Is self driving still hopeless?
Waymore CEO said they don’t expect level 5 is feasible. Uber basically failed. Tesla... don’t know much. Any other play in the field you guys know? Thought it’d be a great innovation but didn’t expect the failure came so quickly.
Have high expectations for Tesla. They are doing really well just with cameras
They'll fail miserably without LIDAR. Yes, it's possible to have a self driving car without it, but it's just increasing the difficulty of the problem for no real reason. Waymo has a five year head start on their competitors, they have LIDAR and massive datasets, tons of ground truth data from street view, and yet they're barely doing point to point navigation between predetermined pick up and drop off points. There's a reason why every self driving car company (except Tesla, if you consider them to be) uses LIDAR, and it's not because they like making their cars more expensive.
Can’t tell if you’re being sarcastic or not about Tesla. Their record speaks for itself: * Broadsiding a semi truck and killing the passenger * Ramming the back of a bright red fire truck * Driving into the gore point at the 101/85s split, ramming the wall at highway speed and killing the passenger
You don’t actually need full level 5 to get most of the benefit. Waymo is running paid trips with L4. Uber hasn’t failed either, ATG is still running quietly and will run paid trips when ready.
Don’t think Waymo will be disruptive. Each of those cars cost north of $250k. If they do go into the market, it will take a long time for them to break even, even if the car doesn’t have a driver.
They’re also running very low volume, a few hundred customers. They’re accepting money for trips and getting people places with no driver though which is kind of big.
Where's the downvote button on this app? You're making some implicit and wrong assertions with your question. There's 30+ viable players and nearly every automaker is throwing vast sums of money at it. There's a ton of progress. I don't agree that level 5 is infeasible. Please supply the source for the Waymo quote?
L4 /L5 will only work in geo fenced cities with HD maps. They will never work in the suburbs and if they do this means we have a real AI and machines will take over.
No reason SLAM can't step in where HD maps are not available. The reason companies are leaning on maps right now is that it simplifies the problem, not because it is absolutely necessary. It may be tough right now but that does not make it impossible.
State of the art SLAM techniques are still nowhere near perfect... SLAM really works in a constrained set of environments
There's a lot of work to do. Other than demo setup or a very controlled route like Waymo tries, the technology is still very much under development. It could be very well 5 years for L5 and in a very limited scope (e.g. very specific routes and under certain time and traffic conditions AND with remote assistance and supervision).
As autonomous displace human driven cars the overall behavior of traffic will become increasingly predictable. This makes the problem easier to solve.
The L system for autonomous driving is a quality meme
First learn the name of company and then make uneducated statements.
Problem in the US is the infrastructure. Research says 70% of the streets are not as good marked as needed for selfdriving
Look up SAE STANDARDS for autonomous systems. L5 means no steering wheel or human control. L1 is your car right now. L2 is Tesla and GM Super cruise.
Cruise Automation is doing terrible, their cars can barely drive in a straight line and have been involved in many accidents.