MoMaStage: Skill-State Graph Guided Planning and Closed-Loop Execution for Long-Horizon Indoor Mobile Manipulation

TL;DR

MoMaStage introduces a structured vision-language framework for long-horizon indoor mobile manipulation that ensures logical consistency and robustness without explicit scene mapping, significantly improving task success in complex environments.

Contribution

It proposes a novel hierarchical skill and skill-state graph grounded in vision-language models, enabling flexible, topologically valid planning and closed-loop execution without explicit scene mapping.

Findings

Outperforms state-of-the-art baselines in simulation and real-world tasks.

Achieves higher planning success and task completion rates.

Reduces token overhead and improves robustness in dynamic environments.

Abstract

Indoor mobile manipulation (MoMA) enables robots to translate natural language instructions into physical actions, yet long-horizon execution remains challenging due to cascading errors and limited generalization across diverse environments. Learning-based approaches often fail to maintain logical consistency over extended horizons, while methods relying on explicit scene representations impose rigid structural assumptions that reduce adaptability in dynamic settings. To address these limitations, we propose MoMaStage, a structured vision-language framework for long-horizon MoMA that eliminates the need for explicit scene mapping. MoMaStage grounds a Vision-Language Model (VLM) within a Hierarchical Skill Library and a topology-aware Skill-State Graph, constraining task decomposition and skill composition within a feasible transition space. This structured grounding ensures that generated plans remain logically consistent and topologically valid with respect to the agent's evolving physical state. To enhance robustness, MoMaStage incorporates a closed-loop execution mechanism that monitors proprioceptive feedback and triggers graph-constrained semantic replanning when deviations are detected, maintaining alignment between planned skills and physical outcomes. Extensive experiments in physics-rich simulations and real-world environments demonstrate that MoMaStage outperforms state-of-the-art baselines, achieving substantially higher planning success, reducing token overhead, and significantly improving overall task success rates in long-horizon mobile manipulation. Video demonstrations are available on the project website: https://chenxuli-cxli.github.io/MoMaStage/.