watching you watching me

[cam235]

A Google StreetView car just drove by. I presume the cameras are looking out of the blue pod on top, but I'm curious about the white-capped black cylinders tilted downwards in front and back just under the camera pod, any idea what that is? Another type of camera, or maybe lidar scanner?

 

[marku2]

Lol California smog detection

 

[Fastb]

According to wikipedia:
"

• Laser range scanners from Sick AG for the measuring of up to 50 meters 180° in the front of the vehicle.[34] These are used for recording the actual dimensions of the space being photographed.
• LIDAR scanners from Velodyne were added in the 2017 update. Mounted at 45° to capture 3D depth information, and used for additional positional information.[33]

"

 

[cam235]

Thanks! I should have known it would have its own wikipedia page.

Reading the referenced articleon the Streetview LIDAR sensor, looks like they may be preparing data needed for autonomous cars:

In fact, Google's setup looks identical to Here's "True Collection Vehicle" which is purpose-built for creating self-driving car maps. Here's setup has four cameras on top and a Lidar sensor mounted at a 45° angle!

Google is certainly making a much more precise map than GPS could make alone. GPS is accurate to a few meters, while a Velodyne sensor can position you within a few centimeters. If all you want to do is locate your Street View pictures on a map though, it sure seems like overkill.

 

[Fastb]

You quoted:
"Google is certainly making a much more precise map than GPS could make alone. GPS is accurate to a few meters, while a Velodyne sensor can position you within a few centimeters. If all you want to do is locate your Street View pictures on a map though, it sure seems like overkill."

For Google Maps, sure it might be overkill. But I think Google has bigger aspirations than just "Maps".
Self driving cars will need very accurate maps. There's big money to be made. The market will be quite large, with exponential growth. Besides pure revenue, there are strategic reasons for Google to control and provide that geographic information.

 

[Mike.in.Minnesota]

A hot babe detector?

 

[Fastb]

I thought an IR cam could catch a Hot Babe

 

[cam235]

In theory, a self-driving car should be able to work without a detailed map, just like people do. But in practice, I agree it's a lot easier and more practical with a very accurate map, and that is likely what google is up to. The LIDAR is probably for the location of the road curb accurate at cm-level relative to camera images of the lane markings and other landmarks, which previously had to be guessed from the camera image alone.

 

[Kawboy12R]

I've wondered what level of GPS corrections they use for their road maps. WAAS isn't that great and has reflection issues. Omnistar has major issues with signal blockage (drive close to a tree or building and you're done), traditional RTK or cell-assisted isn't everywhere, but companies like Trimble have come out with satellite-driven RTX solutions (RangePoint, CenterPoint) that are nearly as accurate and repeatable as dedicated RTK with base stations (repeatable to about an inch). Good for just about everything except water work like drainange. Definitely accurate enough to map roads and keep self-driving cars in their lanes.

 

[cam235]

Good question about GPS corrections, I have no idea what Google does (n addition to terrain shadowing there is also intentional jamming*). There are two different map problems: 1) making a good map with whatever data you can get beforehand, and then 2) using the map with live sensor data to locate and drive a car in real time.

If your mapping car has at least occasional good GPS fixes, and you also have a lot of really high-resolution visual + LIDAR data (and a lot of CPU power), you can piece it all together like a jigsaw puzzle using self-consistency into a map with very good local (relative) accuracy, and good-enough absolute accuracy based on GPS. Likewise the self-driving car could work even with iffy GPS, if visual and other sensor data could be used to key into the map well enough.


* It's an interesting article:
"...most of the jamming Gostomelsky saw was done for the sake of profit or paranoia. Gostomelsky lives near Interstate 476 in Pennsylvania, and would regularly detect GPS jammers in use as tractor-trailers went through the toll booths. Because the toll-taking for commercial trucks relies on GPS tracking, they can avoid paying through jamming. If a $45 device made your daily commute free, you too might be tempted to commit a federal crime. [...]

Other jammer use-cases that Gostomelsky saw were more mysterious. A couple of jammers who were flagged by his system in Philadelphia had unusual electronic signatures emanating from their cars. They appeared to have both smartphones and burner phones, based on their communication with cell towers, which he was also monitoring. He was intrigued, so he started moving his listening stations around to figure out the cars’ driving patterns, to zero in on where these jammer users lived. He wound up tracking one of the jammers to a house in New Jersey. He went digging through public records to see who lived there and figured out the property owner was related to a police officer.

He is now convinced that some of the people using illegal jammers were police officers. They may not have seen a 2014 warning from the FCC that the prohibition on jammer use applies to “state and local government agencies, including state and local law enforcement agencies.”

“I think they’re using GPS jammers on their vehicles so that they can’t be tracked home,” Gostomelsky said. An undercover officer, for instance, might worry about a suspicious target sticking a GPS tracker under his bumper to find out if he’s a cop. " https://gizmodo.com/jamming-gps-signals-is-illegal-dangerous-cheap-and-e-1796778955

 

[code2]

Lol California smog detection

Best answer right here