With many kinds of Aussie and Brisbane signs on the road, many are wondering how self-driving cars will handle operations. To test this, an AI-driven electric vehicle is going on a 1,200km (745mi for Americans), three-month journey across Queensland.
The e-car, a zero-emission iteration Renault ZOE, will be mapping roads in the northeastern state of Queensland, while manned by researchers from the Queensland University of Technology, based in Brisbane.
Queensland’s Minister for Transport and Main Roads, Mark Bailey, described the process, saying that researchers will be driving the vehicle across the state, wherein its on board sensors will examine roads, and Brisbane signs in order to create a virtual map that AI-equipped vehicles can then use to operate safely in the state’s roads.
Bailey says that, while the technology is still in its earlier stages, AI tech and smart road infrastructure could, potentially, change how people travelled in the state, making the roads safer and reducing accidents and trauma on the road.
The project is aimed at examining how the vehicle and the AI system onboard responds and adapts to road markings, lights and signage, with the additional aim of working past the common problems that GPS systems run into when dealing with built-up areas and tunnels.
A researcher from the QUT’s Australian Centre for Robotic Vision, Michael Milford, said that, while they were driving the vehicle, they’d be determining how it would handle conditions human drivers commonly deal with. He says that, while there’s a lot of excitement regarding autonomous vehicles, there’s a long road of work ahead.
Motorists on rural roads know that they should stick to the road’s left side, or imagine that there’s some division on the road, splitting it down the middle. People can and do cross this imaginary division to get past or around obstacles, but automated vehicles have more trouble with this.
Millard says that past studies, including experiments conducted by the QUT, show how automated cars have trouble with dealing with small things, like a pain spill on the asphalt, or a lack of road markings, can confuse AI. He adds that the primary aim of the project is to see how current developments in technology can enable AI cars can look and see the road signs that humans take for granted, and follow on instinct.