How Pittsburgh Became Uber’s Driverless Testing Ground | Hacker News Reader

How Pittsburgh Became Uber’s Driverless Testing Ground

nytimes.com

70 pointssethbannon41 comments

I can't tell from the picture, but their test track looks like where we used to test self-driving cars when I was a graduate student at Carnegie Mellon. Except when I was doing it, it was basically dirt (most of it not even roads) in a super-sketchy part of the city. The Roundhouse, where we kept the cars, looked like an abandoned building - in fact, and I am not exaggerating, 1/3 of the roof had collapsed in on the building.

It'll be interesting to see what impact this has on the city of Pittsburgh. Mayor Peduto is definitely a progressive guy who is very gung-ho about reinventing the city.

Here's another interesting fact - almost none of the initial 40 engineers that Uber poached from NREC had any previous experience with self-driving cars. Those guys mostly went to Google, and a handful created a startup called Ottomatika which was acquired by Delphi.

Did you do any tests at Allegheny airport?

I never did, but I don't know too much about what they did after I left. We did most of our testing on a construction site near a defunct steel mill. Otherwise we sometimes drove around the city, usually Southside as it was nearby and great for testing. There was a media release about driving the car to the Pittsburgh Internatioal Airport.

http://www.cmu.edu/news/stories/archives/2013/september/sept...

I wonder how people felt when the automobile revolution was taking place. Probably a mixture of all the same things people are feeling now, human nature being what it is. Could probably look up the newspapers of the time to get a feel.

I like that bit in the article where it says the local politicians have stayed out of the way - that's the sort of thing politicians mostly should do.

For me the real turning point will be when the first passenger hails an autonomous vehicle that doesn't have a human behind the wheel, I'll be satisfied with a remote operator, because that will be truly novel as current laws asume a human behind the wheel.

For a perspective on how the automobile revolution was viewed: http://99percentinvisible.org/episode/episode-76-the-modern-...

The auto revolution was awful.

We can hope the self-driving revolution is something of a counter-revolution, turning the auto into a better-behaved tram, allowing driving time to become free time, etc.

> a better-behaved tram

I have always felt this way. If everyone driving looked at the bigger picture, it would be a much more efficient and pleasant place. Road rage, traffic jams, death, all these things are mostly the result of human error and poor mental judgement.

I agree that mostly it was people, not the machine. That's the basis of my love affair with autonomous cars - remove humans from the equation and watch casualty numbers plummet.

https://www.researchgate.net/publication/236825193_Street_Ri... covers almost precisely the same material as that podcast episode, for those who prefer to read.

The paper is from 2007 and podcast 2014, neither covers potential of ending individual car ownership and human drivers for getting the streets back for people - not that they should have - the history is more than enough material.

The concentrated interest of car dealers and their leverage as advertisers in newspapers is mentioned by the paper (and I think podcast) as important for driving people from the streets. How close are we to destroying local dealerships now? Why has the internet not done this yet? I've seen headlines about Tesla fighting with dealerships but haven't read. Can we end this powerful but reviled sector?

If a car can drive itself around Pittsburgh, it can drive itself anywhere.

Seconded. Of all places in North America PGH for me is the most stressful place to drive. The use of GPS for someone not familiar with the city is obligatory and even then you need to stay exactly in the right lane at the right time. Other places in the world can present different type of stress like hectic streets of India, China, or South East Asia, but there you have an option to slow down and do a lot of honking. Not so much in PGH. It is combination of three rivers, hilly terrain, complicated interchanges and numerous bridges, but knowledgeable locals just vroom by.

I am sad about the slow but steady loss of the few remaining cobblestone streets around Pgh.

But have you seen the wooden street in Shadyside? That's still there.

http://pittsburgh.cbslocal.com/2013/06/05/construction-begin...

At least they are still keeping the brick road in Oakmont. I had to check online to make sure.

First time I ever went to Pittsburgh, driving w/o a gps, we were trying to find 6th street. We were on an avenue hoping to see numbered cross streets.

We ended up at the intersection of 5th avenue and 6th avenue (https://goo.gl/maps/MSJC93Z4rLC2). I remember just laughing. Which way do we go? Any way will do I guess.

This was in Austin but I always got such a kick out of it :)

The intersection of Barton Hills Drive and Barton Hills Drive:

https://goo.gl/maps/HaA3iSvxZm22

Your experience in North America is mundane compared to what the world has to offer. Quito, Rio, Lagos and the other mega-cities of the developing world are so much more chaotic. I look forward to driving in the big cities of India one day :)

No doubt.

Between all the bridges, narrow roadways, winding hills, inclement weather and mislabelled streets we (Pittsburgh) present a big challenge for human drivers let alone robots. Glad to see all the Uber SD cars buzzing around shadyside the past few months though.

Eh. Test it in the rural mountain roads of West Virginia.

And Boston while you're at it.

Singapore already has some autonomous taxis in test.[1] They have a backup driver on board, as do Uber and Google.

[1] http://mashable.com/2016/08/25/driverless-cars-starts-singap...

step 0: poach an entire department from CMU.

To be clear, they hired most of their talent away from CMU's National Robotics Engineering Corporation, CMU Robotics Institute's commercial arm -- not actually faculty.

They hired some faculty as well, although the vast majority were senior personnel at NREC.

I lived in Pittsburgh from 1989 to 1991, very close to Carnegie Mellon University. One of their technical departments (CMU) had a self-driving car (really a van like a UPS van) even back then ; we saw it driving around our neighborhood sometimes. I think it always had an operator standing by inside it, for override.

That was the NavLab, an autonomous vehicle with a crew of 5. I was in it once when I visited CMU. Rack-mounted workstations in the back with up to 3 operators, a driver, and someone with a keyboard in the passenger seat. Top speed about 5 MPH. First 3D LIDAR.

Good ideas, nowhere near enough compute power or good enough sensors. Or precision GPS, which modern self-driving cars lean on too much.

> That was the NavLab, an autonomous vehicle with a crew of 5.

And a "Nobody on board" yellow diamond sign in the window. Pfui.

They had to start somewhere. But progress was too slow for years. Tony Tether at DARPA came up with the DARPA Grand Challenge because he thought the DARPA-funded automatic driving projects needed a serious kick in the ass. It worked.

(The reason it worked is not well known. Some major research universities were quietly told by DARPA that if they didn't do well in the Grand Challenge, their DARPA funding would be cut off. Suddenly entire CS departments were being devoted to automatic driving. Levels of effort went from 5 people to over 100, and not at DARPA's expense.)

Non-mobile link: http://www.nytimes.com/2016/09/11/technology/no-driver-bring...

Thanks, we updated the submission from http://mobile.nytimes.com/2016/09/11/technology/no-driver-br....

Why is it OK to protect American IT jobs with limited (or expensive) H-1Bs, but it's Ok to replace commerical drivers with autonomous vehicles? Why is one kind of protectionism Ok and the other overdue for being squashed?

Because regulating who comes into the country is generally seen as the government's job, but regulating what labor-saving technology is used is generally not.

And government can always tax automation to provide an income to those who are redundant if other jobs are not available.

Technology replacing jobs doesn't directly benefit outsiders. Foreign labor replacing local labor feels unfair because foreigners are outside of the local people's psychological tribe. That explains why people are happy to have children even though those children are bound to grow up and take old people's job, possibly forcing them into early retirement and poverty. At least the job is being taken by another member of the same in-group so it doesn't feel so painful.

It would be a fair question, if the H1B programme actually did protect US IT worker jobs but look at this list:

http://www.epi.org/blog/top-10-h1b-guestworker-offshore-outs...

The top 10 users of it are all "consulting" firms.

One doesn't involve a flood of immigrants basically working as slaves.

People don't care about immigrants working like slaves because that's what they'll be doing in their own country anyway.

Do you really not see a distinction between allowing sewing machines and allowing unlimited immigration of sewers?

Because it's not actually about protecting jobs, it's about white supremacy.

I've seen a lot of white truck drivers if it makes you feel better. I'm even related to one.

Reply on news.ycombinator.com