Dr John Baruch: Robot cars - a vision for the future? - Spring 2008

In January, Rick Wagoner, the Chief Executive of General Motors, told the global Consumer Electronics Show in Las Vegas that his company wanted to revolutionise the car industry by dropping reliance on fossil fuels and handing over control of cars to 'robot drivers'.

The car, nicknamed Boss and part of the Cadillac range owned by General Motors, is being developed in conjunction with Carnegie Mellon University, with reported plans for public testing by 2015 and full production three years later.

Here, Dr John Baruch, from the University's School of Informatics, says robotic driver systems for cars could slash the death toll on the roads, but will the car manufacturers welcome it?

Earlier this year, The Guardian newspaper reported on the development of Cadillac's Boss car, and implied that its driverless feature was a mere afterthought.

Driverless cars are not an afterthought and would make the biggest impact on our lives of any of the current technologies. They would end 90 per cent of vehicle accidents and deaths, remove parked cars from our streets, accelerate hydrogen fuel for vehicles and make an immense saving of the time now spent on congested roads. Driverless cars or cabs could be parked off the streets waiting for their owner or a user to call them.

The Guardian quoted a General Motors spokesman saying: "The technology exists right now to move cars without a driver. This [model] would know where all the vehicles are around it, dramatically reduce accidents and even reduce congestion."

He failed to say that VW and Mercedes are believed to have had robot cars on the road for a year or more. The motor companies appear to oppose robot cars because they would completely change the market's business model. You could no longer sell the 'driving experience' and it would blur the divide between public and private transport - it would make no difference whose robot took you to work or to the pub.

There are predictions that driverless cars would become part of a social network, exchanging information about road conditions - just as is becoming possible for owners of the BMW series 7.

The difficult parts of all this include moving away from the thinking of transport experts who predict intelligent roads, which would tell drivers and eventually cars what to do - a lucrative future that road building companies are planning. Another problem is the need to develop the law to license software for driverless cars.

The key technologies are familiar to British universities, like Bradford. We can construct the systems for robot vehicles using strengths in mathematics, physics, law, communications, computing and engineering. The objective of the collaboration would be to make money for the universities. A key problem is the law.

The road to robot cars

The law must be changed to allow driverless cars on the road and the development of software licensed to drive them. The software development will merge intelligence, knowledge of the road system and a library of driving situations and what should be done. Once the software had been developed, the licensing process would take at least a year with the software taking vehicles through many different driving situations on many types of roads.The two components of a legal framework and a licensing process will pave the way for robot taxis and cars in the UK. Other countries will follow.

Most of the engineering would be to ensure that every aspect of the vehicle control is completely electronic. There is also the challenge of rethinking how people would sit in a vehicle where there is no driver. These teams will not be led by vehicle engineers but by the sort of people who have put together our phone networks. We already have the basic vehicle that uses electronic systems to control its operation. We can purchase vehicles with driverless parking from Lexus, Mercedes and BMW with their electronic control systems. That is the basis for a robot vehicle.

The next step is to sensorise the vehicle so that there can be a record of everything that it does. It would have cameras and radar systems to map out the road ahead and the space around it. Sensorised cars would be especially attractive to insurance companies or people with high insurance premiums since they could replay accident scenarios and show where the responsibility lay.

With sensorised vehicles it becomes possible to develop the software. The key is the mathematics of relative motion. UK researchers lead the world in this. A group in Newcastle showed how locusts could fly with complete collision avoidance with only one brain cell. Robot vehicle software would copy these techniques using different software for moving off and for travelling.

Computer games could support the programme with software learning the basics of how to drive cars in special computer game scenarios, before it is tried out on vehicles in simulated road situations. Of course there would be sat nav for navigation.

Accidents would trigger a learning process to ensure such an accident did not happen again. There would be updates to the situation libraries in the black boxes. Most accidents would still be with human drivers like today who were not totally sober or just driving too fast. The key job for the systems engineers is to put together these building blocks for autonomous vehicles and win the confidence of the public to let these vehicles greatly improve our quality of life especially our quality of transport.

Expertise at Bradford

We run the only autonomous robot available to all on the internet, which is changing the way school children do experimental science.

At present the real world facility is the Bradford Robot Telescope operating by itself 3,000 km away, high up on the island of Tenerife. The robot telescope currently services three groups of people. There is thehttp://schools.telescope.org which is a subscription site to support the National Curriculum studies of English school children from the age of 9 to 16.

There is also a subscription site for Scottish schools and there is a free site without the UK curriculum support material, which is used by amateurs around the world, by many students studying A level and by many astronomy undergraduate programmes.

To date the Bradford Robotic telescope has serviced over 50,000 requests and has over 14,000 English school students registered to use it with over 500 school teachers. The system is totally autonomous. If you ask the telescope to observe for you there is no human intervention and most school children receive their requested observation in time for their lesson the following week. Bradford understands about autonomous robots and with the law, maths and physics departments of our sister universities we could deliver robot taxis in a three year programme.

So why isn't this at the top of the agenda for our research councils? Well, now that universities have to respond to commercial interests, their vehicle and transport sections are dominated by the business models of the major car manufacturing and road building firms.

Driverless vehicles are about thinking outside the box; if you always ask the same people the same questions, driverless cars will remain an afterthought.