There’s a quiet war taking place behind the wheel of today’s vehicles between drivers and semi-autonomous technology designed to keep them from crashing, or worse.

Often, drivers annoyed when automatic lane centering turns the wheel while they’re holding it, or self-driving systems want to steer in a certain direction at odds with the driver’s wishes, simply turn off the devices.

Indeed, it comes down to a basic relationship problem between driver and technology and new research by the University of Michigan released Thursday aims to cool the conflict.

“We want our humans and the automation to work as team together and if you constantly have the automation pushing the driver, you’re just going to turn it off,” said Hannah Baez, a PhD student in robotics at the University of Michigan’s School of Engineering, in an interview.

Working with Prof. Brent Gillespie with funding and collaboration with the Toyota Research Institute , Baez and team created a simulator featuring a steering wheel with a set of bladders and bubbles that inflate or deflate to inform the driver an automated operation’s intent.

If the driver squeezes the correct bubble or bladder, it can negate the action or indicate disagreement without completely turning off the automated system.

“In a in a sense, we’re trying to get them to leave them on, and we’re giving them different ways to interface or to control them, not just shutting them off,” said Gillespie, in an interview. “It’s to increase the communication and put the communication on different levels at the same time.”

That doesn’t mean the driver always wins, for instance, in the case where the automation wants to steer the vehicle in a certain direction, but the driver would prefer to go another way.

“Then you can squeeze the bladder, which will inflate to tell you what direction it plans to go, and it will then respond with either quickly deflating it and inflating the alternate one to say, okay, fine, that’s all right, we’ll go in the direction you’d like to go, or it would pulse to say, I know that there’s construction in the direction you want to go, so let’s not do that,” explained Baez.

The experiments were conducted to examine three conditions involving 30 participants taking 39 turns at various simulated intersections:

  • Two-way communication between automation and driver
  • One-way communication between automation and driver
  • No communication between automation and driver

The result was two-way communication led to “significantly improved driver performance when turn intents did not match,” as stated in the study titled “Haptic Shared Planning and Control: Enabling Coordination of Future Actions in Human-Autonomous Vehicle Teams Through a Haptic Negotiation Interface,” published in the journal Human Factors.

As part of the experiment, after every three turns participants were asked to rank their trust in automated systems between one and five.

“We found that in comparison to no-way communication, and in comparison, even to a display where the driver gets a heads up, two-way communication significantly reduced the amount of swerving on the road because they were able to negotiate and come to a consensus,” said Baez.

Advising Baez on the project was industrial engineer Nadine Sarter who specializes in similar issues related to human-to-machine interactions in aircraft—basically involving pilots and automated systems, Gillespie pointed out.

As much progress as the research has made, it’s currently unfunded after years of backing by Toyota, leaving a future phase, well, on the bubble.