Robust Control for Systems with Input Saturation

You are reading version 1 of this direction, frozen 2026-10-04. The latest version is version 2.

Version 1 of 2181 papers215 conceptsfrozen 2026-10-04

In classical control theory, we often design algorithms under the implicit assumption of "infinite authority." We assume that if a controller calculates that a drone needs 500 Newton-meters of torque to stay level in a hurricane, the motors will simply provide it.

In reality, every physical actuator—whether it is a quadrotor motor, a robotic joint, or a chemical valve—has a "ceiling." This is input saturation. When a controller demands more than the hardware can give, the mathematical guarantees of stability often vanish. Modern research in this field focuses on baking these physical limits directly into the control logic, ensuring that systems remain stable and performant even when they are "floored."

The Stability Gap in Saturated Systems

The fundamental problem with saturation is that it introduces a non-smooth nonlinearity that breaks the linear relationship between "command" and "result." When an actuator hits its limit, the feedback loop is effectively broken: the controller keeps increasing its demand to fix an error, but the output remains flat.

As Xi Ma, Hongbo Li, and Bing He point out in Attitude control of rigid body with inertia uncertainty and saturation input, ignoring these constraints leads to a discrepancy between commanded and actual torque that can destabilize a satellite or drone. This often manifests as "integrator windup," where the controller’s internal error-summing mechanism grows to astronomical levels while the motor is stuck at 100%, leading to massive overshoots once the error finally begins to decrease.

Integrating Limits into Lyapunov Stability

To fix this, researchers are moving away from "clipping" the output of a standard controller. Instead, they are redesigning the stability proofs themselves to include saturation.

One common approach is to incorporate a saturation function directly into the Lyapunov stability analysis. A Lyapunov function is a mathematical "energy" function used to prove that a system will eventually settle into a desired state. By accounting for the saturation function within this proof, researchers like Eduardo Mojica-Nava in Robust synchronization and policy adaptation for networked heterogeneous agents ensure that the system remains stable even when agents operate at their maximum input capacity.

Similarly, Lujie Yang demonstrates in Lyapunov-stable neural-network control that physical limits can be modeled as piecewise linear clamping functions. This allows for formal verification—essentially a mathematical guarantee—that a robot will not crash even when its motors are saturated.

The Performance-Safety Trade-off

A recurring theme in recent work is the inherent conflict between aggressive performance and physical boundaries. If you want a drone to track a path with centimeter-level precision, the controller will naturally want to use high gains. However, high gains lead to "spiky" control signals that hit saturation limits almost immediately.

Peng Wang 0039, Shuzhi Sam Ge, and Xiaobing Zhang address this in Balanced control between performance and saturation for constrained nonlinear systems. They propose a Performance Safety Evaluation (PSE) function. When the system is far from its limits, it prioritizes high-speed tracking. As the actuators approach saturation, the controller dynamically "softens" its performance requirements to ensure it never loses control authority. This prevents the system from "tripping" over its own hardware limits.

Constraint-Aware Learning and Optimization

As control moves toward data-driven methods like Reinforcement Learning (RL) and Model Predictive Control (MPC), the challenge is ensuring these "black box" or optimization-based models respect saturation.

Why This Matters for the Real World

This research direction bridges the "reality gap." Whether it is preventing "jerk" during electric vehicle gear shifts (Marc-Antoine Beaudoin, Fundamental limitations to no-jerk gearshift in electric vehicles) or ensuring underwater robots don't suffer from thruster "buffeting" (Xinyu An, An Improved ESO-Based Line-of-Sight Guidance Law...), the goal is the same: moving from controllers that work in perfect simulations to controllers that understand the physical exhaustion of the machines they command.

Papers in this version

Showing 60 of 181 papers.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 294 papers
  2. Version 1 · 2026-10-04 · 181 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_actuator_saturation_08266176,
  title = {actuator saturation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/actuator-saturation/rdr_08266176b14d42239e9f650b39803f53},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}

This exact version (version 1, frozen text):

@misc{lacuna_robust_control_for_systems_with_input_sa_08266176_v1,
  title = {Robust Control for Systems with Input Saturation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/robust-control-for-systems-with-input-saturation/rdr_08266176b14d42239e9f650b39803f53/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-22},
}