Pages

Saturday, June 11, 2011

The Self-driving Car

The Future of the Self-Driving Car

All over the world engineers and computer scientists are in full swing, trying to produce some of the most amazing cars. Among these is the driverless car…and exactly as the name indicates, no one needs to drive it. Using a massive array of different sensors to judge when to turn, stop, accelerate, indicate and break, the driverless car navigates itself through any street it comes across, and previous tests show that scientists may be close to success. That’s the exciting news. 

So what’s the bad news? Driverless cars are supposedly banned in many countries so testing and selling of this concept isn’t going to be quite as fast as we hope. And there’s another thing too. How accurate can a machine be at making decisions? There are many circumstances where driverless cars could save lives, but there are also occasions where we must make quite difficult, spontaneous decisions, which a machine may be unable to make in the required time, especially during early stages of deployment. A lot of testing will be required to perfect this system. Nonetheless though, it’s coming….and it’s coming very soon. And the cars aren’t all that slow either.

Why have a self-driving car?

 

You may have had plenty of moments when you’ve gone out with friends and needed to take a taxi, or a bus, or a train to get home. You may need to be picked up by family. Or maybe you’ve been to the pub and had something to drink and so can’t drive home. You probably would much rather the comfort of your own car than being stuck in a jam-packed train carriage or wait in the freezing cold not knowing when the next bus would come.
Well, if you just took out your mobile, accessed an app that connected to your car’s built in computer system and instructed your car to drive over and pick you up then all would be sorted. It pulls up beside you, you get in, and you sit back and relax as it takes you safely home.
No need to touch the steering wheel. In fact…do you need to do anything? It’s like your own personal taxi service. Would you even need a driver’s licence? 

Mobility on Demand
As the first video showed, as a taxi service, this could be very useful. You may not even need to buy a car. There could just be a pool of many cars in a garage somewhere in the centre of town and you could just call them up and choose your destination and it will come and whisk you off. There is a name for this.
MIT (Massechussets Institute of Technology) have called it Mobility on Demand. Why have MIT given it a name? Well, because MIT are in fact working on their own prototype of a driverless car, which they name CityCar. Amazingly, they predict they can get the first cars in to commercialisation by 2012. On top of that, they expect to sell 100,000 by 2014. No doubt they will come at a very high price and will be quite slow in their first versions.
However, to understand why this concept is coming so soon, it’s important to know what CityCar is and what it’s aims are.


These plans hope to greatly reduce traffic congestion because the computer in the car can take into account traffic to continuously calculate the best possible routes. Then, this could be taken to an even greater extent whereby the cars can actually communicate through signals with each other and with a big server system located somewhere away from the city to find out, in real time, when accidents occur or where there is traffic build up for example, to dramatically reduce traffic congestion.
Parking also appears to be highly improved if the cars can just reduce their length to fit into tight spaces. And, as the wheels are controlled robotically, the car can turn in ways it can't for a normal car. It can turn on the spot 360 degrees rather than having to be turned through a U-turn.

How is it done?
This technology is all possible now and we know how it’s down. It just has to be tested and improved until it is extremely, and hopefully 100% accurate (although some may say that that’s never possible).


Data is picked up from radar, scanners, cameras, laser scanners and using GPS. Motion sensors, object recognition and heat sensors are integrated into more accurate driverless cars. The data put together is mapped to form an overall image of the area and any surrounding obstacles. An image can be produce such as the one below to represent this information.


Google’s attempt for a driverless vehicle uses Google Street View in fact. Below, the google car has a 360 degree view of the area by using a camera on top of the car that spins round incredible fast:


Google has been lobbying as of April this year for the Nevada State Government to legalise driverless cars and to be the first US state to do so. This would be great for Google as they could finally test legally on the streets on a much grander scale and start selling their driverless cars. Around 8 months ago, their cars were reported as having gone 140,000 miles with only one accident – it knocked into the car in front when stopping at a traffic light. The man trying to push it through – Sebastian Thrun suggests how much more we could do with our car journeys, perhaps from watching videos, playing games, even exercising or maybe just working. The presentations to the Nevada government will continue till mid-June when the verdict will be given.

The ethical, personal and economic issues:
The Google Car aims to keep drivers in control by suggesting that the driver still mainly controls the car. The self-driving function can simply be used when the driver needs to answer the phone for example or perhaps on a motorway or in traffic. However, some companies wish for the car to drive by itself without any need for driver interaction at all. The issue here is about machine breakdowns for example because, after all, we have computers crashing or machines failing all the time in our lives. The problem is, on the roads it can be deadly. Is it really worth placing our lives in the hands of machines when on the roads? Maybe. Accuracy is key and many designers will probably ensure that if something goes wrong, a backup system in the car brings the driver to safety.
Personal preference brings up a lot of debate too. Who wants the same car as everyone else when out an about? Many people don’t care, especially if they are don’t drive much. There are many car fanatics out there. Wouldn’t allowing driverless cars on the road mean that non-driverless cars will be banned? Or can they work together? Can we have both types on the roads? If not, what would happen to people like Jeremy Clarkson? What do people who actually like driving say about this?

Economic issues come into play too. Why buy a car if there are plenty of driverless cars on demand in the streets which you can just hop in, use an oyster card like thing to pay for them and drive off for a while? You may buy a driverless car to benefit from the different colours, specifications and models but is it really going to be worth it? Won’t it cost far more than necessary? The car manufacturers could suffer greatly even if they do produce their own driver-less cars and many countries benefit from car production...so that’s a lot of economic problems and a massive shift required in industries.


When will we see these cars?
Very soon. Very soon indeed. Google and MIT reckon a couple of years. International organisation Bosch reckons within the next ten years. Who’s right? Well perhaps both. We may be seeing driverless cars used in city environments to take people around a little bit like trams at first...or maybe more like the Boris Bikes in London. A few years may pass before we witness any private ownership of driverless cars so it will be a while before you have one in your garage. And it will probably cost a fair deal too.


More information:
This is a very interesting video to view. Look up the guy and the google project. And look up about MIT CityCar. The results are pretty interesting. Keep updated on the lobbying too on the Nevada government which is due to end this month.