Driverless Cars Could Save Hundreds Of Thousands Of Lives

FROM THE ECONOMIST INTELLIGENCE UNIT

Commercial airliners largely fly on autopilot. Would bypassing the driver also improve safety for cars?

The 2004 film I, Robot suggested that autonomous or self-driving cars would become widespread by 2035. While full-scale adoption is some way off, recent displays at the Consumer Electronics Show (CES) and the Detroit Motor Show suggested that the technology is starting to become available.

On show at the CES in Los Angeles, for example, was a Toyota Lexus LS car, which uses light detection and ranging technology (LIDAR) combined with stereo cameras to measure distances and note changes in its environment. The car is currently being tested at a new 8.6 acre facility in Japan, where Toyota has tried to replicate real-life driving conditions.

While most of Toyota's technology is simply bolted onto a normal Lexus, Germany's Audi has gone one better and developed a prototype car that uses a compact laser sensor small enough to fit within the car grille. Daimler and Nissan are also working on self-driving projects. Nissan's uses an innovative fly-by-wire approach that bypasses the steering column entirely.

The idea of autonomous cars is far from new: it first emerged in 1939, when General Motors demonstrated an automated highway system at the World's Fair in New York. Even though the technology had not yet been created, the exhibit envisaged a radio control system that would guide the car, taking into account the distance from the car in front. By 1958, a GM prototype could travel for one mile unassisted.

Since then, various companies have tested various vehicles in various conditions. Sweden’s Volvo, for example, began its Safe Road Trains for the Environment project in 2011, backed by the European Commission. The system, called “platooning”, uses inbuilt vehicle sensors to keep cars at a safe distance from each other, travelling as a road train. Volvo claims this increases road safety and cut down on CO2 emissions. The project successfully drew to a close in September 2012, when a road train drove down a public motorway in Spain.

Then two years ago, internet giant Google kick-started its driverless car project, with a fleet of seven Toyota Prius cars that use Google Street View data and onboard radar combined with LIDAR to navigate. According to Google, the cars have logged over 300,000 miles of testing without incident, and have managed to negotiate San Francisco's Lombard Street, with its eight hairpin bends, as well as the Golden Gate Bridge and the Pacific Coast Highway.

Different Strokes

Some of the differences between these projects point to different aims. Nissan sees its system, which uses a converted Leaf, as a way to enhance the driving experience rather than replacing the driver. It would kick in if, for example, someone stepped out in front of the car. Daimler's R&D chief agrees with this approach, pointing out that the advanced cruise control in some of its Mercedes models already brings the car to a controlled halt if the car in front stops abruptly.

Google, meanwhile, is interested in automating the whole driving process, leaving the driver free to do something else. Even so, most carmakers stress than ultimate responsibility for the car lies with the driver. Toyota is particularly cautious. Its North America president, Jim Pisz, says that its Lexus "has the ability to drive itself, but we won't allow it". The company is developing a virtual co-pilot to monitor driver behaviour.

Safety will certainly be the big selling point of autonomous cars, particularly in emerging markets where the rapid take-up of driving has led to a horrific rise in road deaths. The World Health Organisation estimates that 1.2m people are killed in road crashes every year, while 50m are injured – figures that are rising rapidly. The International Organisation for Road Accident Prevention estimates that human error accounts for 90% of road accidents worldwide. In the airline industry, automation has already helped to reduce crash rates significantly.

In an ideal world, autonomy would combine with cloud-based intelligence to create a system that allows cars to monitor their surroundings and communicate with each other, while roadside sensors monitor traffic flows. Networking firms are already waking up to the opportunities. Both Cisco and NXP Semiconductor this month decided to invest in an Australian firm, Cohda Wireless, which specialises in inter-vehicle networking.

All these projects suggest that autonomous technology really can work, and that it can improve safety, as long as the technology is reliable. The US states of Nevada, Florida and California have already passed laws permitting the use of driverless cars and Washington DC looks likely to be next. This will pave the way for mass production and commercialisation.

Nevertheless, plenty more development will be necessary, as well as legislation to establish exactly who is liable if there is a crash. Even then, it will be difficult to introduce hundreds of autonomous cars into a global road network that is already crowded with normal cars. Self-driving cars could eventually take the stress out of driving, but until they are well-established they are likely to make traffic management a bigger headache than ever.

Click here to subscribe to The Economist