Pith. sign in

REVIEW

LQR with Tracking: A Zeroth-order Approach and Its Global Convergence

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 2011.01815 v2 pith:ANY4KJMH submitted 2020-11-03 math.OC cs.LGcs.SYeess.SY

LQR with Tracking: A Zeroth-order Approach and Its Global Convergence

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

There has been substantial recent progress on the theoretical understanding of model-free approaches to Linear Quadratic Regulator (LQR) problems. Much attention has been devoted to the special case when the goal is to drive the state close to a zero target. In this work, we consider the general case where the target is allowed to be arbitrary, which we refer to as the LQR tracking problem. We study the optimization landscape of this problem, and show that similar to the zero-target LQR problem, the LQR tracking problem also satisfies gradient dominance and local smoothness properties. This allows us to develop a zeroth-order policy gradient algorithm that achieves global convergence. We support our arguments with numerical simulations on a linear system.

discussion (0)

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