The Society of Automotive Engineers (SAE) J3016 standard has defined six levels of automation ranging from level zero (no control automation, only warnings, e.g. lane departure warning systems) to level five automation (fully automated vehicles under all conditions). Currently no production level four or five vehicles exist yet, but automakers have begun to offer active driving systems at level two automation in some newer model year vehicles.
However, despite their increasingly commercial availability, some safety advocates and others have raised concerns about the reliability of these level two active driving systems for two main reasons. First, drivers are supposed to remain fully attentive and in control of the vehicles even when the automation systems are engaged. Yet, these systems might lull drivers into a false sense of security and cause drivers to become inattentive. Second, the actual efficacy of level two automation systems under real world driving conditions is not entirely clear. Real world driving conditions may present these systems with substantial challenges such as faded lane markings, challenging traffic conditions, stopped vehicles and other situations that could cause these systems to fail.
To help answer some of the questions around performance of level two active driving systems, a research team from the American Automobile Association (AAA) Foundation for Traffic Safety tested level two systems under real-world highway driving conditions, and used a closed course to test in some uncommon but realistic situations. The team selected five different 2019 or 2020 model year cars, each from a different manufacturer and equipped with a different form of level two automation. Second, the team equipped each vehicle with cameras and various on-board sensors to record acceleration, deceleration, speed and other parameters during testing. For the closed course testing, the team drove all five vehicles on a test track and presented the vehicles with various staged challenges, such as stopped vehicles, using a representative but strikable target vehicle. Finally, the team drove each vehicle from San Francisco to Los Angeles along freeways and recorded vehicle behavior during these real-world test runs.
• Currently available level two active driving systems are not capable of sustained vehicle operation without constant driver supervision.
• During naturalistic evaluations, an adverse event (primarily problems with lane-keeping) was observed on average, once every 8 miles
• On fresh pavement with well-defined lane markers, all systems consistently traversed the testing lane with some lateral offsets relative to the lane center.
• Evaluated systems made contact with a staged disabled vehicle 66 percent of the time on a test track.