Pith. sign in

REVIEW

Robust Approximate Simulation for Hierarchical Control of Linear Systems under Disturbances

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2003.04136 v1 pith:ZIK4T34Z submitted 2020-03-09 eess.SY cs.SY

Robust Approximate Simulation for Hierarchical Control of Linear Systems under Disturbances

classification eess.SY cs.SY
keywords simulationapproximatedisturbancessystemssystemconcretecontrolrobust
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Approximate simulation, an extension of simulation relations from formal methods to continuous systems, is a powerful tool for hierarchical control of complex systems. Finding an approximate simulation relation between the full "concrete" system and a simplified "abstract" system establishes a bound on the output error between the two systems, allowing one to design a controller for the abstract system while formally certifying performance on the concrete system. However, many real-world control systems are subject to external disturbances, which are not accounted for in the standard approximate simulation framework. We present a notion of robust approximate simulation, which considers external disturbances to the concrete system. We derive output error bounds for the case of linear systems subject to two types of additive disturbances: bounded disturbances and a sequence of (unbounded) impulse disturbances. We demonstrate the need for robust approximate simulation and the effectiveness of our proposed approach with a simulated robot motion planning example.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.