Friday, 9 November 2012

HUMAN BEATS ROBOT CAR IN RACETRACK DUEL


A human has beaten a robot car in a race around a circuit. The competition was held at the Thunderhill Raceway in California. The man v machine duel now has a definitive winner.
The car used for the track battle was a 272hp Audi TTS that can “drive itself”, while a racing instructor from the circuit used an identical vehicle – apart from the computer controlled guidance, of course.

It was a hollow victory for the organic pilot, however. As well as knowing the circuit – unlike the digitally controlled car, which had to learn it – the human only beat the silicon chip by a few seconds.

The race was part of a research programme by the Centre for Automotive Research at Stanford University looking at how to develop control systems to make the humble car more autonomous.
The Audi – which is called Shelley – is fitted with an array of sensors that calculate the vehicle’s position on the racetrack. Data concerning the tyres’ grip levels and many other parameters is fed back to the car’s ‘brain’, helping it work out the best route to take around the circuit.

Once familiar with the three-mile course the car was raced against a Thunderhill Raceway instructor. The autonomous vehicle recorded a top speed of 115mph and a lap time only a few seconds slower than the racing driver.

The project has real relevance to road-going vehicles however, according to head of the Cars Lab at Stanford University, Professor Chris Gerdes:
“The basic idea here is also applicable to safety systems. If we can have cars that drive up to the limits and recover if they go past, this is something that could help ordinary drivers – for instance on a slippery road.”

Scientists observed that the computer-controlled car predictably did not have the finesse of a human driver. According to Prof Gerdes:
“In situations where the car is being driven at the limit of the grip of its tyres, the car cannot be turned via the steering wheel. Instead, race drivers use the brake and the throttle to force a car round a corner.

“What human drivers do consistently well is feel out the limits of the car and push it just a little bit further and that is where they have an advantage.”

Researchers have already begun work on learning what the best human drivers do with a car’s controls, incorporating the counter-intuitive movements into new algorithms for the car.
In theory this will overcome the drawbacks of current vehicle stability and assistance systems, says Gerdes, allowing manoeuvres the best drivers use to get out of trouble.

No comments:

Post a Comment