Queensland researchers are teaching robot drivers to work with human roads.

Researchers from the Australian Centre for Robotic Vision are taking an electric car fitted with high-tech sensors and computers on a 1,200km road trip in which it will be tested on a wide range of road and driving conditions.

“Engineers at QUT’s Research Engineering Facility have developed a research car platform equipped with a range of state-of-the-art camera and LIDAR sensors used on autonomous vehicles,” project leader Professor Michael Milford said.

“So, as we drive, AI will watch and determine if it could perform the same as a human driver in all conditions.

“During this trip, you could say AI will become our ultimate back-seat driver.

“The big problem that faces autonomous vehicles right now is that at the moment they don’t drive as well as humans in all possible conditions.

“We’re targeting how the car might use infrastructure, such as lane markings and street signage, to help it to drive well.”

Professor Milford said current autonomous car systems, when faced with some of the road conditions Australian drivers deal with daily, either refused to go into autonomous mode or hand control back to a human driver.

This research project will look at how an autonomous vehicle’s artificial intelligence systems copes with Australian road conditions in four main areas: lane markings, traffic lights, street signs and how to determine a vehicle’s exact position despite errors that occur with GPS systems in highly built-up urban areas or poor reception areas such as tunnels.

Professor Milford said past studies and his team’s initial experiments show that autonomous cars could have difficulties on rural roads which often lacked lane markings on the side or even a centre line.

“A human driving down a rural road knows to stick on the left and they infer or imagine that there is a line in the middle of the road,” Professor Milford said.

“But they will also cross that imaginary line to go around obstacles quite freely. That’s very hard for an autonomous car.”

Professor Milford said early testing of the system had already revealed how a paint spill on the road from the back of a truck could confuse a self-driving AI system into wrongly identifying it as a lane marking.

“The primary goal of our research is to determine how current advances in robotic vision and machine learning – the backbone of AI – enable our research car platform to see and make sense of everyday road signage and markings that we, as humans, take for granted,” he said.

“So, safety is an obvious off-shoot, but not the focus of this particular study. What’s important is understanding how AI performs and potential improvements to both the technology and physical infrastructure as the autonomous car revolution unfolds.”

The pilot project is part of the Queensland Government’s wider Cooperative and Automated Vehicle Initiative (CAVI).