マルチスケール拡散による拡張可能な計画
Extendable Planning via Multiscale Diffusion
拡散ベースの計画器の長期的タスクにおける限界を、軌道の漸進的拡張と階層的マルチスケール拡散を組み合わせて解決し、単一モデルで長期的意思決定を可能にした。
著者: Chang Chen, Hany Hamed, Doojin Baek, Taegu Kang, Samyeul Noh, Yoshua Bengio, Sungjin Ahn
分類: cs.LG, cs.RO
原文アブストラクト
Long-horizon planning is crucial in complex environments, but diffusion-based planners like Diffuser are limited by the trajectory lengths observed during training. This creates a dilemma: long trajectories are needed for effective planning, yet they degrade model performance. In this paper, we introduce this extendable long-horizon planning challenge and propose a two-phase solution. First, Progressive Trajectory Extension incrementally constructs longer trajectories through multi-round compositional stitching. Second, the Hierarchical Multiscale Diffuser enables efficient training and inference over long horizons by reasoning across temporal scales. To avoid the need for multiple separate models, we propose Adaptive Plan Pondering and the Recursive HM-Diffuser, which unify hierarchical planning within a single model. Experiments show our approach yields strong performance gains, advancing scalable and efficient decision-making over long-horizons.
