SAE 2020 Government/Industry Meeting: Level-2 Driver Assistance Systems May Be Working
Research on production-vehicle ADAS systems suggests the technology may indeed be reducing accidents – and so far, not encouraging inattentiveness.
Advanced driver-assistance systems (ADAS) are designed to enhance safety and reduce driving stress. Based on the results of two recent studies meant to assess ADAS effectiveness, the systems may be living up to their billing. At this week’s SAE Government/Industry Meeting in Washington, DC, researchers presented largely positive results of two studies of production-vehicle ADAS that provide driver assistance at Level 2 in the SAE J3016 standard for describing automated-driving functionality.
Level 2 automated-driving functionality essentially indicates that “support” features to control speed and braking (longitudinal control) as well as lane-keeping assist (lateral control) can be operating alone or jointly, but require the driver’s constant supervision and ability to re-assume control at any time.
“Naturalistic Study of Level 2 Driving Automation Function ,” conducted by the Virginia Tech Transportation Institute and underwritten by the National Highway Traffic Safety Administration (NHTSA), observed 120 drivers during a total of 216,585 miles in ten separate vehicles. Of that mileage, slightly more than 70,000 miles were traveled with both adaptive cruise control (longitudinal control) and lane-keeping support (lateral control). Some 50,000 miles were driven with one of the two features engaged.
Each of the vehicles used in the study (below) was equipped with an industry-standard data-acquisition package to record driver action, driving environment and vehicle data.
- 2017 Audi Q7 Premium Plus 3.0 TFSI Quattro (Driver Assistance Package)
- 2015 Infiniti Q50 3.7 AWD Premium
- 2016 Mercedes-Benz E350 (Driver Assistance Package)
- 2015 Tesla Model S P90D AWD (Autopilot Convenience software version 8.0)
- 2016 Volvo XC90 T6 AWD R with Design and Convenience Packages
“This project is the first study sponsored by NHTSA to review driver interaction with vehicles that include lateral and longitudinal automated features in real-world settings,” said the report, which was intended to provide insight into four areas of ADAS functionality: system performance (during unscripted, on-road driving); driver-system interaction; driver performance; and driver engagement.
The key findings, according to Virginia Tech’s Sheldon Russell:
- Of the 71 recorded safety-critical events (SCE’s) in the study, “No statistical relationships were observed between SCE rates and feature activation level,” seemingly indicating drivers were not over-reliant on the ADAS capabilities. And for none of the SCE’s had the automated systems requested the driver to intervene (known as an RTI) – which seems to indicate, the study said, that when a critical situation presented itself, drivers were paying sufficient attention.
- Drivers did not evidence a statistically significant higher degree of inattention when using one or both Level-2 support features.
- “The analysis of environmental factors observed indicated that, in the majority of cases, participants were operating the driving automation system-equipped vehicles in a manner consistent with manufacturers’ intended use.
- In numbers that will be important to researchers studying so-called Level 2-3 “handoff,” the NHTSA study recorded mean reaction times to RTIs at 0.94 sec., Russell reported. Mean time to intervention was 1.79 seconds.
- “When operating the vehicles above 40 mph, drivers typically drove with both features active. Drivers were less likely to activate the systems in heavy traffic, on non-interstate roads, and in rainy weather conditions.”
- “Overall, drivers appeared to trust the driving automation systems, and were comfortable using them,” although “Driver interviews and trust ratings gathered at specific intervals during the 4-week participation period suggested that there was little change in trust in the lateral systems, although summarized comments also indicated that there were situations reported where the lateral systems did not function as expected.”
IIHS study: reduced crash claims. Maybe.
At the 2020 SAE Government/Industry Meeting, David Zuby, executive vice president and chief research officer at the Insurance Institute for Highway Safety (IIHS), said that although the data remain in relative infancy, the IIHS’s recent study of insurance-claim frequency for vehicles fitted with Level-2 ADAS capabilities indicates there is a possibility that the systems are having a positive effect.
Zuby said the Level-2 systems in the vehicles studied might – emphasis on “might” – be associated with a lower frequency of crash claims against insurance coverage. Specifically, IIHS is estimating collision coverage claims to be reduced anywhere from 1.2% to 13% and property-damage liability to be in a potential range of 1% greater to 15% less. He added that rear automated braking is demonstrating high effectiveness in reducing claims.
Next steps according to Zuby include an effort to conduct a Euro NCAP evaluation of Level-2 systems with features not available in Europe and to jointly release the results. He also noted that IIHS has determined the wide variation in how auto companies name their systems has been found to have a direct effect on how much trust users have in the system’s ability to handle complex tasks. The effect is most notable, Zuby said, for Tesla’s Auto Pilot, for which participants in a task survey disproportionately indicated they expected the Tesla system to be more capable of handling complex situations.