What happened
David Kidd and Watson Spivey of the Insurance Institute for Highway Safety (IIHS) tested pedestrian automatic emergency braking at night on a track. They used a 2023 Mazda CX-5, a 2023 Honda CR-V and a 2023 Subaru Forester. An adult-sized mannequin crossed the road as each car approached at 40 km/h. It wore black clothing, black with a reflective jacket, black with reflective strips on its limbs or white clothing. Light towers gave 0, 10 or 20 lux in the crosswalk. The cars used low beams throughout and high beams with the mannequin in black and no roadway lighting. A collision occurred in 2% of trials with the Forester, 84% with the CR-V and 88% with the CX-5. IIHS reports that the Forester's one collision came with reflective strips under 10 lux and that it still slowed by more than 80%. The CR-V and CX-5 did not slow at all for the reflective strips. Traffic Injury Prevention published the study in November 2024, and IIHS described it on 9 January 2025. IIHS lists EyeSight as standard on the Forester. No crash avoided on public roads is reported.
Sources
-
Assessment report, by authors with no tie to the system.
Kidd, D. G. and Spivey, W. A case study of nighttime pedestrian automatic emergency braking performance under different roadway lighting and pedestrian clothing conditions. Traffic Injury Prevention, November 2024. Abstract in the IIHS research bibliography.
A collision occurred in 84% of trials with the CR-V, 88% of trials with the CX-5, and 2% of trials with the Forester.
-
Assessment report, by authors with no tie to the system.
Insurance Institute for Highway Safety. High-visibility clothing may thwart pedestrian crash prevention sensors. News release, 9 January 2025.
In contrast to the two other vehicles, the Subaru came to a complete stop without hitting the pedestrian dummy in every trial except one run in which the dummy was wearing clothing with reflective strips and the roadway was illuminated to 10 lux. Even in that run, it slowed by more than 80%.
-
Assessment report, by authors with no tie to the system. Context or contrary evidence.
Insurance Institute for Highway Safety. 2023 Subaru Forester 4-door SUV, front crash prevention: pedestrian (night). Vehicle ratings page, read on 1 November 2026. Cited for the name of the standard system.
Front crash prevention: pedestrian (night) System details standard EyeSight
Details
AI system
- Workflow
-
An adult-sized mannequin crossed the road as the car approached at 40 km/h. The trials varied the mannequin's clothing and the roadway lighting at 0, 10 or 20 lux. The car used its low beams in every combination and its high beams with the mannequin in black and no roadway lighting.
Follow-up
- Conditions for replication
-
The pedestrian crosses the car's path, since every trial used a perpendicular crossing at 40 km/h. The pedestrian does not wear reflective strips on the limbs, since the Forester's one collision came with such strips under 10 lux.
Entry details
- Sector
-
Road vehicle crash avoidance.
- Purpose
-
Subaru built EyeSight to warn the driver and brake automatically when a collision with a pedestrian ahead is likely.
31 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). EyeSight on a Forester hit a crossing dummy at night in 2% of trials (AIBD-0978). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-0978.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)