Saving Lives in Hazardous Spaces: The Flyability Story with Adrien Briod

Yeah, that's exciting.

People that wouldn't be in the industry, I don't think they would really appreciate all what you just said and the complexity.

And people might say it's already solved, drones are around, drones are flying.

What else can you do there? And let's actually then separate those two things, the hardware element and then the software element.

On the hardware element, especially kind of coming from my comment before, we have drones, they're flying, like how much innovation can actually still happen? What are the technical limitations today, maybe starting from that angle, from a pure hardware sensor, motor, whatever you pick here? What are the limitations today where you still cannot get the full potential out and they are still working on better versions, better technologies? better durations more more power more whatever it might be to get to the ultimate value for your customers

right I think actually it seems already that things are very advanced today but the way I see them is that we're just getting started in robotics in terms of all of the intelligence that still can be brought to robotics.

I would say the challenge, of course, is working with the physical world.

There's so much uncertainty, unlike code running on a laptop, right? That just has a very clear hardware to work with and very limited interaction with the physical world.

So I would say in terms of constraints, there's been tremendous progress on sensors and onboard compute in the past 15 years.

When I remember at the beginning of my PhD, we were really struggling with In terms of hardware, we wouldn't even dream of the hardware we have today.

But I still think that in 15 years, we'll see as much progress in terms of sensing.

Right now, there's incredible progress on the LiDAR front.

Also, I benefit a lot from all the progress happening in terms of battery technology.

Of course, the massive turning point is with AI.

This is technology that will transform robotics and there will be both hardware and software innovations that will unlock so much.

Like when I mentioned that our drones are mostly manual, at the moment, I think AI will go a long way into helping drones become really like assistants to our users that receive instructions, not joystick commands, like understanding the world and taking decisions about the the world, it's really bringing the agents that you know today that are stuck on your 2D screen, maybe writing code for you or doing some actions for you, booking hotels.

This is bringing these capabilities to the real world.

You'll be able to prompt the real world, right? So I think we've not seen anything yet in a sense in terms of robotics and combination with AI technology, which I think will be a massive, massive unlock for in the next 10 years, not just for drones, right? For anything interacting with the physical world.

For sure.

No, for sure.

And I think that's in this buzzword category of physical AI, right? Everything that's robotics and that uses AI and connects a bit the real world with the digital world, as you described it, is now summarized under this physical AI category.

What does that really mean for you today? And you mentioned sensors, of course, more data, you can create other decision trees and AI can react differently if they have more data.

All of that is clear.

I'm curious on physical AI and from two lenses.

One is, do you have the right hardware in place and drones are ready to leverage the technology? what physical AI is and maybe you should describe what physical AI means for you but do you have that already or do we need the next generation of sensors first to really get the physical AI benefits like autonomous actions for example what you described or I'll be pretty good on the hardware side and we can then I already get the benefits out of that physical AI.