Controller

P: propoartional to the current error; $K_p$

I: summation of previous discrete errors in a time interval; $K_i$

D: derivative of the current error; or, the finite difference of the current and prevous states; $K_d$