Hey everyone, thanks for watching this video! If you have any questions or comments that you'd like me to see, please leave them under this comment so that I get notified and can respond. Cheers!
Hi Brian, Thank you for the videos. That was a huge help to kick-start my RL project. Do you know any RL open-source projects which is using the last mentioned RL approaches in which Gains of a classical controller are learned? Because that would be pretty interesting to see.
@@BrianBDouglas Thank you for your videos! I have a question: what would be a good heuristic for learning, say, PID gains? would it be error integral, weighted {response time, oscillation, steady state error} costs, or something else entirely?
Thanks a lot Brian, this is very good and useful series. Please, can you help me with some more details about using RL as optmization tool for a calssical control, particularly PI control. I have studied some of the mathwork example, but non of them use the RL as controller-parameters optmization tool. Many thanks.
Interesting how control theory and AI are overlapping. IN my view they both have differente potentials but complement each other, so I would say both are neccesary. But at the end, is RL leading to AGI? Or it's just another engineering tool? Great series of videos!
That final hybrid solution is awesome! Is it worth implementing on a quad and PID controllers? It would be interesting to have an autotuner which is pretty black-box
You may refer to this paper of mine which deals with autotuning of PID controllers using Reinforcement Learning - ieeexplore.ieee.org/abstract/document/8973068
Thank you for your interesting explanation of RL. In the case of RL tuners, specifically for LQR, what do you think would be the best representation of the states? In contrast to RL controller, I don't think dynamics states would be a good choice.
Hey everyone, thanks for watching this video! If you have any questions or comments that you'd like me to see, please leave them under this comment so that I get notified and can respond. Cheers!
Please update faster, quite enjoy watching your video, haha
@@hanlinniu I appreciate that! I wish I was faster but I'm afraid this is top speed for me.
@@BrianBDouglas ok, no worries, thank you very much anyway!
Hi Brian, Thank you for the videos. That was a huge help to kick-start my RL project. Do you know any RL open-source projects which is using the last mentioned RL approaches in which Gains of a classical controller are learned? Because that would be pretty interesting to see.
@@BrianBDouglas Thank you for your videos! I have a question: what would be a good heuristic for learning, say, PID gains? would it be error integral, weighted {response time, oscillation, steady state error} costs, or something else entirely?
Combining a traditional control with RL is very fresh to me.
and always thanks to you for uploading a good video.
Amazing video, if its possible, in another video you could show us one example of a process control in Matlab, thanks!!
I think a follow up video updating any changes that affect this learning sequence. It is five years old, 35 dog years and probably 10 RL years.
Thanks a lot Brian, this is very good and useful series. Please, can you help me with some more details about using RL as optmization tool for a calssical control, particularly PI control. I have studied some of the mathwork example, but non of them use the RL as controller-parameters optmization tool. Many thanks.
nice ...very nice Thanks a lot
Interesting how control theory and AI are overlapping. IN my view they both have differente potentials but complement each other, so I would say both are neccesary. But at the end, is RL leading to AGI? Or it's just another engineering tool? Great series of videos!
That final hybrid solution is awesome! Is it worth implementing on a quad and PID controllers? It would be interesting to have an autotuner which is pretty black-box
You may refer to this paper of mine which deals with autotuning of PID controllers using Reinforcement Learning - ieeexplore.ieee.org/abstract/document/8973068
@@soutrikband Great, thank you! Will have a read!
Thank you for your interesting explanation of RL. In the case of RL tuners, specifically for LQR, what do you think would be the best representation of the states? In contrast to RL controller, I don't think dynamics states would be a good choice.
How to increase robustness? You said dynamically change some parameters. How do you set that up?
Love you brian, you arre the besttt
sa