Underactuated Control (MIT Tedrake)
Phase diagram is particularly important in nonlinear dynamic.
–> more useful in control to move phase diagram
小學生 level: linear feedback controller (P, PI, PID)
中學生 level: optimal control (maybe linear or nonlinear controller depends on the cost function)
Optimal control ————–> w.r.t COST function
大學生 level: optimal control based on dynamic programming or … Boyd (convex optimization, CVX)
Pro: different cost function since it’s done by computer!!
advance level: machine learning? reinforce learning?
–> add constraint, optimal control
Input is always a function of (x, and x_dot), not any random function in control sense!!!!!!!!!!!!!!!!!!
–> introduce cost function in vector filed phase diagram
Based on MIT OCW Russ Tedrake lecture (YouTube)
and Princeton Strogatz lecture (YouTube)
Use double integrators and simple pendulum as an example for dynamic programming
Line to Quantum Mechanics