Driverless: Intelligent Cars And The Road Ahead
In this book, Hod Lipson and Melba Kurman offer readers insight into the risks and benefits of driverless cars and a lucid and engaging explanation of the enabling technology. Recent advances in software and robotics are toppling long-standing technological barriers that for decades have confined self-driving cars to the realm of fantasy. A new kind of artificial intelligence software called deep learning gives cars rapid and accurate visual perception. Human drivers can relax and take their eyes off the road.
Driverless: Intelligent Cars and the Road Ahead
But the field has made significant leaps forward in recent years, and today there are multiple forms of autonomous driving technology actively being tested on the road including driverless delivery vehicles and software and hardware platforms designed to transform traditional trucks and cars into self-driving machines. But how close are we, really, to becoming a society of passengers?
We spoke to the company at CES this year, where it revealed that we could be 10 years from self-driving cars being on public roads . But the company also said it's seeing rapidly increasing demand for autonomous vehicles in non-public situations.
A major challenge for cars that drive themselves is winning social acceptance. They would have to share the roads with normal cars with drivers as well as with pedestrians, animals and unexpected objects.
Regardless of how zanily futuristic and even dangerous such machines might feel, especially the idea of sharing roads with driverless cars, that era is inevitable simply because artificial intelligence is far better at avoiding accidents than human drivers, said HIS analyst Egil Juliussen. It just might take some time, such as until the 2030s, he said.
As a specialist in artificial intelligence, I have been conducting research on autonomous vehicles since 2003. My students and I have developed new protocols for cars to coordinate movement through intersections that promise to be dramatically more efficient than traffic signals and stop signs when most of the cars are autonomous. We participated in the famous 2007 DARPA Urban Challenge in which fully autonomous cars were demonstrated to be capable of driving in (sub)urban settings, while following the rules of the road.
When we began, even our forward-looking colleagues in computer science looked at us incredulously when we said that we thought autonomous cars would eventually become commonplace. Research on autonomous cars was thought to be well ahead of the curve. However, last year we retired the car because of how much progress is being made in industry. It is no longer a topic of futuristic research; autonomous driving is becoming a marketable product.
On the other hand, my colleagues and I have been actively considering the opportunities for reducing congestion, and thus indirectly reducing fuel usage and emissions. At intersections, cars may be able to coordinate so finely with one another that they barely need to decelerate when approaching an intersection (and seldom, if ever, need to stop). The direction of lanes could be changed minute-by-minute based on how many cars are travelling in each direction on a road. And novel pricing mechanisms may make it possible to incentivize more off-peak travel, and to encourage distribution of trips so that fewer roads get clogged.
Of course all of these changes will need to take place in such a way that the roads remain friendly, or ideally become friendlier, for bicyclists and pedestrians. Even if demonstrated to be safe, will they feel comfortable walking and riding in the presence of autonomous cars, or will we need to build new special lanes, bridges, or tunnels? 041b061a72