It was great to have Karl visit us back at MIT, now with Oscar, to provide a nice mix of the big picture view of autonomy & specific applications of deep learning to driving data.
This was easily my favorite talk in the series, Lex. Thanks for continuing to host such great speakers time and time again. Also, Kudos to Aptiv for their contribution back to the world of academia. In a lot of ways these benchmark datasets seem to augment the open source research community. It is clear that advancements made with this dataset could very easily benefit Aptiv itself. We are all working together to build something incredible.
I’m just a blue collar tradesman, but listen to npr stories about the emerging transportation systems. Is anyone studying smart roads that can better interact with smart cars? It seems like self driving cars are being made to conform to old modalities, and new infrastructure isn’t being designed for their new uses. Is it appropriate to push out legislators to reform to public policy around the areas of transportation and technology, or is it too early to try to meet the needs of an uncertain future?
It's still too early. And the shortcut is to improve the tech in cars which has been stagnant for years. Revamping the roads and adapting them might be costly to test and costly to implement afterwards. Just my 2 cents.
I'm also just an interested layman, but it seems to me that at some point a standards body has to get together and do that, sensors and communications in traffic signals, signs, the road itself, traffic cameras, other cars. Eventually I can imagine that all cars essentially have 100% awareness for blocks or miles around them due to the network effect. If we add smartphones in to that we have some people added to the network of awareness. I can see how AI driving will quickly become many many times safer than human drivers.
@@xpkareem "2 years" later I still agree with you but I worry that people will *never* Vote in favor of creating the data, so we'll have trouble building the smart roads.
This is all very complicated and uncertain. How about this? Aging folks become unreliable drivers. Introduce autonomous cars in a setting with such old folks. Like parts of Florida. Study the accident rate. It goes down, it is a winner. It goes up, we are doing something wrong. IMHO.
It was great to have Karl visit us back at MIT, now with Oscar, to provide a nice mix of the big picture view of autonomy & specific applications of deep learning to driving data.
This was easily my favorite talk in the series, Lex. Thanks for continuing to host such great speakers time and time again. Also, Kudos to Aptiv for their contribution back to the world of academia. In a lot of ways these benchmark datasets seem to augment the open source research community. It is clear that advancements made with this dataset could very easily benefit Aptiv itself. We are all working together to build something incredible.
Thanks Tim!
Best lecture till yet. The point of working with point clouds allows for very great data augmentation is really nice ! Kudos to the Aptiv team !
I think this is the first AV company presentation that mentions actual technology behind it.
Fact.
Exactly. Most of the interviews here are just genéric pr. Those guys showed something real at least.
Wonderful piece of Information thanks
It’s a huge effort , Cost , time involved in re validating entire modules for small change in Code
I’m just a blue collar tradesman, but listen to npr stories about the emerging transportation systems.
Is anyone studying smart roads that can better interact with smart cars? It seems like self driving cars are being made to conform to old modalities, and new infrastructure isn’t being designed for their new uses. Is it appropriate to push out legislators to reform to public policy around the areas of transportation and technology, or is it too early to try to meet the needs of an uncertain future?
It's still too early. And the shortcut is to improve the tech in cars which has been stagnant for years. Revamping the roads and adapting them might be costly to test and costly to implement afterwards. Just my 2 cents.
I'm also just an interested layman, but it seems to me that at some point a standards body has to get together and do that, sensors and communications in traffic signals, signs, the road itself, traffic cameras, other cars. Eventually I can imagine that all cars essentially have 100% awareness for blocks or miles around them due to the network effect. If we add smartphones in to that we have some people added to the network of awareness. I can see how AI driving will quickly become many many times safer than human drivers.
@@xpkareem "2 years" later I still agree with you but I worry that people will *never* Vote in favor of creating the data, so we'll have trouble building the smart roads.
Though it's common for folks to click agreements to private companies I think that in general "we" generally concern ourselves a lot in re: privacy.
This is just a big thanx for providing the nusense data! Thanx.
Wondeful lecture from aptv members.
Will you interview Patrick Winston?
Excellent!!!!!!!!!!!!!!!!!!!
great!!
i hope you come to Algeria 😢
think you sir
wow very nice course
great talk !!
Cameras are built for human, lidar for machines!
you best of the course
al over your teacher not best course
only best course
i am ml engineer ok sir
sir one prolme
This is all very complicated and uncertain. How about this? Aging folks become unreliable drivers. Introduce autonomous cars in a setting with such old folks. Like parts of Florida. Study the accident rate. It goes down, it is a winner. It goes up, we are doing something wrong. IMHO.
You humans Are so funny I thought smart people came from MIT If this is the best you've got we will conquer your planet in no time