Separate names with a comma.
Discussion in 'Autonomous Vehicles' started by Jeff N, Aug 18, 2016.
Uber reveals a fleet of human-supervised self-driving cars available to customers in Pittsburgh
Just saw that too. My buddy who works there told me to be sure and check the Uber news this morning.
Very exciting, can't wait to see reports on how often the driver needs to take over. And also what the next city is going to be and when.
It's got a LIDAR on top of a SUV, looks pretty tall, will it be able to get in a normal garage? I'm also wondering how much that costs. Volvo did a smart thing to partner with Uber and release it where not many people will complain the lack of a roof rack mount. Maybe Tesla should have done the MX FWD as a special edition for Uber/taxi/limo.
The news reports sound more like a press release than anything substantive.
Looks like they didn't Photoshop that image quite right. The charge port for the XC90 T8 should be on the left side. Maybe they'll be using autonomous UK versions on the streets of Pittsburgh? Maybe they replaced the charge port with side radars?
Pretty wild but I won't get excited until I find out more about how good this system works, esp. at what speeds.
Google car limited to 25 mph last I heard ? That's not the taxi I want, unless the speed limit is that or lower of course.
It sounds like a test fleet but unlike Google that only allows employees in the fleet, Uber takes customers with them.
My impression is: the test procedure goes further than Google's because it requires at least 2 employees for each trip: a Safety Driver who can do a manual driving and a Co-pilot who takes notes.
Not as tall as in this picture:
I think garage doors come in either 7' for 8' high so may be they fit Uber after all.
It will be very interesting to see how autonomous driving fares in real-world Uber situations. As an Uber driver, some of the most challenging driving I ever do is during Uber pickups in hectic urban settings. Especially after ball games and concerts, or during the Saturday night club scene, the task of dodging pedestrians, trying to spot the rider, finding a legal place to pull over, and trying to keep pedestrians and bike riders from being "doored" is a challenge that will take a long time for autonomous systems to solve.
That said, I've been driving Uber riders on driver-assisted autopilot for months!
Not really.. they limit the car so much when no-one is in the car.. when someone is in the car, they don't have limit.
For the 2 people.. i would say that theyr system isn't good enought or the recording system or similar, since everything can be recorded and can be replayed in a more comfortable enviroment why would you do your check on the road?
Maybe they don't want to do the intensive teaching they do at google so they simply put a driver with no experience behind the weel and a semi-programmer on the other side.. what a waste..
It's just interesting to see what will come out of it, surely uber would make a lot of miles and in stressing enviroment
This is just a clever way for Uber to get PR from essentially an engineering effort. The cars in Pittsburg next month might not even be as good as Tesla's AP now, but the two employees in each car will be gathering tons of engineering data to build their own AP.
And I agree with Bizjet, Uber driving is the most challenging urban driving there is. Pickups and dropoffs at crowd scenes are common. My prediction: those engineers are going to find out that urban driving is really hard to automate!
Not only that they have to pay 2 employees for each use, they don't get any income from their customers (free autonomous rides until further notice.) No income, only expenses!
Thus, it sounds more like a research orientated project and it is not profit oriented at this period.
Volvo Cars and Uber join forces to develop autonomous driving cars
"Both Uber and Volvo will use the same base vehicle for the next stage of their own autonomous car strategies. This will involve Uber adding its own self-developed autonomous driving systems to the Volvo base vehicle. Volvo will use the same base vehicle for the next stage of its own autonomous car strategy, which will involve fully autonomous driving."
It doesn't sound like Volvo would work on a joint autonomous research project with live customer riders.
Anyone knows how much an Uber car drives every week? Because a fleet of 100 cars is just a small sample of data, compared with the tons of it Tesla gets every day from they AP enabled cars.
And that is not something you can just brush aside. We live in a time where data is power. Just ask Google.
You are right about Tesla who gets the most data, more than longtime Google even.
Eventually, Tesla will surpass its semi- and autonomous performance but for now, Tesla system still has many limitations:
It can't see a giant white tractor-trailer against bright sky.
It doesn't brake for a Lateral Turn Across Path (LTAP) big tall 18-wheeler.
In some cases, it doesn't brake for stationary vechicles when drivers thought Automatic Emergency Braking should.
In some cases, it can't handle twisted roads and it crashes.
In some cases, it crashes on winding roads where driver thought Automatic Emergency Braking should slow down or stop the car...
All true. But do we have any system in the market that excels in those situations? Tesla is paying the bill for getting the cars in extreme situations where their level-2 (remember this) is put to the test.
It is true that there's currently no system in the market to compare with Tesla.
However, Google system has been around for quite a while and it has disclosed all accidents which have been so minor.
The latest was a swiping with a bus in which "programing to act like a 'normal' driver" was blamed (illegally lane sharing in California when it saw a long waiting line of cars in front and passed those cars within the same lane to get to the front, then unsuccessfully cut off a bus in the same lane because that is what buses do to your cars, don't they?)
Google has been doing well with construction zones while Tesla doesn't do too well with cones as seen in this thread.
Google has covered winding roads as well.
It can map in real time 360 degrees of different images from trucks to bicycles and bicyclist hand signals as well, so I expect it would not have any problem with Lateral Turn Across Path (LTAP) big tall 18-wheelers.
so when the human is deemed no longer necessary, what happens when a pedestrian steps off the sidewalk.
does the car swerve into the next lane to avoid the pedestrian and cause an an adjacent vehicle to swerve into oncoming traffic into a fatal rta? or does it just kill the pedestrian because it can't know the consequence of swerving out of lane.
I will take some convincing away from Autonomous == Folly on the public street.
That depends on the ethics of a programmer to decide which one must die if there is no other choices.
In a traditional manual car situation, I would spare pedestrians' lives because the driver is the one who brought in a 5 thousand pound deadly car into the scenario.
So, may be occupants of autonomous cars will sign an acknowelegement that pedestrians' lives will take priority over the occupants of an autonomous vechicle.
Check out post #23 (Carnegie Mellon Driverless car in action, Sept 2013):
Why Tesla will win the Autonomous Driving race
Thanks to @cluster for posting!