The take-up of automated vehicles is likely to lead to much more congestion on roads in the major cities of Australia with the total number of all vehicles in a "mixed fleet" to rise to as much as 30 million by 2030, an economist has forecast.
Brian Haratsis, executive chairman of MacroPlan, said on Thursday there was a substantial shift in the economics of personal transport which would emerge as costs eventually dropped and consumers took far higher numbers of personal transport journeys.
"Congestion is going to get a lot worse," Mr Haratsis told The Australian Financial Review after giving a presentation at an International Driverless Vehicle Summit in Adelaide.
"Roads will become real estate."
He said astute residential real estate buyers were already factoring in the long-range value upside for properties located in good suburbs close to freeways.
Transport users would value the extra time they were able to spend doing work on laptop computers in automated vehicles and "robo-taxis", or in ride-sharing vehicles still physically being driven by a human driver, as they headed to city workplaces.
The total number of vehicles on Australia's roads is about 19 million currently. The number of traditional vehicles would keep growing, but then the total "mixed fleet" including both traditional vehicles and autonomous vehicles could expand to 30 million in just over a decade.
He acknowledged he had contrarian views to many experts in the field. But the proliferation of Uber vehicles and those in other ride-sharing firms would accelerate as the "costs per kilometre" of taking trips dropped. Automated taxis would also trigger an expansion.
Ecommerce weighs on congestion
Online retailing was rapidly growing, and he predicted that by 2025, about 25 per cent of all non-food retail would be happening online.
Even the best online retailers such as Amazon were still trying to perfect the "last mile" - where smooth and efficient deliveries of the goods ordered by online shoppers arrived at their houses.
Autonomous vehicles would play a big role in improving that service - but would also be a big contributor to the growing congestion on roads.
There would also be room for extra drivers of traditional vehicles because of the emphasis on getting the "last mile" right. "Professional driver demand is actually outstripping demand," he said.
He said the average cost of owning and running a vehicle currently was about $8,000 per annum in Australia.
The business models of traditional manufacturing of vehicles was being overturned as technology companies and software experts steeped up their development of autonomous vehicles and it was important to remember that it was the private sector driving the industry.
Science was colliding with economics.
"Driverless vehicles and autonomy is being driven by the private sector not the government," Mr Haratsis said.
Cost key to usage
Makers of autonomous vehicles, once they got the technology absolutely right, could see the extra demand which will be created by the advent of autonomous vehicles if they are priced astutely because the appetite for more personal journeys is there.
"The cost per kilometre needs to halve," he said.
Mr Haratsis said between the start of January 2016 and mid-2017 in Sydney ride-sharing companies such as Uber snared about 15 per cent of total taxi journeys away from traditional taxis, but the overall market for personal journeys had increased by 35 per cent.
ASX-listed Cabcharge is changing its name as the taxi group fights ride-sharing firms, which also include Taxify and Chinese new entrant Didi.
"It's actually about the price per kilometre," he said. While there have been complaints about a surge in prices at Uber during busy periods, ride-sharing firms had shown there was latent demand.
"If the price drops, people will travel more," he said.
Earlier, NSW chief scientist Hugh Durrant-Whyte told the conference he didn't think there would be widespread take-up of autonomous vehicles in his lifetime, and that the problem wasn't in sensor technology, but in the algorithms behind them.
"The problem is the algorithms fail," he said.
Makers need to be absolutely confident that the mathematics behind their algorithms had factored in every possible outcome, and could be completely trusted, and this would take a long time.