Christian Hubicki
speaker
633 appearances
1 recordings
1 series
first heard Jun 2026
last heard 5 Jun
Christian Hubicki’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 1 in all, peaking in Jun 2026 with 1.
Appearances
Guess what?
You probably do if you have a cell phone, right?
It's the thing that determines if you're in portrait or landscape mode because it's determining the acceleration of gravity, you know, and which way it is down.
So it also senses all kinds of accelerations.
Like if you ever were in a car and you fell asleep and you suddenly felt like the car come to a stop, like you've been parked, right?
You felt that inertia.
That's kind of what an accelerometer does.
That's acceleration.
you can integrate that up to position using calculus, then you can know where you are.
And you combine these two things together, the visual and inertial, you can get some pretty good estimates of where you are.
That's half of it.
The other part is deep reinforcement learning, and I'm sure this has been covered many times on the podcast, right?
That basically, once you know where you are...
And what do you do?
And so what they did is they trained this drone in a computer simulation.
They built a computer simulation of the drone, and they would try 100 different drone races at a time.
And they would, in reinforcement learning, you give it what's called a reward signal.
What's good and what's bad, right?
And that's the art of reinforcement learning.
Part of it is saying, hey, what's good and what's bad?
Showing 501–520 of 633 · page 26 of 32
← Previous
Next →