this study was to evaluate driver behavior when there was not a need to stop for approaching traffic, only instances where there was no cross-traffic present were analyzed. To calculate the vehicle speeds, a distinct point on each vehicle was tracked during its route leading up to and through the intersection using a camera-tracking and image analysis engine software (The Pixel Farm, 2020). The lowest speed for each vehicle was identified.

The Analysis

Explico’s analysis showed that out of the 62 vehicles tracked, only four drivers came to a complete stop when there was no cross-traffic present. The lowest speeds for vehicles that did not come to a complete stop at the intersections ranged from 0.4 to 12.2 mph. 

Explico’s evaluation of naturalistic driver behaviors at stop-sign controlled intersections agreed with the notion that drivers do not always comply with roadway regulations. Generally, drivers do not come to a complete stop in the absence of cross-traffic, and of those non-stopping drivers, most slow down to less than 10 mph. 

Application

For accident reconstruction, a naturalistic study such as this could be used to identify possible pre-impact behaviors. As an example, take a scenario where a driver (Vehicle 1) does not identify any oncoming or approaching traffic and makes the decision to proceed into an intersection. Vehicle 1 enters the intersection, and suddenly they are hit by another vehicle (Vehicle 2), whose path they unknowingly crossed into. For purposes of the example, assume there is no EDR data, testimony, or other means of documentation as to how fast Vehicle 1 was traveling just prior to impact. It is expected that a driver that does not detect approaching traffic would perform similarly to the drivers observed in this study. An analyst could utilize the speeds derived from this study to help determine a reasonable and data-based range as to how fast Vehicle 1 could have been going before entering the intersection and into the path of Vehicle 2.