The Milieu, Science & Logic of Feedback Control
Robert R. Bitmead

TL;DR
This paper explores the philosophical and practical aspects of feedback control, emphasizing when data-driven control design is appropriate without detailed plant modeling, supported by industrial examples.
Contribution
It investigates the circumstances under which direct data-driven control design is viable, balancing robustness, confidence, and risk in practical applications.
Findings
Data-driven control can be effective with sufficient confidence in data and plant assumptions.
Industrial examples demonstrate practical viability of direct control design methods.
The approach depends on the risk appetite and the quality of available data.
Abstract
'The cardinal sin in control is to believe that the plant is given' Karl Astrom. Astrom, a towering figure of control theory and practice and awardee of the 1993 IEEE Medal of Honor for his work on adaptive control, provides this assessment of the obstinate part of realizing a feedback controller. And yet we are exhorted to rely on solely-data-driven methods of control design skipping the modeling and plant identification phases entirely. What is going on? Whom should we trust? How do we reconcile the implied ease (or indeed avoidance) of modeling with the steely focus on robustness of the control and the capacity of feedback to accommodate uncertainty? This paper seeks to investigate this subject with the objective of appreciating not whom to trust but what are the circumstances where the direct paradigm of control design from any lightly qualified data set provides a sensible way…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsComplex Systems and Decision Making
MethodsSparse Evolutionary Training · Focus
