This article is more than 1 year old

Register lecture: Teaching self-driving cars how to be more human

Road rage against the machines

Autonomous vehicles have been given the green light – according to HM Government – which confidently expects they'll be pounding Britain's potholed highways by 2021.

But before AVs take off, they must be extensively tested – and therein lies the rub: those who are building smart-car systems require highly realistic simulations and testing conditions to ensure their creations are safe. That demands a really solid understanding of human road behaviour – with all its foibles, unpredictability and aggression included. And, frankly, we just aren't there yet.

Sit back and enjoy our Reg lecture from two academic and car-industry experts on what's missing and the work to close that gap.

Dr Gustav Markkula, a University of Leeds associate professor who is studying human behaviour and cognition in road traffic, tackles the science part. A car-industry veteran with 10 years at Volvo, Gustav discussed the complex work he and others are involved in, attempting to mathematically model the behaviour of human road users.

Joining Gustav is Nick Reed, Bosch head of mobility R&D. Nick, who led Transport for London's £8m Innovate UK-funded GATEway public trials of automated vehicles in Greenwich, who discusses how this feeds into practice.

Diving into our psyches, Nick talks about what makes humans hate the idea of a computer being given complete authority over several tonnes of fast-moving metal. He touches on why it's critical that manufacturers, governments and regulators nail these concerns or risk a backlash against driverless vehicles.

You will hear from Gustav and Nick about why the barrier to entry for robot drivers is higher than for humans, what the machines have to learn from us and how it's hoped decision-making models will let the transport systems of tomorrow be delivered in a socially responsible and human-centric way. ®

More about

TIP US OFF

Send us news


Other stories you might like