Increasing Trust in Automated Driving
“Where’s my flying car?” drivers mutter as they wend their way through commuter traffic. “The Jetsons promised me a flying car by now.” Yet some of the same drivers have been hesitant to adopt nascent automated driving features available in today’s vehicles. OEMs are motivated to offer advanced driving features — with an emphasis on advanced — but adoption has been slower than many predicted.
What is holding drivers back from having enthusiasm for the technology? What can be done to encourage them to trust it? And how can OEMs benefit from the results? The answers, in short: Share information, demonstrate how well the technology meets expectations for a humanlike driving experience and encourage drivers to adopt the features gradually.
The autonomous driving features have to work
Plenty of us get into a car with someone whose driving acumen we distrust. Aunt Sandra imagines that she is a race car driver, cousin Ron has had a disconcerting number of fender benders, and buddy Todd persists in texting while behind the wheel. Nevertheless, we climb into the passenger seat.
But we expect more from automated driving systems. The automotive industry is working intensely to improve software-defined vehicles, including the AI and machine learning (ML) algorithms that enable them. OEMs are making ADAS more adaptive and developing better reasoning machines.
However, trust is about perception as much it is about objective reality. It takes time for new technologies to be accepted and then to become commonplace. In 1983, only 14 percent of drivers were using seat belts, despite plenty of research showing that the equipment saved lives. It took education, technological improvements and regulations to reach today’s 92 percent usage level.
It is one thing for automated driving to be safe;
it also has to be perceived as safe.
To earn trust, an ADAS-equipped vehicle needs to deliver its promised user experience by making good, safe decisions while proactively and transparently communicating with occupants. Failures scare people away.
Offer gentle introductions
OEMs can help people learn what advanced driving features can accomplish by demonstrating them in low-stress situations.
For instance, automatic parking happens at a slow speed. If drivers do not like how the vehicle performs this maneuver, it is easy for them to cancel the operation and take over. At some point, after seeing the system’s success, drivers may conclude that the auto-park feature is quicker and more accurate than they are. They are able to gain confidence gradually. That is how people learned to accept cruise control: They knew that they could resume manual control at any time with a tap on the brake pedal, and soon they learned that they did not need to.
One way to ease people on this journey is to moderate the output. Implementing an “explainer mode” in advanced driving systems can show a hesitant driver what the vehicle would do if full automation were turned on. The reasoning machine does its job, but it offers suggestions rather than taking over, much like a GPS that suggests an ideal driving route but does not compel a driver to follow it.
Another step is to tell the driver what the vehicle is doing. People may lack confidence that their ADAS will respond appropriately — Will it stop in time? — which puts them on edge. Vehicles can be designed to reassure drivers the same way humans reassure their passengers (like a parent saying, “Hang on, kids!” before making a hard turn or rapidly accelerating to pass another vehicle). With an explainer mode in place, the vehicle can say, “I see the bike on the side of the road. Don’t worry; I’ll stop if there’s a problem.”
With those guardrails, the driver eventually gains enough confidence to adopt ADAS suggestions. They can accept that the system is trained on actual human driving behavior from competent drivers and conclude, “I can enable the system on my commute, and I don’t have to worry about it.”
By offering people entry-level advanced driving capabilities, they can get used to such features before they explore more complicated options. This suggests market opportunities in scaling ADAS features from premium vehicles all the way to entry-level models. Doing so can democratize these features and help them gain broader adoption.
Use data transparency to share technology progress
ADAS features are advancing at an impressive rate, especially when it comes to safety. OEMs ensure that their vehicles comply with regulations and safety policies, though consumers notice such things only in their absence. They should tell people where the technology is — honestly and accurately — as well as where it is headed.
One virtue of AI/ML in automated driving is that these systems can learn. Data is collected and analyzed offline, using thousands of anonymized examples. Autonomous system efficacy thus improves over time, with later iterations making decisions that are safer, more efficient and so on.
OEMs should tell people when it does. Share hard numbers that make automated driving more compelling. Brag about technical improvements.
OEMs already collect and analyze a lot of data. They know how many miles people drive, how many of those miles are traveled using advanced driving features, the number of disengagements and in which scenarios the technology operates most efficiently. They can emphasize how much the automated driving features are tested — both in the real world and with synthetic miles — and back up that assertion with data.
OEMs can build on data-driven processes to improve confidence in autonomous transportation systems. We get into the car with Aunt Sandra despite her questionable driving skills because she has a driver’s license (though it might have been renewed three eye exams ago). Appeal to that reliance on external assurance of safety by considering an industry effort to adopt principles of human-based licensing processes for certification.
Factual information reassures people. It helps them make better decisions, particularly when it is time to choose a new vehicle. Setting expectations accurately can convey how powerful and safe advanced driving systems are without suggesting that the technology is perfect.
An operational experience that combines ADAS features with user experience can create confidence among drivers, encourage adoption of these features and build trust over time.