Self-triggered output feedback control for nonlinear networked control systems based on hybrid Lyapunov functions*

Michael Hertneck, Frank Allgöwer

TL;DR

This paper introduces a self-triggered control (STC) strategy for nonlinear systems that only have access to limited output measurements rather than the full system state. By combining an asymptotically stable observer with a dynamic variable that "remembers" past system behavior, the authors developed a method that significantly reduces network communication frequency while mathematically guaranteeing system stability.

What Problem Does This Paper Address?

In Networked Control Systems (NCS), controllers and sensors talk over a digital network with limited bandwidth. To save resources, researchers use Event-Triggered Control (ETC), where the system only sends an update when a specific error threshold is crossed. However, ETC requires constant monitoring of the system state. Self-Triggered Control (STC) is more efficient: at each update, the controller calculates exactly when the next update should happen, allowing the communication hardware to "sleep" in between.

The catch? Most existing STC methods for complex, nonlinear systems assume the controller can see the entire system state (e.g., every position, velocity, and internal variable). In the real world, we usually only have "output feedback"—limited sensor data like a single position or a voltage. Bridging the gap between theoretical full-state STC and practical output-feedback scenarios is the primary challenge this paper tackles.

Methods

The authors propose a framework that integrates three main components:

  1. State Observer: Since the full state isn't measured, a continuous-time observer is used to reconstruct the plant's internal state based on the available output.
  2. Hybrid Lyapunov Functions: The system is modeled as a "hybrid system"—one that has continuous "flows" (the physics of the robot or machine) and discrete "jumps" (the moments data is sent over the network). Lyapunov functions are used to prove that the system's energy stays bounded and eventually goes to zero (stability).
  3. Dynamic Variable ($\eta$): This is the "secret sauce." Instead of just looking at the current state, the controller maintains a dynamic variable that encodes a discounted average of past performance. If the system has been performing well (low error) in the past, the STC mechanism can "borrow" from that history to stay quiet for a longer period before the next update.

The core of the paper is Algorithm 1, which computes the next transmission time by checking multiple sets of stability parameters and choosing the one that maximizes the time until the next update without risking instability.

Datasets

The paper does not use external datasets. Instead, it evaluates the approach using a numerical simulation of a single-link robot arm, a standard benchmark in control theory.

Key Insights and Ideas

  • Decoupled Design: A major "aha moment" is that the design of the STC mechanism is largely independent of the specific observer used. As long as the observer is "asymptotically stable" (meaning its estimation error eventually goes to zero), it can be plugged into this framework.
  • Non-Conservatism: The authors show that moving from full-state feedback to output feedback doesn't require stricter assumptions on the plant or the controller. This makes the approach highly versatile.
  • Memory Matters: By using the dynamic variable $\eta$ to track past behavior, the system can achieve much longer transmission intervals than "static" rules that only look at the current error.
  • The Fallback Strategy: The algorithm includes a "fallback" transmission time ($t_{min}$) to ensure that even if the system state grows faster than the observer can keep up initially, the system remains stable.

Key Results

The authors compared their STC approach against traditional periodic sampling (where updates happen at fixed intervals) using the robot arm model:

  • Communication Reduction: In a 50-second simulation, the proposed STC required only 92.4 transmissions on average, compared to 282 transmissions required by periodic sampling to guarantee the same level of stability.
  • Efficiency Gain: This represents a 3x reduction in network load.
  • Stability Proof: The paper provides a rigorous mathematical proof (Appendix A) that the system achieves Uniform Global Asymptotic Stability (UGAS), meaning it will return to its equilibrium point regardless of the starting conditions.

Limitations and Future Work

  • The "Initial Spike" Problem: The authors noted that if the observer starts with a large estimation error, the system may trigger very frequent updates at the beginning of a run. They suggest that tuning the initial values of the dynamic variable $\eta$ could mitigate this, though it might slow down the initial convergence speed.
  • Observer Complexity: While the framework works with any stable observer, designing such observers for highly complex or "chaotic" nonlinear systems remains a difficult task in itself.
  • Network Imperfections: The current model assumes a relatively clean communication channel. Future work could involve extending this output-feedback STC to handle significant network delays, jitter, or packet loss.

Summary updated

Abstract

Most approaches for self-triggered control (STC) of nonlinear networked control systems (NCS) require measurements of the full system state to determine transmission times. However, for most control systems only a lower dimensional output is available. To bridge this gap, we present in this paper an output-feedback STC approach for nonlinear NCS. An asymptotically stable observer is used to reconstruct the plant state and transmission times are determined based on the observer state. The approach employs hybrid Lyapunov functions and a dynamic variable to encode past state information and to maximize the time between transmissions. It is non-conservative in the sense that the assumptions on plant and controller are the same as for dynamic STC based on hybrid Lyapunov functions with full state measurements and any asymptotically stabilizing observer can be used. We conclude that the proposed STC approach guarantees asymptotic stability of the origin for the closed-loop system.
BibTeX
@inproceedings{hertneck2023self,
  author = {Michael Hertneck and Frank Allgöwer},
  title = {{Self-triggered output feedback control for nonlinear networked control systems based on hybrid Lyapunov functions*}},
  booktitle = {IFAC-PapersOnLine},
  year = {2023},
  doi = {10.1016/j.ifacol.2023.10.165},
}

Loading related content...

Figure