Skip to content

Driving a human-machine collaboration | MIT News

Driving a human-machine collaboration | MIT News

Here is the scenario: the driver falls asleep at the wheel. But their car is equipped with a dashboard camera that detects the driver’s eye condition, activating a security system that immediately directs the car to a safe stop.

This is not just an idea on the drawing board. The system, called Guardian, is being improved at the Toyota Research Institute (TRI), where MIT professor John Leonard is helping direct the group’s work, while on leave from MIT. At the MIT Mobility Forum on Friday, Leonard and TRI President, Avinash Balachandran, gave an overview of their work.

Presenters provided ideas on multiple levels around automation and leadership. Leonard and Balachandran discussed specific systems for TRI while also suggesting that – after years of touting the possibility of fully automated vehicles – a more realistic possibility might be to deploy technology that helps drivers, without replacing them.

Anticipating this kind of future might make sense, “rather than thinking of an all-or-nothing world, in which there is complete automation or exclusive human control,” said Leonard, professor of mechanical and ocean engineering in the Department of Mechanical Engineering. “Is there some kind of future where there is a better mix of the best of man and machine?”

Leonard guides the audience through the steps the Guardian system takes, from realizing the driver’s unconsciousness, to taking control of the vehicle, and eventually reaching the point – given the alert driver – the system no longer operates the vehicle itself.

“Now the guardian has intervened; the system is working on the screen,” Leonard said, explaining that “now at some point you will be offered to return it to you.” He added, “Our vision is that we can use some advanced technology from automation to further assist drivers.”

The presentation, titled “Human-Centered Leadership Research at the Toyota Research Institute,” was held online Friday in front of an audience of more than 200 people. The MIT Mobility Forum is a weekly series of talks covering all aspects of transportation. It is presented by the MIT Mobility Initiative, an institute-wide effort to spur innovation and strategic thinking around developing sustainable, accessible, and safe transportation.

In his notes, Leonard positioned himself as an advocate for the auto industry’s efforts to develop high-tech vehicles, and as a realist about the way these technologies are likely to arrive in everyday life.

“I was a bit skeptical about full automation in terms of timelines. [It] It’s going to take longer to get this kind of ubiquitous automated taxi, since, you know, today’s teen will never need a driver’s license or ever need a real human Uber driver, because all the cars will drive themselves independently,” Leonard emphasized.

It likely suggested, “We can expect multiple paths and a longer time horizon than many would have predicted to start implementing this.”

However, both Leonard and Balachandran highlight the development of current technologies. Balachandran showed the public a demonstration video of the TRI system that can help prevent collisions when a vehicle has drifted dangerously and is about to leave the road due to inattention, poor conditions or other factors.

This system is intended to “human amplify, for better overall system performance. This includes things like safety, but also other elements we believe [make driving] Balachandran said, calling this type of human-machine collaboration a “HCID” for intelligent human-centered driving.

In the TRI video about handling the drift, the car navigates a winding test track with obstacles placed on the circuit as well. If the vehicle is about to drift off the road, the system can react and quickly control the vehicle more effectively than many drivers would do in a time of crisis.

“This gives us access to a capability that is beyond what the average driver could do,” Balachandran said.

The event was hosted by Jinhua Zhao, associate professor of transportation and city planning in the MIT Department of Urban Studies and Planning, and director of the MIT Mobility Initiative.

At the start of the Q&A portion of the event, Zhao inquired about the relative challenges of the everyday safety issue on city streets, rather than improving the handling capabilities of the high-end vehicles found in Forumula One drivers, for example.

“When a lot of people think about uncertainty, they think about uncertainty about the perception of the environment and predicting factors,” Balachandran said. “It is true, [it] It is a very difficult problem.”

However, Balachandran added, the car’s ability to perform object recognition is “a much more specific kind of problem. … dynamic uncertainty [of driving maneuvers] It is actually much higher.” Balachandran noted that in every area “it’s relatively easier to reach 90 percent capacity. It’s very hard to get 90 to 100 percent.”

For his part, Leonard emphasized that the need to improve traffic safety is an urgent need, given developments such as the increase in traffic deaths per mile driven since the start of the COVID-19 pandemic.

“The sudden increase in traffic accidents and deaths since Covid is really heartbreaking,” Leonard said. “I am really disappointed…because we haven’t found a way to use this advanced technology to make cars safer. And so I think the human piece is important.”

And while it may be attractive to see fully automated vehicles cruising the streets without any problems, it seems very likely that hundreds of millions of drivers will remain on the roads for the foreseeable future. So there is an alternative vision that helps these people lead as effectively as possible.

“I strongly believe, personally, that we want to use this technology to help human drivers,” Leonard said.

Source link