Like many people driving long commutes daily, I’ve been doing some thinking about whether there is a better way. All too often I find myself sitting in traffic jams contemplating the amount of driver error that has contributed to the motorway ‘incidents’ holding us all up. I have often imagined how the driverless car might take some of this pain away. After all official figures tell us that driver error is a factor in 90% of car accidents today.
It is not simply that we are getting worse at driving but the roads in front of us are filling up. Many of our roads are very close to capacity and we cannot keep widening them as vehicle numbers continue their inexorable rise. The amount of vehicle miles car drivers are putting in all over the UK is rising at a rate of over 2% per year and now stands at 244.5 billion vehicle miles. As our roads fill up, tolerance for driver error falls and more accidents occur. Sadly injuries and deaths in those accidents are also now creeping up year on year despite increased vehicle safety features.
But what if more of us left the driving to our autonomous vehicle? Would we see accident numbers fall, road safety levels increase and journey times fall? Technology companies like Google and Bosch; and car manufacturers such as Toyota and Volvo (with its Drive Me Project) have been pumping millions into development to answer these questions in the affirmative.
By June this year, Google announced that its 23 driverless vehicles have now driven over a million miles ‘equivalent of 75 years of typical U.S. adult driving’, and that in the process they had encountered 200,000 stop signs, 600,000 traffic lights, and 180 million other vehicles.
These cars which are being tested on public roads have been involved in a total of 12 accidents over the last six years driving around on public roads. All of these accidents involved being rear-ended or side-swiped by other vehicles. One suspects most of these incidents were caused by impatient drivers as the trial cars have been limited to 25 mph.
But is the technology just way too expensive for you and me? Google's cars have about $150,000 of equipment on-board including a roof-mounted $70,000 laser scanner system. LiDAR, as it is called, is a 64-beam laser that measures distances by illuminating a target with a laser and analysing the reflected light. It allows the vehicle to generate a detailed 3D map of its environment.
The car then takes these generated maps and combines them with high-resolution maps of the world, producing different types of data models that allow it to drive itself. Several U.S. states have already changed their laws to allow driverless cars on their public roads and Google expects to be able to offer its first driverless car to the world by 2020.
In an interesting twist, we will be testing cars as well as drivers for readiness to drive on public roads. There is talk of driverless cars having to do 10,000 miles of ‘safe road driving’ before being given a road license in the States.
You could argue that the driverless car, or at least elements of it, are already here. Many high-end cars already come pre-fitted with lane-keeping systems. Others also offer self-parking systems, perhaps the most well-known of which is Volvo Auto Parking which is controlled via an app on your smart phone.
Self-driving trials are now underway in the UK. In Milton Keynes, courtesy of the UK Automotive Council and the Department for Business, Innovation and Skills, the LUTZ Pathfinder project has delivered the UK’s first three driverless electric-powered ‘pods’ which can take you from Milton Keynes Central rail station to MK’s Hub shopping mall.
However the big issue appears to be - when will we be able to trust driverless cars to drive better than the average car driver on our roads? The boffins in places like Oxford University have been working hard on the issue of machine learning – applying neural computing techniques (based on integration of deep learning and reasoning models) to autonomous vehicles. The idea is that the computer in the driverless car will be able to exercise judgement; thereby compensating for the fact that they cannot do what humans do, e.g. send non-verbal signals to other drivers in order to negotiate who goes first to avoid collisions in traffic or at junctions.
Machine learning aside, Google tells us that driverless cars could be on our roads within just four years. That leaves a few questions for car dealerships. Just how disruptive will the driverless car be for dealerships? Will autonomous vehicles be programed to automatically book in for servicing when they fall due or maintenance issues have been detected? Will workshop dealers need to upgrade their IT skills to run IT health checks and firmware upgrades at the same time as changing the oil? Logically driverless cars will be able to sell with more ‘infotainment’ offerings when new cars are being spec’d up. Are the manufacturers that you currently sell preparing for these changes? Maybe 2016 is the year to answer some of these questions.