Robust Control vs Nominal Control

Which Controller Survives Model Uncertainty?

Consider a simplified high-altitude platform aircraft flying at near-constant altitude. The control system was tuned on a clean nominal model: nominal mass, nominal damping, nominal drag, and nominal actuator response.

The real aircraft may not match that model. Payload changes, uncertain aerodynamic drag, changing atmospheric conditions, structural flexibility, actuator limits, and wind gusts can all change the closed-loop response.

Excellent Nominal Response Can Create False Confidence

A tuned PID can look almost perfect when the aircraft exactly matches the assumed model. But if mass increases, damping changes, or drag is lower than expected, the same controller can become too aggressive, too weak, oscillatory, energy-expensive, or saturation-prone.

Lateral / Altitude Deviation Under Uncertain Mass and Drag

The aircraft is represented by a second-order station-keeping model. The state is deviation $x$ and velocity error $\dot{x}$. The control input is $u$, and $d(t)$ represents wind disturbance.

Here, $m$ is uncertain effective mass and $c$ is uncertain aerodynamic damping or drag effect. Higher mass makes the aircraft respond more slowly. Lower damping can make oscillations last longer.

The Controller Sees One Aircraft; Reality May Be Another

The nominal model uses $m_{nominal}$ and $c_{nominal}$. The real model is modified using uncertainty factors.

The simulator lets you change mass uncertainty from lighter to heavier aircraft, and drag/damping uncertainty from low-damping to high-damping response.

Fast, Sharp, and Optimized for the Expected Aircraft

The nominal PID is tuned to perform well on the expected model. It can give low error and fast response when the model is correct, but it can over-command when the model changes.

• Fast response in the nominal case
• Low error when model assumptions are correct
• Can overshoot under uncertainty
• Can saturate actuators
• Can consume more energy
• Can suffer integral windup

Slower Nominal Response, Better Uncertainty Survival

The robust controller is intentionally conservative. It uses lower proportional aggression, relatively higher damping emphasis, smaller integral action, actuator-aware limiting, anti-windup, and control smoothing.

It may be slower when the aircraft is perfectly nominal, but safer across uncertain conditions.

The Aircraft Cannot Apply Unlimited Control

Both controllers are limited by the same actuator authority. The commanded control may be larger than what the aircraft can actually apply.

Anti-windup prevents the integral term from growing dangerously when the actuator is already saturated.

Robustness Should Include Energy Demand

A high-end control comparison should not only show tracking error. It should also show how much energy the controller spends fighting disturbances and uncertainty.

A controller that uses extreme actuator effort may look accurate but be unsuitable for long-endurance aircraft.

Robust Control vs Nominal PID Simulator

Adjust uncertainty, wind, actuator limits, gains, robustness level, sensor noise, and initial deviation. The simulation compares the nominal PID and robust controller on the same real aircraft.

Six High-Value Test Cases

Use these presets to quickly demonstrate how nominal tuning and robust design behave under different uncertainty conditions.

Simulation Figures

Controller Performance Metrics

How the Controllers Fail

Nominal Performance vs Robust Survival

The Stronger Controller Is Not Always the Fastest One

A controller should not be judged only by how well it performs on the nominal model.

For high-altitude long-endurance flight, mass, drag, wind, actuator authority, and environmental uncertainty all change the closed-loop response.