Audi’s Self-Driving Car Hits 150 MPH on an F1 Track | Hacker News Reader

Audi’s Self-Driving Car Hits 150 MPH on an F1 Track

wired.com

24 pointsdenzil_correa14 comments

Not sure how impressive this is. Isn't this just processing power and different performance profile data? If they could solve the driving in a busy urban environment in torrential rain I would be more impressed.

It's impressive. Making anything completely driverless working at incredibly high speeds is challenging.

Much easier to test a race track than a busy street too.

The impressive part about Audi's technology is that it doesn't use very detailed track data to achieve this but is mostly depending on a variety of sensors to anticipate to current road conditions 'on the fly'.

From the article:

While the RS7’s computer had a detailed digital map of the track and was following an optimized path through it, Audi did not preprogram things like torque or steering.

"Audi did not preprogram things like torque or steering."

And that was my point, unlike BMWs famous testrun (with Jeremy Clarkson) where the car was programmed to steer and give gas for every inch of the track, the Audi is actually able to make those decisions on the fly with the information it gets from its sensors.

EDIT: So yes it does use GPS data, but does not rely on it as much as others (including Google).

You missed the part about the F1 test track?

This obviate the need for deeper GPS depenendence.

You have a 3D track designed with graduated radius turn transitions, smooth topological surfac transitions, and predictable fiction levels. All in a general layout that is suited to safety at speed.

Try running this car at 150mph on glendora mountain road, at night, during a bout of santa-ana cross winds.

To put the 2:10 minute round time in perspective http://en.wikipedia.org/wiki/Hockenheimring#Record_laptimes

I wonder why they didn't use a http://en.wikipedia.org/wiki/Audi_R8 if the goal was racing?

Not really fair to compare F1 and a street car, though.

My guess is it's because the current R8 has been in production since 2007 without any significant upgrades and therefor lacks the tech to achieve full autonomous driving, also the RS7 tops R8s performance (out of factory). As an owner of the latest model A6 (which the A7 shares most of the components with) I can attest that the factory car is already almost capable of driving itself (but I guess purposefully limited because of legal concerns of taking it on public roads). For example, the A6/A7 is factory equipped with camera's all around, radar and ultrasonic sensors that allow for things like 'staying in it's lane' and adaptive cruise control.

I think the RS7 has the engine but not handling performance.

The R8 was based on the Gallardo chassis, whose successor has already appeared. Presumably a Huracan-based R8 successor can't be far behind. Unfortunately RWD sports cars still out handle the Huracan, though by a lesser margin than previously.

They're likely not working with the R8 because their self-driving capabilities are a marketing tool to compete with the likes of the Mercedes S-Class.

The R8 is also at the end of its product life-cycle so they're probably trying to keep it out if the limelight as much as possible.

And yes, formula 1 comparisons are unfair, but if they're actually within 30 seconds of DTM cars that seems pretty impressive.

>And yes, formula 1 comparisons are unfair, but if they're actually within 30 seconds of DTM cars that seems pretty impressive.

The comparison that's really missing is to put a good racing driver in the same car and same conditions and seeing how much faster he can go. We're not testing the car here after all just the self-driving tech.

Here's more comparable lap time figures:

http://fastestlaps.com/tracks/hockenheim_gp.html

Still clearly some room to improve, but damn good for an autonomous car.

Reply on news.ycombinator.com