For autonomous cars, ancient Greece is the word
Fundamental science will keep them on the right track, writes Australia’s Chief Scientist, Alan Finkel.
Driving from Rome to Tuscany in a rental car, we were at the mercy of the satellite navigation system. The first crisis was when it indicated we were on the exit ramp but we knew we were still on the freeway. So we ignored it. After a few hundred metres, it nonchalantly showed us we were back on the freeway.
Human judgment saved the day. But what if this had been a self-driving car? The problem is that the civilian satellite navigation system as implemented in most cars has wiggle room; it’s only accurate to about 10 metres.
Fortunately, a state-of-the-art technology upgrade is on its way. Surprisingly, delivery of more accurate mapping will also depend on the ancient science of geodesy, born in ancient Greece from the struggle to decide between Homer’s view that the world was flat and Pythagoras’s view that it was spherical.
Geodesy is not a science that I often think about, but it is now in my mind because I chair the selection panel for the Prime Minister’s Prize for Science, and the 2018 winner was one of the world leaders in geodesy, Professor Kurt Lambeck.
When he started as a young geodesic scientist, Professor Lambeck never dreamed his chosen scientific discipline would be a critical enabler of self-driving cars.
Driving is a complex challenge, and autonomous vehicles need everything working for them. They use video to help them follow lanes, and radar and lidar (similar to radar but using light instead of microwaves) to measure distances to other cars and objects.
Autonomous vehicles will eventually talk to all the other cars in their vicinity to determine their positions, speeds and intentions, as well as to the city computers that control the traffic signals, temporary speed limits and road repairs.
But what happens in snowy, wet conditions on a country road on a moonless night? With all its sensors blinded the car will be totally reliant on its sat nav. Relying on today’s systems will likely see many autonomous vehicles drive unwittingly into a ditch.
Tomorrow’s sat nav will shrink that wiggle room a hundredfold to 10 centimetres, the width of a hand. It will keep cars well clear of ditches.
To achieve such stunning positional accuracy, self-driving cars will employ ground-based stations to augment the satellite position data. Car receivers will hone their accuracy by combining signals from the American and European satellite systems and perhaps the Russian and emerging Chinese satellite systems.
The invisible physics behind all this is Einstein’s theory of special relativity and atomic clocks that enable the satellites to know exactly where they are relative to each other and the mass of the Earth. But Einsteinian physics and extraordinary technology are not enough. If you want to know exactly where an object is on Earth, you need a perfect map.
The original design of the GPS satellite navigation system was based on the advice of geodesic scientists working with the US navy. Mathematical modelling of the shape of the Earth by geodesic scientists has been essential to knowing exactly where a grid point on Earth lies relative to an orbiting satellite.
And it works in reverse. By observing the slight shifts in the orbit of a satellite, geodesic scientists have been able to accurately measure the flattening and bulging of the Earth as it responds to the forces constantly acting within, and from the pull of the Moon.
Besides our planet’s shifting dents and bulges, the continents are on the move. Australia, for instance, is moving north at seven centimetres per year. This means that the precision of the future satellite navigation systems will be wasted unless the terrestrial maps can be routinely updated to an equivalent level of accuracy.
Geodesic scientists call upon a huge number of tools and skills to make this possible, such as very long baseline interferometry to measure the relative motions of ground stations around the planet with accuracy better than 1 millimetre per year.
From their roots in ancient Greek natural philosophy, with a nod to Einstein’s special theory of relativity, today’s geodesic scientists call on a proud tradition of fundamental science to generate the precision Earth maps that will underpin tomorrow driverless cars.