Brilliant video!

first

Young Angela Johnson Kevin Hall Ruth

How were you able to map the region first with proper loop closures ??????

can you provide the codebase of the model if available?

This is so good, damn

Love this!

The virtual experiment shown in the video was created using which software?

The virtual experiment shown in the video was created using which software?

how i can find the best value of the constante

Fuzzy Logic.....

Interesting work

One of the best PID explanations that i have ever seen...

This is probably the only video that gives a proper intuition about PID control. I wanted to develop that intuition to tune the roll, pitch, and yaw controls of my drone. None of the drone experts know how to properly explain the impact of changing P, I, and D.

the best PID algorithm tutorial I have seen. The visual aids makes it so much easier to understand and give an intuition

looks like the Pipe-Windows screen saver from windows 95 😅

daz impressive!

thats dope

The last part got me😂

good explain, but, it would be better if you explain how the reinforcement learning role

I support the dev of VPP quadcopters (in a 5-7" prop size). Because it's in my interest. I fly 3D mode FPV quadcopters, but currently with FPP. There may be an additional advantage of FPP quadcopters over FPP helicopters. I noticed that almost all helicopter FPV videos show very strong vibrations in the camera recording. Either this is due to poor attachment or it is generally the case that FPV is much more suitable for quadcopters than it is for helicopters. With quadcopters you can get very smooth, vibration-free shots. It's probably not for nothing that helicopters were never used as small camera drones.

amazing vid

very gud and simpel

for those who want to implement this in python a code here you have the simplest way to do this: class PIDController: Kp Ki Kd error_sum = 0.0 error_rate = 0.0 _prev_error = 0.0 def __init__(self, arg_p, arg_i, arg_d): self.Kp = arg_p self.Ki = arg_i self.Kd = arg_d def update(self,target,current): error = target - current self.error_sum += error self.error_rate = self._prev_error - error self._prev_error = error return self.Kp*(error) + self.Ki*(self.error_sum) - self.Kd*(self.error_rate)

Can you please name those advanced controllers u mentioned at 4:10 please

very well explained, i love when the complex topic is explained via breaking in down into simpler pieces

Surely accepted. Sadly didn't get a chance to review it. Did you cite our encirclement ICRA paper this year?

so cool as always. it's my dream to work in this lab one day lol

So. where to buy one??

😈 promo sm

It is very nice, and it is not believable you researched it 8 years ago, thanks for sharing, and now you can use RL to more enjoy the results.

god this is so cool. also congrats on your thesis defense jesus!!

this is exactly what i was looking for. Im in design school prototyping a autonomous beer buggy

great research

Great. How did you test your model on the physical robot?

amazing

Bro just did a gymkhana

Very nice presentation, thank you for uploading! :) Do I understand it correctly, that the NN controller solely determines a future control input based on observed system outputs? So no desired values for the output or other information is taken into account?

Is source code available somewhere.

Props for them to use an actual rc car, trackers and a projector instead of just a simulation

How to make it?

What a phenomenal video.

This is the best

This is literally just steering behaviors from game ai. Am I missing something here?

Amazing

Amazing Video!!!!!

always a joy to see SLAM at scale.

well done!

This is the best control theory video I’ve ever seen.

paper link？