Using behavioral science to accelerate consumer adoption in automated mobility
What would it take for you to be comfortable riding in a self-driving car? How do you think you’d react seeing the wheel turn without human hands, navigating through urban traffic better than you?
It’s an area Aptiv has been studying for several years now – the psychology of self-driving cars. We’ve been observing human reactions, emotions, likes and dislikes when they interact with a self-driving car for the first time.
Aptiv knows consumers need to trust the technology before wide-spread commercialization takes off. Additionally, people need to feel comfortable when they ride in an automated vehicle – and we’ve been working on that too.
Aptiv has been digging deep into consumer relationships with automation technology, key behavior shifts and the dynamics of building trust. Here’s a behind the scenes look.
A System in Transition
Today, we are very much designing for a system in transition. It starts with consumers gaining first-hand experiences with autonomous vehicles. At roughly 20+ automated rides, they begin to internally adjust and transition their human-robot trust. They gain experiential knowledge of an automated vehicle’s capabilities, limitations, behaviors, and driving style “personality.”
As new advances in automation technology occur (e.g., expanding operational domains, better sensor resolutions), they cause consumers to update their mental models of how an automated vehicle works and how accurately it perceives situational risks.
That’s why, at Aptiv, we adopt a phased transition-friendly mindset in our product philosophy as we take consumers through the various stages of onboarding and record their personal journey of establishing trust with the autonomous vehicle.
Future automated ride sharing passenger’s experiences have been compared to that of a commercial airline passenger. It’s true that we’ll all eventually become seasoned riders (much like in aviation), but it’s going to take a calibrated approach to get there. We are not seasoned riders in vehicle autonomy just yet.
At Aptiv, key foundational interactions are introduced in the initial phases of product deployment and certain elements are slowly phased out or refined as consumers transition in their subjective assessment of trust.
The establishment of trust, maintenance of trust, and in certain situations, the regaining of trust revolve in a continuous loop. Ground-truth reliability and actual performance of an autonomous vehicle might only be a few of the many characteristics affecting a person’s trust and notion of safety.
Research shows that transparency in decision-making and perceived understanding of an autonomous maneuver plays an important part in the overall calibration of trust. For example, the use of forewarnings helps humans adjust their trust accordingly. In one of our on-road sessions, participants experienced a set of scripted autonomous vehicle limitations (e.g., sudden hard braking in the middle of a turning maneuver). Participants who were not given any indication or prior warnings of upcoming situational anomalies reported a loss in trust (an average three points decrease on a scale of one to 10).
They also perceived a high level of environmental risk or system malfunction for each scenario. Participants who were made aware of possible upcoming system limitations and/or situational anomalies calibrated their trust accordingly and reported lesser deviations in trust (an average one point decrease on a scale of one to 10).
Key trust-building interactions also consider the cognitive and affective experiences of a passenger, especially when the dynamics of automation are involved. While situational awareness is a well-researched field for manual driving, in a shared mobility ecosystem we need to take a closer look at another phenomenon called “passenger situational awareness.”
While not responsible for actual latitudinal and longitudinal control of a vehicle, passengers take on certain supervisory functions in a ride, subconsciously maintaining an internal orientation and a high-level status of their journey. Put another way, passengers regularly look outside their windows to orient themselves in an environmental context.
Our on-road sessions repeatedly show that passengers heavily rely on vehicle speed variations and the visceral experience of vehicle motion to intrinsically assess certain driving contexts and states that might need their attention.
There are two main challenges when we perform any kind of consumer research in the field of emerging technologies. First, it is extremely hard for participants to simply “imagine” the future and their expected behaviors around such systems. It is comparable to asking someone who has never seen an iPhone about how would they use such a device.
Second, what participants “say” about their beliefs and expectations can be drastically different than what they actually “do” or “feel.” We have to actively frame participants in a future context in which we chip away at any superficial responses.
For example, certain initial views on autonomous systems can be bifurcated: either it’s a perfect system or an unreliable system. In later sessions, however, we have found that participants are eager to develop a human-robot team relationship and certain system behaviors no longer erode trust.Here are some methods in action:
- Co-creation exercises: In this scenario we ask participants to design a given solution and actively participate in the process. While such exercises might seem fun and playful they often offer an excellent reflection into the consumer’s mind and how that person frames a given problem.
- Simulation with a twist: Once we actively place participants in a future context, we use simulation to run through a set of repeated controlled road scenarios. Driving simulators are used for driver-focused research and training all around the world. We additionally run simulated sessions that focus on the passenger and collect data on situational awareness, trust, line-of-sight, and a range of other passenger behaviors in an automated context.
- On-road sessions: While a simulator setup allows for road scenario controllability and eases data collection, on-road automated sessions bring us much closer to the dynamic road context and true passenger reactions. Watch a video of one of our road sessions in action here.
It is highly likely that before reaching 100 percent market share, autonomous vehicles and shared mobility will follow Rogers bell-shaped adoption curve. The initial customer segment of innovators and early adopters are characterized as risk-takers, technology-focused, and experimental in their adoption approach.
Leveraging initial building blocks of trust and an understanding of consumer behaviors within this segment will accelerate our understanding of future segments of majority adopters.
This research helps us improve our design, functionality and user experience in our autonomous mobility platform. When combined with our commercial operation in Las Vegas, where riders on the Lyft network can hail an Aptiv self-driving car – we’ve already delivered 5,000 rides with a rating of 4.96 / 5.0 – Aptiv is ahead of the competition.