Volvo Leads the Shift to ‘Driver Understanding’
Volvo research might indicate we’ve been approaching ADAS all wrong.
In recent weeks, Volvo started teasing out some of the details of its all-electric replacement for the XC90 SUV, dubbed the EX90. As expected from the Swedish brand, safety is front and center. While we’ve known for more than a year that the vehicle now branded as EX90 would feature a more robust exterior sensor suite, including Luminar-developed lidar, what really caught my attention is the driver-understanding system.
In 2022, driver-monitoring systems (DMSs) aren’t particularly unique anymore. General Motors included an infrared camera to ensure the driver was watching the road when it debuted “hands-free” Super Cruise in 2017. Similar DMSs now are found on vehicles from many manufacturers, including Ford, Mercedes-Benz, BMW, Nissan and Toyota, with more to come in 2023 and beyond. The DMS that is in the market today looks only at the driver’s eye gaze and head position to detect visual distractions such as looking at a phone, infotainment screen or passenger.
Volvo’s announcement highlighted that the EX90’s DMS also will be looking for cognitive distractions, that is where the driver’s eyes are on the road, but their mind might be elsewhere. We’ve all experienced situations, especially on a long drive, where you suddenly realize you can’t remember the last several minutes of the trip. That can be dangerous.
But according to Volvo cognitive scientist Mikael Ljung-Aust, it might not always be necessary for the vehicle to intervene in such instances. Ljung-Aust and his colleagues have been conducting a series of experiments over the years to understand driver behavior and responses to various inputs. It turns out that much of the driver’s task consists of automated responses: as drivers become more experienced, they react to most situations without consciously thinking about it, much as we don’t actually think about the motions of walking while doing something else.
Naturalistic driving research has shown that over time, drivers’ eye motions and viewing patterns are surprisingly consistent; within a bounding box of behavior, drivers can reliably respond to virtually all driving scenarios. So, in deciding when to alert the driver or otherwise intervene, Volvo’s driver-understanding system instead looks for eye and head motions that go outside of expected behavior.
Interestingly, the goal with driver understanding is not necessarily to have the system intervene earlier or more often, but later and less frequently. Numerous studies have shown that advanced driver-assistance systems (ADAS) that are overly persistent or intrusive annoy drivers to the point that they frequently disable the system. Lane-keeping assist systems are a common source of complaints for drivers — and often are switched off.
The situations in which a driver is unable to recover from an error actually are rare, which is consistent with mileage-per-crash statistics: there are more than 3.2 trillion vehicle miles traveled annually in the U.S and about 6.5 million crashes, or approximately one every 30 years for the average driver. However, if ADAS features are disabled, they certainly can’t help in the rare instances when they are needed.
Ljung-Aust said that the goal of driver understanding is to “arbitrate if you need help” and create a better user experience. If ADAS is always there when needed but stays out of the way the rest of the time, drivers are less likely to disable it and safety should improve. I can’t wait to try it to see how well it actually works.