This dude needs a dumber version of himself to explain his project to Dumber like me. I have followed Tesla’s FSD and admit that I doubt it could navigate London streets. I get that his software is smarter and more intuitive re what other drivers could do whereas Tesla’s FSD seems to take its actions in a more awkward step-wise fashion. Would love to see longer driving demos. If I understand correctly, he is getting good results using less data than Tesla by making the core intelligence of his software more predictive as opposed to seeing and reacting which I think Tesla FSD does more. Would like to see this discussion and get some Tesla FSD users to try this program. I DO agree that all the other software companies using Lidar have it all wrong and try to rely on smarter sensors to make up for less smart neural networks. I was very sceptical but think his efforts are extraordinary. Congrats esp for a Kiwi which is where I am now and although I driven a Tesla, think a rollout here is many years off. Would love to try his software in my car.
![[CVPR'24 WAD] Keynote - Alex Kendall, Wayve](http://i.ytimg.com/vi/AEfq5nFi7s8/mqdefault.jpg)
![[CVPR'24 WAD] Keynote - Alex Kendall, Wayve](/img/tr.png)
![[CVPR'21 WAD] Keynote - Andrej Karpathy, Tesla](http://i.ytimg.com/vi/g6bOwQdCJrc/mqdefault.jpg)
![[CVPR'23 WAD] Keynote - Andreas Geiger, Universität Tübingen](http://i.ytimg.com/vi/x_42Fji1Z2M/mqdefault.jpg)
![[CVPR'22 WAD] Keynote - Yuning Chai, Cruise](/img/n.gif)
Thanks so much for your insights
This is truly cutting edge. Wish there was a way to invest.
Very interesting work on high-dimensionality representation of the world!
Wish I can join that team
This dude needs a dumber version of himself to explain his project to Dumber like me. I have followed Tesla’s FSD and admit that I doubt it could navigate London streets. I get that his software is smarter and more intuitive re what other drivers could do whereas Tesla’s FSD seems to take its actions in a more awkward step-wise fashion. Would love to see longer driving demos. If I understand correctly, he is getting good results using less data than Tesla by making the core intelligence of his software more predictive as opposed to seeing and reacting which I think Tesla FSD does more. Would like to see this discussion and get some Tesla FSD users to try this program.
I DO agree that all the other software companies using Lidar have it all wrong and try to rely on smarter sensors to make up for less smart neural networks.
I was very sceptical but think his efforts are extraordinary. Congrats esp for a Kiwi which is where I am now and although I driven a Tesla, think a rollout here is many years off. Would love to try his software in my car.
Interesting thoughts, especially vs. Tesla FSD