Speeding Up Consumer Acceptance of Autonomous Vehicles
Last week, Intel released details of a small-scale, qualitative study it carried out recently to explore the theme of human trust in the acceptance of self-driving cars. The company believes this is the first study published about the subject.
The technology for fully autonomous cars is close to being ready, but factors such as regulation mean that the concept is unlikely to become mainstream quickly. Adoption won’t go well if consumers don’t trust the technology. Many people believe that driving requires human understanding, so it’s reasonable to expect them to have a low level of trust in self-driving technology at first, especially owing to its complexity. A recent survey by the American Automobile Association found that 75 percent of Americans would be scared to ride in a self-driving car.
In carrying out its study, Intel worked on a vehicle’s human-machine interface to ensure screens showed what the car was doing, and developed a voice system to keep passengers informed about what the actions the car would perform next and to receive instructions from passengers.
Intel briefed the 10 passengers in the study group at a high level about the technology in the car and, in broad terms, how it works. The passengers then called the vehicle using a smartphone app and went for a pre-planned ride around a test circuit. They were asked to make changes to the route during the ride, such as stopping for coffee. For safety reasons, there was a person in the driver’s seat, who could take control of the vehicle if necessary, but they weren’t allowed to speak or take part in the experiment.
The study highlighted seven interesting areas of tension in the participants’ attitudes that had an impact on their confidence in self-driving cars. Firstly, although they initially had a low level of trust in the technology, they understood that most of the dangers on the road come from other drivers. This means that if they could trust a self-driving car, they would acknowledge that it ought to be safer. For example, participants recognised that autonomous cars have a 360-degree vision of its surroundings, whereas a driver only looks in one direction at a time.
Secondly, participants liked the personal space that a self-driving car offers, with the opportunity to work, make phone calls or even sleep. Busy parents thought that it could be a productivity gain if the car could drop the kids off at football training, while they do other things. However, there was also unease about what happens if something goes wrong — for example, how to get help. In a ride-sharing context, people also expressed concerns. For example, they said that they enjoy chatting to an Uber driver and that it’s often useful to be able to fine-tune the details of the drop-off point or route by talking to the driver.
The passengers in the experiment also appreciated the car’s ability to talk to them. But they found that, once they were comfortable with the vehicle’s capabilities, they wanted it to stop talking so much and let them concentrate on other things. There was also an expectation that if the car could talk to them, then they should be able to speak to the car to give it instructions and ask questions. Voice is a natural element of the user interface and Amazon’s Echo digital assistant has clearly raised expectations in this area.
There was an interesting discussion about legacy controls. Participants found it spooky to see the steering wheel turning by itself and questioned if it was needed at all in a fully autonomous car. But if all the legacy controls are removed, what mechanisms should there be? What’s the new design and user interface language for controlling self-driving vehicles?
Furthermore, people found that they were interested in how the car works, and liked the initial briefing as well as the display in the vehicle showing all its functions, but most of the time, they preferred to look out of the window and watch the road. The implications of this finding are quite important for the sales process of autonomous cars, at least in the early years.
Lastly, the vehicles in the study were set up to be strict in following rules about stopping at stop signs, speeding and others. But this isn’t always how humans drive. So, at one level, people found the driving style in the experiment rather unnatural. However, this was offset by a view that driving should actually be better and safer that way, and this highlighted the need to let go of preconceptions about driving behaviours.
This was a fairly small study to see how useful its findings might be. Intel is encouraged by the results and says it will explore further aspects including a more diverse set of routes, different urban and rural environments, reactions of children and millennials who are more comfortable with technology, how things play out when there are multiple passengers in a car, regional differences and more.
I believe this is a very worthwhile project that raises a lot of important questions that must be addressed before self-driving cars come to market in a big way. Indeed, the area is so significant that it’s quite remarkable the technology has been allowed to develop as far as it has without someone doing this type of research before. Intel’s effort to start the process is commendable.
Although this is an important start, it’s still only part of the picture. Building consumer confidence in self-driving technology doesn’t just involve the occupants of a car; there are many aspects of trust relating to how people outside a vehicle interact with it. These include the subtle signals a driver sees when a pedestrian wants to cross a road, and those the pedestrian detects when the car slows to let them walk across. They also include the different forms of communication between drivers to allow (or obstruct) cars pulling into traffic, changing lanes, overtaking and so on. Intel is aware of these issues, but isn’t focussing on them yet.
The challenge of autonomous car design is significant, but even greater is how the vehicles will interact with the broader environment — everything from pedestrians and other cars, to parking spaces, traffic management infrastructure and the many components of a smart city. All these areas will have considerable bearing on consumer trust and adoption, not to mention investment and business models. The industry is making great strides, but truly autonomous vehicles and all-important consumer trust will require far more than radar, lidar and 360-degree cameras. This is just the start.