00:00Thank you again for this low-tech part of innovation here on the ground at VivaTech.
00:05Now we're going to switch to a totally different story, super high-tech by the way,
00:10the future of computer vision for autonomous driving.
00:14I'm sitting live with Julius Kummerle, you're the CEO of SafeAD.
00:19You're definitely on a mission which is shaping the future of computer vision for AD.
00:26How do you intend to achieve this mission?
00:30Yes, pleasure to be here.
00:32So, yes, like the whole autonomous driving space, this is not a challenge to solve alone.
00:37So this is a very open community, like software, hardware players,
00:45and they have to play all the same game to really win that for everybody.
00:51And so where we position ourselves is as a very agile, flexible and quick startup
01:04and also take the newest hardware platforms and bring them together very quickly
01:11to a really well-developed prototype that we then put to the OEMs and tier ones
01:20and, with their help, make a product out of this.
01:23Wow.
01:24Well, so what you're building is a high-definition map.
01:29Does it sound as simple and straight to the point that it is?
01:32So high-definition maps is a really important topic in the whole autonomous driving space.
01:38This has to do with safety.
01:40So safety has, like, primary importance for the whole autonomy space in general.
01:48And so these maps, they are very, very precise and giving additional information
01:55that the car has a prior understanding of the scene,
01:59which it then combines with the online data collected
02:03and to have a really, really redundant and safe system.
02:09And so these HD maps, yeah, these we are providing.
02:13And a big challenge now, really, is to scale these maps up.
02:19And you have lots of technology in your hands to actually do that.
02:23LIDAR, RADAR, AI, 3D, name them.
02:26Exactly, yes.
02:28So, you know, the big challenge is to reduce the technology,
02:34especially the sensors, to a minimum to make the cars very cheap.
02:39So this is the big challenge.
02:41And so our approach is really to get the best out of the video data.
02:47Yes, LIDAR is a nice technology, but it is expensive.
02:52And computer vision with new results from the community,
02:59from the research community, there's a lot more possible.
03:02In addition, with the increased computational power we have on the edge, on the car,
03:08where you can do crazy things.
03:10So I do understand you have to crack actually many different challenges,
03:14how to make the technology, I mean, like efficient,
03:20actually happening and affordable at the same time.
03:23Now, autonomous vehicles were pure sci-fi a while ago.
03:28They definitely do exist, like now in this very life,
03:31already operating under very controlled conditions from metro systems to space flights.
03:39I mean, autonomous vehicles are already in our lives.
03:41Yes.
03:42So this is exactly the point.
03:45We have also some robo-taxis in very constrained regions.
03:50And the really, really big thing now is really to scale the technology up.
03:58And so there are two different ways.
04:00One is hitting directly the fully autonomous vision.
04:05The other thing is step-by-step making it assistant and then autonomous, really autonomous.
04:13But the real value lies in scaling it up.
04:16Then you have the real value proposition for humanity.
04:21Well, I mean, another question for humanity is definitely autonomous cars.
04:26I mean, there are still many challenges to be facing in the growing and making of complete autonomous cars.
04:34There's different levels when it comes to their autonomy.
04:37We're about to actually achieve level four, if I'm not mistaken.
04:42The last level is five.
04:44Do you think we will come up and overcome the challenges to make it to level five someday?
04:50Level five?
04:52Nobody can answer that question, I guess.
04:55I hope that we at some point achieve it.
04:58That will save humanity a lot of time.
05:01We also have the vision zero, so no fatalities from traffic.
05:09Anymore.
05:10And of course, this is the big vision.
05:12Nobody can answer it when this will really happen.
05:16Well, actually, thank you so much, Julius, for your perspective on autonomous driving.
05:22And of course, what you're bringing to the market with SafeAD.
05:27Thank you very much.
05:28Thank you again.
05:29And now, Luke Brown is waiting for us with the winner of the LVMH Innovation Awards, Asna Koudra.
05:36Bye-bye.
05:36Bye-bye.
05:36Bye-bye.
05:36Bye-bye.
05:36Bye-bye.
05:36Bye-bye.
05:37Merci.
Commentaires