Prescribed Performance Control via Error Transformation

20 concepts14 papers1 proposal

In classical control theory, proving that a system is stable usually means showing that its tracking error will eventually converge to zero or remain within a small bounded region. For instance, a controller might guarantee uniform ultimate boundedness, ensuring long-term stability as discussed by Helwa et al. in Provably Robust Learning-Based Approach for High-Accuracy Tracking Control of Lagrangian Systems.

However, standard stability proofs are often silent about the transient phase—the period between initialization and steady state. During this window, a system might exhibit massive overshoot or painfully slow convergence. Historically, engineers attempted to mitigate this by aggressively tuning local controller gains to shrink the error envelope, a strategy explored by Fukao et al. in Decentralized Adaptive Control with Improved Transient Performance. Yet, high gains can saturate actuators and risk instability, and they still fail to provide hard mathematical guarantees on the error trajectory.

Prescribed Performance Control (PPC) solves this problem by enforcing a strict, time-varying "performance funnel" or envelope that the tracking error is mathematically forbidden from escaping.


The Performance Funnel

To dictate exactly how a system must behave, we define a time-varying performance function, $\rho(t)$. This function acts as a dynamic boundary for the tracking error $e(t)$. Typically, $\rho(t)$ starts out large to accommodate initial tracking errors and smoothly decays over time toward a small positive value $\rho_\infty$, which represents the maximum allowable steady-state error.

As highlighted by Chen et al. in Practical Prescribed Tracking Control of n-DOF Robotic Manipulator with Uncertainties via Friction Compensation Approach and Xi et al. in Disturbance Rejection Control for Autonomous Trolley Collection Robots with Prescribed Performance, this framework forces the tracking error to satisfy strict inequalities at all times:

$$-\rho(t) < e(t) < \rho(t)$$

By choosing the decay rate and the final value of $\rho(t)$, an engineer can directly prescribe the maximum overshoot, the minimum convergence speed, and the ultimate tracking accuracy of the system.

Error e(t)
  ^
  |  \   Upper Bound  \rho(t)
  |   \________________________
  |     *   *   *
  |   *   *       *  *  *  *  *   <-- Tracking Error e(t)
  |____________________________
  |   /  Lower Bound -\rho(t)
  |  /
  +-----------------------------> Time (t)

The Core Mechanism: Error Transformation

Enforcing these hard inequalities directly would normally require complex, computationally expensive constrained optimization algorithms (like Model Predictive Control) or barrier Lyapunov functions. PPC bypasses this complexity entirely through a mathematical trick: error transformation.

As explained by Wang et al. in Adaptive Prescribed Performance Fault Tolerant Control for a Flexible Air-Breathing Hypersonic Vehicle With Uncertainty and Li et al. in Robust Finite-Time Control of a Multi-AUV Formation Based on Prescribed Performance, the constrained tracking error $e(t)$ is mapped into an unconstrained state variable $\epsilon(t)$ using a smooth, strictly increasing bijective function $S(\cdot)$.

We can express the relationship as:

$$\frac{e(t)}{\rho(t)} = S(\epsilon(t))$$

A common choice for $S(\epsilon)$ is a sigmoidal or logarithmic function that maps the open interval $(-1, 1)$ to $(-\infty, \infty)$. For example:

$$S(\epsilon) = \frac{e^\epsilon - e^{-\epsilon}}{e^\epsilon + e^{-\epsilon}} = \tanh(\epsilon)$$

Let's look at the intuition behind why this works:

  1. Warping Space: The transformation maps the bounded, safe region inside the funnel ($-\rho(t) < e(t) < \rho(t)$) to an infinitely wide, unconstrained highway ($-\infty < \epsilon(t) < \infty$).
  2. The Infinite Barrier: If the physical tracking error $e(t)$ drifts close to the upper performance boundary $\rho(t)$, the ratio $e(t)/\rho(t)$ approaches $1$. Looking at the mapping, as $S(\epsilon) \to 1$, the transformed error $\epsilon(t)$ shoots toward $+\infty$.
  3. Simplified Control: As noted by Zhao et al. in Adaptive neural network control of a 2-DOF helicopter system considering input constraints and global prescribed performance, this completely transforms the constrained error dynamics into a new bounded variable. If a standard, unconstrained controller (like backstepping or sliding mode control) simply keeps the transformed error $\epsilon(t)$ stable and bounded, the physical error $e(t)$ is mathematically guaranteed never to touch or cross the performance boundaries.

Advancements and Variations

While the foundational error transformation mechanism is elegant, researchers have introduced several key innovations to handle real-world complexities.

Prescribed-Time and Fixed-Time Convergence

Standard performance functions decay asymptotically, meaning the error technically takes infinite time to reach its absolute steady-state bound. In time-critical applications, such as aerospace maneuvers, this is unacceptable.

To address this, Xu et al. in Reinforcement-Learning-Based Tracking Control with Fixed-Time Prescribed Performance for Reusable Launch Vehicle under Input Constraints and Sui et al. in Prescribed-Time Dynamic Positioning Control for USV with Lumped Disturbances, Thruster Saturation and Prescribed Performance Constraints developed fixed-time and prescribed-time performance functions. These functions are designed to hit their final steady-state value exactly at a user-specified time $T$, completely independent of the system's initial conditions.

Singularity-Free Design

A historical vulnerability of error transformation occurs at initialization: if the initial tracking error is too close to the boundary, $\epsilon(0)$ becomes extremely large, causing numerical instability or actuator saturation (singularities). To circumvent this, Cao et al. in Prespecified-Performance Kinematic Tracking Control for Aerial Manipulation proposed designing a specific trajectory for the tracking error to follow toward zero, bypassing traditional barrier functions and ensuring smooth stability near the boundaries.

Multi-Agent Coordination

PPC has also been scaled from single robots to distributed networks. Chen et al. demonstrated in Consensus Control for Leader-follower Multi-agent Systems under Prescribed Performance Guarantees and On topological conditions for enabling transient control in leader-follower networks that leader-follower formations can maintain strict relative distance envelopes during transitions.

To make this seamless across complex networks, Li et al. introduced a unified performance function in A Novel Edge Laplacian-Based Approach for Adaptive Formation Control of Uncertain Multiagent Systems With Unified Relative Error Performance that allows engineers to flexibly specify asymmetric relative error constraints independently of the underlying network topology or agent dynamics.

Robustness Against Faults

In physical deployments, sensor noise and actuator failures can threaten to push a system out of its performance envelope. To guarantee safety under these conditions, PPC is frequently paired with advanced observers. For instance, Zhao et al. in Sampled-data self-learning observer based attitude tracking control against sensor-actuator faults combined error boundary enforcement with a sampled-data self-learning observer, ensuring that tracking errors remain safely within the performance envelope even when sensor data is intermittent and actuators are faulty.

Key Foundational Work

Consensus Control for Leader-follower Multi-agent Systems under Prescribed Performance Guarantees (2019) — contributed 1 concept elements to this direction.
Adaptive Prescribed Performance Fault Tolerant Control for a Flexible Air-Breathing Hypersonic Vehicle With Uncertainty (2019) — contributed 1 concept elements to this direction.
Decentralized Adaptive Control with Improved Transient Performance (1999) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_prescribed_performance_control_bf8e6fe5,
  title = {prescribed performance control},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/prescribed-performance-control/rdr_bf8e6fe5e2414360b5cada82f9b6afd2},
  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_prescribed_performance_control_bf8e6fe5_v1,
  title = {prescribed performance control},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/prescribed-performance-control/rdr_bf8e6fe5e2414360b5cada82f9b6afd2/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-22},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake