By Chris Sheedy

A new and wondrous breed of vehicle is currently travelling a 3.4 km route in Perth, Western Australia.

The RAC Intellibus, a fully driverless electric shuttle bus, carries up to 11 passengers along roads that it shares with cars (including parked ones), trucks, motorcycles, bicycles and pedestrians, at the sedate pace of 15 km/hr.

The vehicle is capable of greater speeds – up to 45 km/hr. But as it is in a test period, its keepers at the RAC are taking a safety-first approach. They need to make sure the vehicle and its technology are up to the job. They also need to convince sometimes reticent members of the community that the vehicle can be trusted.

“We’re developing a fundamental understanding of how the vehicle behaves in all environments before we can increase the speed,” says Anne Still, RAC’s General Manager, Public Policy. “Ninetyper centof all crashes are caused by driver error, so taking the driver out of the equation can be a good thing. But we need to make sure it’s a safe transition and we want people to understand the positives.”

The bus uses LIDAR (remote sensing that uses pulsed laser light to measure range), cameras, GPS, odometry (which uses data from motion sensors) and emergency braking to safely drive and navigate. Amazingly, the mix of technologies meant no changes needed to be made to the roads of Perth before Intellibus was set free.

“The software we use doesn’t recognise the white lane lines because we mainly use GPS for positioning,” Still says. “But in the next phase of our trial we will be looking to test technology that recognises lines, because if you’re running an autonomous vehicle under heavy tree cover, or in streets between tall buildings, you can lose a GPS signal.”

That’s the big issue for autonomous vehicle researchers right now – different environments require different types of technology. Some of these technologies rely more heavily on road markings and signs, some thrive in a heavily built environment and others flourish on the open road.

Preparing our roads

Henry Petroski, the Aleksandar S. Vesic Professor of Civil Engineering and Professor of History at Duke University, and author of The Road Taken: The History and Future of America’s Infrastructure, says there are several issues around public policy and legality that will need to be worked through and smoothed out before autonomous cars can be let loose in large numbers on roads around the world.

Experts disagree about when that might be – Petroski suggests ten to 20 years – but in the meantime, the roads themselves can be prepared for the new form of traffic. But what changes might be necessary? “There are two basic technologies that we’re talking about,” Petroski says. “One is the technology that is on the vehicle, the software, cameras, radar etc. Then there is the technology on the roads, and it is possible that some of these vehicles will rely very heavily on the roads being close to ideal.”

“In some cases, the lines on the road must be clear and sharp for the autonomous vehicle to operate most reliably. Some autonomous vehicles constantly monitor the boundaries of the lane. But if there are conditions such as heavily worn lane markings the autonomous vehicle could become confused. If it’s raining and the slickness of the road causes reflection, or moisture splashes onto the camera lens, this could also create problems. There’s not one answer for all technologies.”

Professor Huei Peng, Professor of Mechanical Engineering and Director of the Mobility Transformation Centre at the University of Michigan (where autonomous vehicle testing groundMcitywas recently created) agrees. Peng says his colleagues are currently working with 3M to design paints that will be far more visible at night and in the rain. These paints are being designed with robot vision, rather than human vision, in mind.

“Currently we’re trying to design robot drivers to simply fit into our current infrastructure, which is the result of trillions of dollars of investment. So we are not yet re-designing the infrastructure for robots, but certainly, when they become a significant portion of the vehicle population, we need to be prepared to make changes.”

Peng says other infrastructural changes will involve the provision of high-power charging stations for fast charging of electric vehicles, signage that contains richer sets of symbols that can be read by robots (similar to the way barcodes work today) and less parking stations and parking spaces, as autonomous cars will likely spend 24 hours on the road, or will efficiently find available spaces.

This story is an excerpt. The original was published as 'Just around the corner' in the April edition of IPWEA's inspire magazine. Read the full version here. 

Image: The RAC Intellibus