gain scheduling

Gain scheduling is a practical trick for controlling a robot whose behavior changes a lot depending on where it is or what it's doing. The strength of a controller's correction is set by numbers called gains — turn them up and the robot reacts harder; turn them down and it reacts gently. The trouble is that one fixed set of gains that feels just right when a robot arm is folded up close to its base can feel sluggish or even shaky when the arm is stretched out far, because a stretched arm swings with much more inertia. Gain scheduling solves this by keeping several different sets of gains on hand and swapping between them depending on the situation.

The 'schedule' is simply a lookup table or chart that says which gains to use in which operating region. The controller measures a scheduling variable — something easy to read that signals which regime it's in, like joint angle, arm extension, payload weight, or vehicle speed — and then picks (or smoothly blends) the gains tuned for that region. It is like the gears on a bicycle: a low gear for climbing a steep hill, a high gear for cruising on the flat, and you shift as the terrain changes rather than forcing one gear to do everything.

This is one of the oldest and most widely used ways to handle nonlinear systems, because each individual gain set can be designed with simple, well-understood linear tools, and the schedule just stitches those local designs together to cover the whole operating envelope. The main care needed is at the seams: if the gains jump too abruptly when crossing between regions, the robot can lurch, so engineers usually blend the gains gradually and check that the system stays stable everywhere in between.