Chris Urmson: How a driverless car sees the road (TEDTalks)

[FrodosPet]

Published on Jun 26, 2015

Statistically, the least reliable part of the car is ... the driver. Chris Urmson heads up Google's driverless car program, one of several efforts to remove humans from the driver's seat. He talks about where his program is right now, and shares fascinating footage that shows how the car sees the road and makes autonomous decisions about what to do next.

[FrodosPet]

1. Google Self-Driving Car Project | SXSW Interactive 2016

Published on Mar 12, 2016

Everyone's talking about self-driving cars these days, but how can you differentiate between hype and reality? In the six years of Google’s project, its vehicles have self-driven over 1.3 million miles, racking up the equivalent of 90 years of human driving experience. Google says its cars can now handle the vast majority of everyday situations it finds on the roads, but what does the path to a driverless future look like? How could it be that self-driving cars be both three and 30 years away?

Urmson shares his stories from the front lines of building the world's first fully self-driving cars.

[longship]

2. No hope here for driverless cars.

Most of the roads are unpaved. And there's snow on them for a significant portion of the year.

Sorry! I want my hands on the wheel.

And Grass Lake Road here... a self-driving car would very likely end up IN Grass Lake.

[Travis_0004]

4. They will figure it out

It will take a bit longer, but there is no reason a computer can not drive in snow, or gravel, or a dirt road.

[longship]

5. Again, no hope here.

First, how in the fuck is a driverless car going to navigate a narrow, unpaved road where there is not only no shoulder, but a ditch on the side. I might add that these roads have occasional rather wide farm implements traveling down them.

And then there's fucking Grass Lake Road, a twisty turny, very narrow road. At Grass Lake, the edge of the road is less than 3 feet from the lake. At one point, the level of the road top is actually below the level of the lake. There is always water on the road there. And upon an oncoming vehicle on that road, barely wide enough to allow such a thing, I want control of my vehicle. In winter, especially so.

In winter, all the roads here are snow covered, for months! There are no visual hints as to where the road goes. One has to be aware of where it goes.

I want nothing to do with driverless cars. They are nothing but a Matrix wet dream by people who live in sunny California. As to that, try getting your driverless car through Cajon Pass during a winter snow storm! That would likely end up being mayhem.

[Rhiannon12866]

3. K&R! Fascinating presentation, have always wondered how these worked...

I used to wish for a self driving car when I had to drive to work in a blizzard. Thanks for posting!

[FrodosPet]

7. They are NOT good in blizzards

The micro-navigation: the exact location of the car within a lane is largely done through lane markers, road edge, etc. In a blizzard, all of that is gone. The snow is going to be more reflective than almost anything else to LiDAR, and the optical cameras and RADAR sensors need to be free of snow.

Being winter ready is one of the hardest challenges that self-driving cars face.

[Rhiannon12866]

8. Thanks!

That's not good for me, since I live in NE NY so we get a lot of snow and that's when I really hate to drive. It's also the hardest challenge that those of us here in the North Country face, LOL, but I appreciate the information. I really like the idea of a self driving car, with all the variables they've taken into consideration and that was a fascinating talk. And even in good weather - maybe especially in good weather - a self driving car has to be a whole lot better than a great number of the drivers I encounter out there...

[chapdrum]

6. From the little search engine company that could.

More nonsense.