AAA testing found that not all automated driving systems detect cyclists at intersections.
AAA
Testing by the American Automobile Association—AAA—has found that not all cars sporting semi-autonomous driving technology spot cyclists on four-way crossings.
“For a cyclist crossing the travel lane of the test vehicle, a collision occurred for 5 out of 15 test runs, or 33% of the time,” states the AAA.
Closed course testing of a 2021 Subaru Forester using “EyeSight” driving assistance found that the proprietary technology failed to spot a dummy cyclist on a crossing. The Subaru car failed to provide any detection alert or initiate any braking in response to the cyclist on the crossing.
Subaru spokesman Dominick Infante said the automaker had improved its EyeSight system for the 2022 model year.
Semi-autonomous
Advanced driver-assistance systems—or ADAs—are electronic technologies that assist drivers in driving and parking functions.
“Our testing demonstrates spotty performance is the norm rather than the exception [for ADAs],” said AAA’s director of automotive engineering Greg Brannon.
While the Subaru failed the test, the dummy cyclist was spotted by a 2021 Hyundai Santa Fe with “Highway Driving Assist” and a 2020 Tesla Model 3 equipped with “Autopilot.”
That some ADAs work while others don’t will be scant comfort to cyclists and pedestrians. And motorists, too, shouldn’t rely on the technology because the AAA’s testing found that a head-on collision occurred during all 15 test runs for an oncoming motor vehicle within the travel lane.
The dummy cyclist used was the official Euro NCAP cyclist, representing an average European adult male on a standard bike. The dummy—disconcertingly called a “target” in the jargon—includes rotating wheels with realistic micro-Doppler spread and other signal characteristics with respect to radar, lidar, camera, and infrared sensors.
A car with Luminar LiDAR based Proactive Safety pre-collision braking crash avoidance technology (R) … [+]
AFP via Getty Images
All three cars—piloted by test drivers backed up with monitoring devices—successfully passed the dummy cyclist in the AAA testing, giving the dummy cyclist adequate room when overtaking.
Thirty-five states and the District of Columbia have laws that require motorists to leave at least 3 feet of lateral clearance when passing cyclists.
“As cyclists are afforded the same privileges to nonlimited access roadways as motor vehicles, it is important to understand ADA system performance in the context of interactions with [cyclists],” states the AAA.
“For a cyclist crossing the travel lane of the test vehicle, two out of three test vehicles detected the cyclist and initiated emergency braking in response,” found the AAA.
“While this is a promising result, the observation that one of three test vehicles failed to detect the cyclist target for any test runs once again suggest that edge-case emergency scenarios are more challenging for ADA systems in general.”
The AAA recognized that ADA systems “continue to improve” but that they are “not capable of sustained vehicle operation without constant driver supervision” and therefore that the driver “maintain situational awareness at all times.”
The auto organization recommends that carmakers “focus on refining ADA system performance in edge-case scenarios.”
