This article also appears in
Subscribe now »
Toyota's Platform 2.0  sensor demonstartion vehicle. (Toyota)

As early autonomy questions roil the industry, Toyota hews to driver-enhancement philosophy

At a recent industry conference, Toyota’s lead for autonomous driving research said the company will continue to focus on a strategy of driver enhancement on the path to high-level autonomous functionality.

Speaking at the June TU-Automotive Detroit conference in Michigan, as part of an aptly titled panel “Getting a Guardian Angel in the Car,” Ryan Eustice, senior vice president of automated driving at Toyota Research Institute (TRI), noted that the focus of Toyota’s “Guardian” automated system will continue to place the driver at the center of the technology.

“When you look at the SAE zero-to-five automation scale, the way we are managing the Guardian system, it doesn’t really map onto the SAE definition because the definitions are about removing or relieving the driver from the task, trying to offer the driver convenience,” Eustice said.

“We’re not arguing for a change in the [SAE autonomy] definitions from the automation perspective, we'd like to have clear adherence to those definitions,” Eustice explained. “I think there's a lot of loose terminology that gets used both within the industry and within the media, so it kind of confuses the landscape. For consumers, do they have the right expectations of what the system can do, and what are the limits of that system?”

One of two paths Toyota is currently pursuing in its autonomous driving research — Guardian and “Chauffer” — Guardian is a foundation philosophy of safety-focused automation in which the human driver maintains vehicle control and the automated-driving system monitors for potential accident or loss-of-control situations, intervening when necessary.

All about stimulation
“Humans drive cars today and there's a sufficient level of stimulation engagement to perform the driving task,” Eustice explained. “One of the challenges you have in these higher levels of automation – Level 2, Level 3, where the human is expected to play some kind of role in the overall system – is how do you keep the human engaged at a level where they provide the level of vigilance required?”

The lack of consistent stimulation appears to have been one issue in the March incident in Arizona in which a Uber autonomous test vehicle struck and killed a pedestrian despite the presence of a human “backup” driver.

“The challenge is in transitions,” Eustice said. “Are they able to safely handle the transition to them being the backup? With Guardian, it's just a very different thinking about how we use that exact same technology that goes into automated driving, where the system works with the human as a partner.”

Eustice noted that Toyota’s direction for Guardian is related to how similar automation is applied today in aerospace. “Think about a fighter pilot. They actually work in conjunction with the flight control systems, to keep the vehicle within a region of stability. Using those similar ideas with a ground vehicle, we call it parallel autonomy.”

“It's a much harder problem on the ground,” he conceded, “because it's not only vehicle dynamics that define this region of stability and safety, but perception and prediction. You need everything that goes into the software and car to understand where the road is, where other cars and pedestrians are, what future interactions might look like. All these things come in to inform how the system works with the human to assess risk, and how we try to moderate that risk.”

One of the main issues, Eustice continued, is how quickly drivers can over-rely on the technology. “It's challenging because humans can very quickly build up an ‘over-trust’ in these systems and begin to mentally zone out,” he explained. “With Guardian we imagine really keeping the driver engaged all the time and the system working with you to make you a superhuman driver.” Continue reading »