PID Control in AI & Artificial Intelligence

The Proportional (P) component is the primary driver of the system. It calculates the Error (Setpoint - Current State) and applies a force proportional to that error. If a drone is 5 meters too low, the P-term tells the motors to spin much faster. If it's only 0.1 meters low, it applies a tiny nudge. The problem with P-only control is Oscillation: the robot often speeds toward the goal so fast that it overshoots, then has to turn around and speed back, creating a never-ending 'Wobble'.

The Derivative (D) component is the 'Dampener'. It looks at how fast the error is changing. If the robot is approaching the setpoint very quickly, the D-term sees the 'Velocity' and applies a counter-force (braking) before the robot actually reaches the goal. This effectively 'Predicts' the overshoot and stops it before it happens, allowing for a smooth, exponential approach to the target. In robotics, PD control is often sufficient for high-speed systems like drones.

The Integral (I) component handles Steady-State Error. Imagine a robot trying to drive up a hill. The P-term might settle at a point where the motor force perfectly balances gravity, but the robot is still 2cm below the target. The I-term sums up the error over time. As long as the robot is even slightly off-target, the 'Integral' value keeps growing until it provides enough extra power to push the robot exactly onto the setpoint. However, too much I-term can lead to Wind-up, a dangerous state where the system over-reacts after being stuck.

How do you keep a robot moving at exactly the right speed, or keep a drone perfectly level? The PID Controller is the universal algorithm for maintaining a stable state through constant feedback.

A PID controller calculates the 'Error'—the difference between where you are and where you want to be—and adjusts the motors to fix it.

Proportional (P) pushes harder the further you are from the goal. Integral (I) fixes persistent errors. Derivative (D) prevents overshooting by braking as you get close.

Checkpoint: What does the 'D' (Derivative) part of a PID controller do?

• →Adds more power
• →It looks at the 'rate of change' of the error and dampens the movement to prevent overshooting the target

Tuning a PID controller involves finding the 'Sweet Spot' for the three K-values. If Kp is too high, the robot vibrates. If it's too low, the robot is sluggish.

By mastering PID Control, you can build machines that are smooth, precise, and capable of balancing in the most unstable conditions.

Checkpoint: What is the 'Setpoint'?

• →The battery limit
• →The desired target value (e.g., speed, position, or angle) that the system is trying to maintain

PID Control mastered! You've learned to balance. Ready to find the best path with A* and RRT?

Compute a Real PID Output. Finish computing a PID controller's output from its proportional, integral, and derivative terms.