Long-Horizon Multi-Object Rearrangement via Distilled Tree Search over Continuous Actions

Authors: Kallol Saha, Yining Hong, Amber Li, Yejin Choi, Maxim Likhachev, David Held · CoRL 2026

Licences

• Open Access
• Paper: arXiv.org perpetual, non-exclusive license 1.0

Under arXiv's licence, everything below is written from the abstract only.

Why it matters

Long-horizon manipulation is hard for learned policies, which must avoid compounding errors while keeping future steps feasible, and search can reason about consequences but becomes expensive in continuous, high-dimensional action spaces. The authors report that DISCO outperforms behavior cloning, reinforcement learning and search-based baselines on simulated and real-world rearrangement tasks.

Proposed solution

DISCO runs progressive-widening Monte Carlo tree search in physics simulation, with a learned policy proposing continuous actions and a learned value function scoring progress, periodically distills the resulting plans into the policy-value model, and deploys that model directly with no simulator-based search at test time.

Abstract

Despite recent advances in policy learning, long-horizon manipulation remains difficult because learned policies must avoid compounding errors while preserving future feasibility. While search-based planning can explicitly reason over such future consequences, it quickly becomes expensive in high-dimensional continuous action spaces. This raises a central question: how can long-horizon reasoning from search be transferred into policies in continuous action spaces? To this end, we propose DISCO: Distilled Tree Search over Continuous Actions. DISCO uses progressive-widening MCTS in physics simulation to generate long-horizon plans. During search, a learned policy proposes continuous actions, while a learned value function scores actions by their expected progress toward task completion. Using GPU-parallelized simulation, DISCO periodically distills MCTS-generated plans into the policy-value model, improving the action proposals and value estimates used in subsequent searches. Finally, the learned policy-value model can be directly deployed without the need for simulator-based search at test time. Across simulated and real-world rearrangement tasks, DISCO outperforms behavior cloning, reinforcement learning, and search-based baselines.

Links

• arXiv 2610.25785: https://arxiv.org/abs/2610.25785
• PDF: https://arxiv.org/pdf/2610.25785
• Code: https://github.com/kallol-saha/DISCO
• Project page: https://distilled-search.github.io/