Menu
Menu
Science has scarcely addressed how autonomous vehicles interact with human drivers: research

Science has scarcely addressed how autonomous vehicles interact with human drivers: research

Warwick Business School research notes major AI concerns could limit the development of these vehicles

'Negotiating the traffic’ is not merely a figure of speech. It involves an actual process of tacit negotiation with other road users in a safety critical environment

Nick Chater and Nathan Griffiths, Warwick Business School

Elon Musk has predicted Tesla will have autonomous robotaxis on the road next year.

However, research from Warwick Business School and Jaguar Land Rover warns the AI needed for autonomous cars to co-exist with human drivers, cyclists, and pedestrians, is pushing the frontiers of what is scientifically possible.

The research authors - Nick Chater, behavioural science professor and Nathan Griffiths, computer science professor at Warwick Business School - argue that while there appears to have been “impressive progress” in cars sensing their surroundings, the challenge of interacting with other road users has scarcely been addressed.

Autonomous vehicles promise great gains in human welfare through improved mobility, safety, and environmental impacts, but also brings to light fundamental challenges for cognitive science and artificial intelligence, they point out.

Human interactions are so effortless that we are unaware of the complex reasoning involved

Nick Chater and Nathan Griffiths, Warwick Business School

These include sensing and control (where machines may potentially exceed human performance such as response times) and mimicking or seamlessly meshing with human behaviour in driving interactions, they state.

This could be the “decisive limiting factor” in developing autonomous vehicles, they conclude in their paper, Negotiating the Traffic: Can Cognitive Science Help Make Autonomous Vehicles a Reality?

“Alan Turing famously challenged future generations to create a machine that would be indistinguishable from a person through the medium of written language,” says Chater.

“The future of fully autonomous vehicles depends on science meeting a similar challenge – creating computer systems that drive in a way that blends seamlessly and safely with human drivers.

Tesla CEO Elon Musk
Tesla CEO Elon Musk

"Negotiating the traffic’ is not merely a figure of speech. It involves an actual process of tacit negotiation with other road users in a safety critical environment. It is crucial that everyone reaches the same – or at least compatible – agreements to prevent a potential accident.

“Creating a machine that can replicate that process involves addressing fundamental questions at the frontiers of cognitive science. Human interactions are so effortless that we are unaware of the complex reasoning involved.

“The rate of progress on this challenge may prove a decisive limiting factor in the development of autonomous vehicles.”

A game of ‘chicken’

In the research paper, they present the following situation to illustrate the complexity of making decisions even in a manoeuver as simple as moving briefly into the on-coming lane to avoid an obstacle.

In 1a. drivers face a problem of joint action. Car A needs to encroach on the opposite lane to pass an obstruction, but could collide with Car C. The cars face a game of ‘chicken’—one must give way; but if both give way, they will fall into deadlock. One natural rule is that priority goes to the driver who stays in lane. But in these scenarios, C is moving slowly, so that, unless C speeds up, A can pass the obstruction successfully. But perhaps C is accelerating? In 1b. C starts indicating left. This would seem to imply an immediate intention to turn left at the point where A would encroach into C’s lane; and be tantamount to a “claim” on the “contested” region of road—which implies that A should give way. In 1c., the same indicating signal may now be interpreted as an intention to turn down the small driveway, and hence communicating the opposite message to A, “ceding” the contested area. In 1d., C flashes its headlights, either to yield, or perhaps to signal “I’m coming”—the interpretation may depend on changes in vehicle velocity as well as local informal norms. Many slight variations on this, and similar, scenarios can change the “natural” interpretations of signals and actions of the drivers. An automated vehicle “impersonating” a human driver will be hazardous unless it follows natural human driving behaviour accurately.
In 1a. drivers face a problem of joint action. Car A needs to encroach on the opposite lane to pass an obstruction, but could collide with Car C. The cars face a game of ‘chicken’—one must give way; but if both give way, they will fall into deadlock. One natural rule is that priority goes to the driver who stays in lane. But in these scenarios, C is moving slowly, so that, unless C speeds up, A can pass the obstruction successfully. But perhaps C is accelerating? In 1b. C starts indicating left. This would seem to imply an immediate intention to turn left at the point where A would encroach into C’s lane; and be tantamount to a “claim” on the “contested” region of road—which implies that A should give way. In 1c., the same indicating signal may now be interpreted as an intention to turn down the small driveway, and hence communicating the opposite message to A, “ceding” the contested area. In 1d., C flashes its headlights, either to yield, or perhaps to signal “I’m coming”—the interpretation may depend on changes in vehicle velocity as well as local informal norms. Many slight variations on this, and similar, scenarios can change the “natural” interpretations of signals and actions of the drivers. An automated vehicle “impersonating” a human driver will be hazardous unless it follows natural human driving behaviour accurately.

“Drivers are playing a game of ‘chicken’—one, but not both, should give way, to avoid either collision or deadlock,” they point out.

Sign up for  CIO newsletters for regular updates on CIO news, career tips, views and events. Follow CIO New Zealand on Twitter:@cio_nz

Send news tips and comments to divina_paredes@idg.co.nz @divinap

Join the CIO New Zealand group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags change managementanalyticstransportationlogisticsroboticsAIautomobilesHealth and safetyWarwick Business SchoolElon Muskdriverless vehiclesautonomous vehiclesdigital disruptionalan turingrobotaxisNathan GriffithsNick Chater

More about JaguarTeslaTwitter

Show Comments