A PID controller drives a system to its setpoint. Tune the gains: too low = slow, too high = oscillation. This is EXACTLY what Corbin's dissertation addresses -- can we PROVE which gains are stable?
u = Kp*e + Ki*integral(e) + Kd*de/dt
Kp: proportional (spring-like, reacts to current error)
Ki: integral (eliminates steady-state offset)
Kd: derivative (damping, prevents overshoot)
Stable: output settles to setpoint. Unstable: output grows or oscillates forever.
Nise, Control Systems Engineering, Ch 1-6 | Corbin Klett dissertation, Ch 3-4 | MIT OCW 6.302