Reproduce “Online Planning for Constrained POMDPs with Continuous Spaces through Dual Ascent” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Arec Jamgochian, Anthony Corso, Mykel J. Kochenderfer
The authors extend online constrained POMDP search to continuous action and observation spaces by combining dual ascent, which adjusts cost emphasis from estimated budget violations, with progressive widening, which controls tree growth. In LightDark, CPOMCPOW’s outcome lies above the convex hull of scalarized POMCPOW outcomes while satisfying the 0.1 expected discounted cost budget; this is empirical, not universal or episode-level safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.