Compositional Monte Carlo Tree Diffusion for Extendable Planning

TL;DR

This paper introduces C-MCTD, a novel framework that enhances Monte Carlo Tree Diffusion by enabling global plan reasoning, parallel exploration, and faster inference through plan composition and caching.

Contribution

C-MCTD extends MCTD by incorporating global plan composition, parallel search, and plan caching, addressing the limitations of local trajectory-based planning.

Findings

Enables globally-aware planning over complete plan compositions.

Reduces search complexity via distributed exploration.

Accelerates inference using cached plan graphs.

Abstract

Monte Carlo Tree Diffusion (MCTD) integrates diffusion models with structured tree search to enable effective trajectory exploration through stepwise reasoning. However, MCTD remains fundamentally limited by training trajectory lengths. While periodic replanning allows plan concatenation for longer plan generation, the planning process remains locally confined, as MCTD searches within individual trajectories without access to global context. We propose Compositional Monte Carlo Tree Diffusion (C-MCTD), a framework that elevates planning from individual trajectory optimization to reasoning over complete plan compositions. C-MCTD introduces three complementary components: (1) Online Composer, which performs globally-aware planning by searching across entire plan compositions; (2) Distributed Composer, which reduces search complexity through parallel exploration from multiple starting points; and (3) Preplan Composer, which accelerates inference by leveraging cached plan graphs.