Approximately a hundred thousand people are killed every month in traffic accidents around the world. If this was happening with aircraft; it is likely they would be all grounded. The number of cars is slated to increase from 1 billion (2010) to over 2 billion in 2020. Almost 90% of accidents are caused by human mistakes. Certainly, when I drive around town (I like to think I am a defensive & cautious driver!), I get alarmed with the quality of driving and the number of accidents I have personally witnessed.
Two years ago, Audi (part of Volkswagen) successfully dispatched a self driving car through a few hundred tight curves over 20kms of sealed (paved) and dirt road in Colorado. Perhaps the driverless car can be a workable solution to our driving woes?
There are now enormous opportunities for engineering professionals in getting into the fast growing field of driverless cars.
The Driverless Car
Certainly, seatbelts and airbags have saved lives. However, the next big revolution promises to be driverless cars. Think of the opportunities to co-ordinate routes, optimise the capacity of roads, save time and fuel. The first commercial driverless car is anticipated by the end of this decade.
Adaptive cruise control has been around since the early nineties – allowing one to use a radar system to track the car in front and to accelerate or decelerate automatically. Autonomous driving is a combination of adaptive cruise-control and lane-keeping systems (warning or adjusting your steering if you move out of your correct lane). This supposedly will only add a few thousand dollars to the cost of a car. There is already evidence of reduced insurance claims for injury of almost 15% in the use of these systems.
The Technology required
As would be expected – the critical components for a driverless car are based around instrumentation and electronic/electrical engineering; not so much mechanical engineering parts. An array of sensors is required to detect the surrounds using cameras, radar and lidar (light emission rather than radio). Ultrasonic detectors for parking and gyroscopes, accelerometers and altimeters provide precise positioning. Communications between neighbouring cars would be another useful component.
The Inimitable Google
You have probably heard of Google’s work in the driverless cars. Their strategy is to build up a detailed map of items along the route in manual mode and then switching into autonomous mode. Over time, the route mapping becomes more and more detailed as the route is covered more frequently (such as going to work).
The challenges are naturally huge. For example, deciding whether an object in the road is a plastic bag or a solid rock can be difficult for software to assess. Or the real edge of the road (e.g. a dirt road after a heavy downpour of rain), humans are still better at doing this type of work (at the moment).
Regulations will need to be put in place to eliminate any hobbyist putting his driverless car out on the road and causing a huge accident without some certification process. We all know how reliable software is and the need for thorough testing….
Driverless Trucks are already being used
In Australia, the Rio Tinto mining company is already using a fleet of self driving trucks (and trains) to haul iron ore (over 300 tonnes per truck). No accidents have been reported to date – probably because of the variety of sensors and interaction (‘they chat to each other’) between the different trucks’ controls systems.
Perhaps in the future, some cities will ban manual driving as a result of proven evidence in saving lives and reducing congestion. Surely the true measurement of the success of this technology (hopefully seen in the next decade) will be in significantly reducing accidents on the road and reducing the overall cost of driving a car.
Naturally, building an effective driverless car will be based on everyone's input. As Eric Schmidt (of Google) remarked: None of us is as smart as all of us.
Thanks to the Economist for interesting reading on the topic.
Yours in engineering learning