Simulation Study of Finland’s Road Network Found That 73 Percent of Crashes are Potentially Avoidable with Fully Automated Vehicles.
An Analysis of 40 Fatal Pedestrian-Driver Crashes in Finland from 2014 to 2016 Evaluated Basic, Advance, and Full Automation Scenarios with and without Driver Bypass Ability.
Finland
The Potential Impacts of Automated Vehicles on Pedestrian Safety in a Four-Season Country
Summary Information
Automated vehicles (AVs) have the potential to enhance traffic safety by eliminating human error, but the operational capability of these vehicles under different traffic and weather conditions is uncertain. This study evaluated the impact of AVs of SAE level 4 and 5 on pedestrian safety. The maximum safety impacts (i.e., the best possible situation had a human-driven car been replaced by an AV) were evaluated in three scenarios based on the AV’s ability to operate in snow, low-light, and without lane markings. In addition, a time-to-collision (TTC) analysis was made for each crash in the 2014 to 2016 dataset. TTC is a surrogate safety measure representing the time remaining before two objects would collide if they continue at their current speeds and paths.
METHODOLOGY
The researchers used data from 40 fatal crashes that included crash descriptions and other characteristics about road infrastructure and weather conditions, evaluating which could potentially have been prevented with AVs. Variables in the evaluation scenarios included weather conditions, road conditions, lightness, visibility of lane markings, estimated vehicle speed, and speed limit. The possibility of an AV to avoid the crash was also assessed using the following method:
- TTCAV = A time distance between a collision point and a point at which an AV would recognize a pedestrian stepping into a roadway.
- TTCP = The pedestrian’s time distance between a collision point and the edge of the roadway.
- If TTCAV was shorter than TTCP, the crash could have been avoided.
Three scenarios were tested: 1) Basic (Level 4 AV, cannot operate in adverse conditions or without lane markings), 2) Advanced (Level 4 AV, operates in low-light and without lane markings), and 3) Full automation (level 5 AV in all weather).
FINDINGS
According to the analysis, 28 to 73 percent of the crashes were avoidable if drivers cannot bypass the system.
- Basic scenario: 28 percent (11 of 40 crashes).
- Advanced scenario: 60 percent (24 of 40 crashes).
- Full automation scenario: 73 percent (29 of 40 crashes).
If drivers can bypass the system, the percentage of avoidable crashes decreases to 20 to 55 percent.
- Basic scenario: 20 percent (8 of 40 crashes).
- Advanced scenario: 48 percent (19 of 40 crashes).
- Full automation scenario: 55 percent (22 of 40 crashes).
Eight crashes had a TTC difference that was negative (evaluated to be unpreventable). However, four of these cases had a difference of -1.0 seconds or less, meaning that the crash fatality risk could potentially have been mitigated.
