Drivers of vehicles with advanced driver assistance systems (ADAS) engaged are 50- to 80-percent more likely to engage in “visual and/or manual visual tasks,” leaving them unprepared to quickly respond to an emergency or automation failure.
“New advances in vehicle technologies and rapid deployment of these advanced systems in vehicles already on the road leave unanswered questions about their impact on driver behavior,” stated a report released in December by the AAA Foundation for Traffic Safety. “Specifically, as drivers are removed from the driving task for periods of time, their ability to retake control of the vehicle in the event of automation failure becomes questionable.
“Numerous studies have indicated that the combination of adaptive cruise control and active steering results in longer response times, increased hard braking and more collisions or near collisions when automation fails, as compared to manual driving,” the report continued.
“A critical issue relative to behavioral adaptation and specific to ADAS use is that drivers may divert their attention away from the driving task to engage in secondary, non-driving related tasks. This could have dire consequences should the ADAS fail or become inactive, meaning the driver needs to quickly redirect his/her attention back to the road and resume manual control of the vehicle.”
The study is titled Understanding the Impact of Technology: Do Advanced Driver Assistance and Semi-Automated Vehicle Systems Lead to Improper Driving Behavior? It is based on two naturalistic driving studies conducted on Virginia roads. One study tracked 50 drivers in their own vehicles for 12 months. The second involved 120 drivers who were assigned specially equipped vehicles for four-week periods.
The investigators suggested a three-phase model of driver interaction with automated systems.
“The novelty phase includes learning and testing the ADAS in real time to understand limitations and capabilities. Trust in the ADAS may be lacking during this phase as the driver has little to no experience with the automation features,” they wrote.
In the post-novelty operational phase, “behavioral adaptation may begin to occur, and over-reliance and over-trust in the automation features may develop. The post-novelty phase is also when driver awareness of the limitations of the ADAS features ideally evolves to avoid an increase in safety-critical events associated with risky behaviors, such as distracted driving.”
Drivers then move to the experienced user phase, “wherein overreliance and work under-load may manifest in the form of drowsiness or inattention.”
The report found that drivers eventually become more comfortable relying upon automation.
“Evolutionarily speaking, humans adapt as they learn; thus, when presented with new technologies, the driver will learn and adapt to using these technologies over time and under changing conditions. Behavioral adaptation can implicate psychological factors, such as motivation, emotions, personality traits, and decision-making that inform how an individual will respond to the change.”
“When operating under a vehicle capable of automated steering and maintaining speed and headway, engagement in secondary tasks increased. The length of drivers’ off-road glances also increased, and they diverted their attention from the forward roadway more than 30% of the time,” the authors explained and cited similar results found in a previous research project.
“The authors suggested that such results were due to behavioral adaptation to the system. A recent survey of owners of vehicles equipped with ADAS found that approximately 30% indicated they were comfortable engaging in non-driving-related tasks when ACC was activated.”
In addition to adaptation, automation may make driving too easy for operators.
“The use of automation features that remove sources of workload from the driver may create issues associated with mental under-load. Reducing mental workload may be helpful, to a point. However, reducing it too much has been shown to negatively impact driver performance. Under-load occurs when the task demands are relatively low and the driver does not need to mobilize too many cognitive resources, or mental effort, to maintain performance.”
Previous studies “on the impact of automation on workload and driver performance… found higher levels of automation were associated with significantly lower workload and poorer response times (1- 1.5 seconds slower) to unexpected automation failure. Under-load has also been linked to monotony and fatigue.”
The AAA researchers referred to driving simulator studies that subjected drivers to “fatigue associated with monotonous, low-demand driving.”
That study “found drivers were slower to respond to an unexpected hazard and were more likely to collide with the hazard. Thus, while there are safety benefits associated with ADAS, there may also be associated under-load and fatigue-related declines in driver performance that need to be investigated.”
The report recommends comprehensive training for drivers acquiring ADAS-equipped vehicles and standardization of operations between ADAS systems.
“Every ADAS-equipped vehicle currently operates differently across functionality, ADAS activation, capabilities, and operating speeds, resulting in a learning curve each time a driver uses a different system. As such, standardization to a greater degree may also be beneficial in that it will limit the impact of novelty symptoms,” it explained.
“Finally, many of these advanced safety systems have limitations and specific conditions under which they are designed to operate optimally. Unfortunately, owners of vehicles equipped with these advanced safety systems typically lack awareness of the key limitations of these technologies.”
For example, the report explained, a recent survey “found more than three-quarters of owners of vehicles with blind spot monitoring systems had misconceptions about its function or were unsure of the system limitations. One-third of respondents whose vehicles had autonomous emergency braking systems did not realize the sensors and cameras on which the system relies could be blocked or obstructed by dirt, snow, or ice.”
The AAA Foundation recommended that the transportation industry learn from previous industries that have incorporated automation into their procedures.
“Decades of prior research in other domains such nuclear power and commercial flight have highlighted the potential consequences of partial automation. Such consequences include operator confusion over control authority, reduced workload resulting in boredom that could lead to drowsiness or fatigue or an over-reliance on a system that results in reduced attention to the primary task.
“One important difference between those industries and the current study is the level of training and practice received. Commercial aviators or nuclear power plant operators, for example, receive thousands of hours of training and practice prior to certification as operators. Conversely, users of automotive technology only receive a minimal set of written and verbal instructions prior to use on public roadways in live traffic.”
The report recommended that drivers undergo extensive training, even after they have spent considerable time in operation, because drivers with more ADAS experience may actually become more dangerous.
In “the experienced user phase the driver has used the system repeatedly, understands the system well (e.g., has “tested” the system and perhaps even over-relies on its limited capabilities), and uses the system frequently. This phase is where the potential for work under-load could manifest itself.”
