I've spent the last six months replicating the paper "Champion-level drone racing using deep reinforcement learning" and now I'm writing down the blog posts I wish I had along the way.
Any feedback is welcome, especially as I'm a bit unsure if I struck the right balance between being concise and not requiring too many prerequisites.
Also if you're working on RL and robotics (especially aerial), let's connect!
dvh•2h ago
Isn't it just bi-copter?
palata•54m ago
No, it's a quadcopter setup, but simulated in a 2D world (I guess for simplicity). A bi-copter would require tiltrotors, which is different.
avidiax•1h ago
I assume you are going to start introducing all the 2nd and 3rd order effects? One big one is ground effect, and another is vortex ring state/settling with power and the related translational lift, and the props themselves have p-factor and the dirty air effect for the rear props.
kabir_daki•1h ago
Physics simulations from scratch are great learning projects.
Did you implement your own PID controller for stabilization?
That's usually where things get interesting — tuning the
gains without it oscillating to death.
daww•2d ago
I've spent the last six months replicating the paper "Champion-level drone racing using deep reinforcement learning" and now I'm writing down the blog posts I wish I had along the way.
Any feedback is welcome, especially as I'm a bit unsure if I struck the right balance between being concise and not requiring too many prerequisites.
Also if you're working on RL and robotics (especially aerial), let's connect!
dvh•2h ago
palata•54m ago
avidiax•1h ago