Semiconductor Technology News & Analysis | The Ojo-Yoshida Report
Get Ready for UX-Defined Vehicles or Not
What’s at stake:
Everyone’s mimicking Tesla, pioneer of the software-defined vehicle. Software will buttress next-generation vehicle architectures. But shouldn’t carmaker imagination reach further, to a vehicle defined by user experience?
The “software-defined vehicle” is a convenient and overused terminology when auto industry types discuss the architecture of future vehicles. The term implies a car whose functions and performance can be patched, fixed, and updated over the air. Such software can alter and improve control, for example, of a vehicle’s firmware and entertainment system.
But c’mon. Can’t we do better than that?
Podcast: How to Build a 'Human-Centered Vehicle'
Bryan Reimer, a research scientist in the MIT Center for Transportation and Logistics, is one of the key architects of the Advanced Vehicle Technology (AVT) Consortium, an academic-industry partnership.
Since 2016, the AVT group has been collecting data on the typical driver’s behavior behind the wheel.
AVT should be the right destination for car OEMs genuinely curious about how their automation features are being used by people who drive.
Given that there are, inevitably, misalignments between what such features were intended to offer to drivers and consumers’ acceptance, AVT’s goal, Reimer told The Ojo-Yoshida Report, is to “enable OEMs to build ‘a human-centered [vehicle] system rather than a system that largely ignores the human’s role.”
The idea of human-centered vehicles shouldn’t be seen as new. And yet, the auto industry has only recently begun to seriously discuss the need to develop “vehicles defined by users” or “UX-defined vehicles.”
Recommended: Tesla Deposition Exposes Disregard for Human Drivers
So, what are examples of surprising driver behavior revealed in the AVT Consortium’s data?
One example among the AVT’s discoveries is that handovers in driving tasks between a machine and a human driver must work well and take place smoothly in both directions — not just from a machine to a human driver, but also from a human driver to a machine.
We tend to worry if human drivers can react appropriately, and more importantly, soon enough, when automation suddenly tells them, “Hey, driver, now you drive. It’s your problem. I can’t handle the situation.” While it is a crucial question that still needs answers, it is equally important for carmakers to design a vehicle in such a way that it makes it easy for human drivers to take over automated driving — at the moment’s notice — without having to fight with automation.
Reimer explained that AVT data – across different vehicle models – indicates that drivers are much more proactive in managing their automation. “The vast majority of transitions, from automated driving back to manual driving, or assisted driving back to manual driving, are really the human operator saying, hey, I want to drive now – for some rhyme or reason,” said Reimer. The reasons might be that drivers are saying “I’m bored of sitting in monitoring [the road], or I’m going to get off an exit. Or, these are situations I know the automation isn’t great at, let me drive because I’m better at this.”
If this is happening, it is crucial that automakers figure out how to more effectively promote collaboration between human drivers and automation. Reimer said, “I think this is a missing point of many of the development systems today.”
As he explained, “We want the driver to become over time an expert on where the automation can help them drive more safely, while not becoming over-reliant on the automation in situations where the human’s going to be more effective.”
Bottom line? Automation needs to support and cooperate with human drivers, and automakers must figure exactly where the human-machine tipping point lies.
Listen to our chat with Bryan Reimer here.
Junko Yoshida is the editor in chief of The Ojo-Yoshida Report. She can be reached at junko@ojoyoshidareport.com.