

ep31 - Miroslav Krstić: nonlinear adaptive control, PDEs, delays, extremum seeking, safety, neural operators for control
Apr 15, 2025
AI Snips
Chapters
Transcript
Episode notes
PhD Eureka in Adaptive Control
- Miroslav Krstić solved an adaptive control problem during his PhD while his advisor was on vacation.
- His solution avoided overparameterization by estimating the parameter vector with a single parameter estimate, boosting his confidence.
PDE Types and Navier-Stokes Control
- Hyperbolic PDEs dominate transport and delay phenomena; parabolic PDEs induce spatial smoothing like low-pass filtering.
- Navier-Stokes equations are complex PDEs representing fluid dynamics with algebraic and integral terms, requiring advanced control techniques.
Traffic Control Collaborative Breakthrough
- Concurrent solutions to traffic flow control by Krstić's group and French researchers led to collaborative publications.
- Boundary control via ramp metering can suppress traffic oscillations upstream, improving flow and comfort.