And, with that, the feeling that Amazon.com will snap this company up for cash and stock is almost palpable. Forget drones, that's PR stuff for now. Uber is the real PTTH (product to the home) service.
And, with that, the feeling that Amazon.com will snap this company up for cash and stock is almost palpable. Forget drones, that's PR stuff for now. Uber is the real PTTH (product to the home) service.
Robo-Uber as I like to call it is all very exciting and transformative. And I have little doubt that it will happen someday. But we're talking decades according to many people who really know the topic. For example, John Leonard who was on the MIT DARPA Challenge team has said "I don’t expect to have truly driverless taxis in Manhattan in my lifetime." http://digitalcommunity.mit.edu/community/featured_content/b...
Next time you drive in a city, think about all the little decisions that you're making constantly. I'd love to have a service like this but it's not going to happen within a time horizon relevant to any current financial investments.
They're solving autonomous vehicles for known environments that are mapped down to extremely high resolution. Once you have that solid ground truth, adapting to changing conditions on top of it is far easier.
Which is why they've already logged 1.1 million kilometers of autonomous driving, including city streets with high complexity.
What the more traditional autonomous vehicle researchers didn't/don't appreciate is that it turns out to be easier to make the whole world[1] into a well-known environment than make a car than can handle unknown environments.
[1] more precisely, the parts of it that the average person would actually drive on, which is much smaller and simpler.
I fully agree. And it's the reason we could probably see fully autonomous operations on designated sections of limited access roads in the relatively near future. The unknowns in that environment--mostly other vehicles--are relatively well-defined.
The reason for the skepticism about more general self-driving without even a human backup available is that even if the roads are precisely mapped the environment in a city is still largely "unknown." (Although obviously some cities are more complex than others.)
"The University of Michigan, along with the local government and major automakers, is building a model town to test a system of self-driving "connected" cars.
.. It's laying the groundwork for the real-world system planned for 2021 in Ann Arbor."