Rotary Inverted Pendulum System Using Reinforcement Learning

[@hw1875]

Very cool! May I know what is the reward function you used? Thanks!

[@ControlEngineeringCh]

good work

[@japlb]

Amazing project, but how do you measure the angle of the pendulum?

[@angelo_pavan]

I think he used the QUBE servo to measure the angle and some physics (lagrangian mechanics I guess). But idk, I didn't make the project.

[@tszulpinedo757]

Y cuando nos van a enseñar a hacerlo?

[@nocknock4832]

very cool!

[@nunobartolo2908]

But why not just use an lqr?

[@michach_164]

You need a model for LQR. Usually reinforcement Learning has the advantage of Model Free design. Thats the only advantage it would have. Also LQR is based upon a model of the real system, which will never perfectly describe the environment. Especially for nonlinear systems, this could provide an advantage over conventional design (Training on real system directly).