New self-perturbing ESC cuts configuration to one parameter per input
Forget complex tuning—this stochastic relay controller needs just one parameter per channel.
A new paper from Timothy I. Salsbury and Min Gyung Yu, published in IFAC-PapersOnLine and available on arXiv, introduces a stochastic relay-based extremum-seeking controller (ESC) designed for multi-input-single-output (MISO) systems. The core innovation lies in dramatically simplifying configuration: the static version requires only one tunable parameter per input channel, and the dynamic variant adds just one extra parameter. This contrasts sharply with traditional ESC approaches that demand extensive manual tuning of many gains and filters, making real-world deployment difficult.
The controller works by injecting stochastic relay signals to estimate the local gradient of the unknown objective function—a key step for extremum seeking—without needing a separate perturbation signal or dither. The authors provide a stability proof for the static map case and extend the method to a general class of dynamic systems via a simple adaptation. Simulation tests on both static and dynamic benchmarks confirm the controller's ability to converge near the optimum with minimal tuning effort. The work was published in the peer-reviewed IFAC-PapersOnLine (doi:10.1016/j.ifacol.2025.01.058) and is aimed at control engineers seeking plug-and-play optimization for HVAC, robotics, and process control.
- Only one configurable parameter per input channel for static MISO systems; one extra for dynamic systems.
- Uses stochastic relay gains instead of traditional dither signals for gradient estimation.
- Includes a stability proof for static maps and demonstrates performance in simulation on both static and dynamic plants.
Why It Matters
Simplifies extremum-seeking control for real-world use, reducing tuning time and enabling non-experts to deploy optimization.