Skip to content
Preprint

X-Planner: Event-Structured Task Planning for Embodied Intelligence

Sep 2026 · 0 citations · 30 references
Computer Science

TL;DR

This work presents X-Planner, a planning front-end that addresses both the supervision and representation of embodied reasoning, and describes planning-text quality and downstream execution.

Abstract

Task planning bridges high-level instructions and executable behavior in long-horizon manipulation, yet modern Vision-Language-Action (VLA) systems often leave this intermediate structure implicit. Existing chain-of-thought (CoT) planners also tend to rely on coarse task-level annotations or serialize long reasoning traces token by token. We present X-Planner, a planning front-end that addresses both the supervision and representation of embodied reasoning. Our planning data combine Ego, UMI, and teleoperation under a hierarchy granularity with source-dependent annotation depth. Takeover-time annotations and human-designed failures supervise ongoing error recognition. On the model side, a shared VLM backbone exposes two event-structured plan forms: a discrete interface that emits interpretable event states and a latent interface that relays continuous CoT states across staggered Transformer depths through Staircase Decoding. A frozen latent-to-text reconstruction objective provides a semantic anchor for the latent representation. Offline two-step planning evaluation places X-Planner second among four evaluated models on both BERTScore-F1 and a judge-based Overall score. In real-robot experiments, respectively, outperforming the evaluated baselines. These results characterize planning-text quality and downstream execution.

View source

Similar papers

Preprint Sep 2026

Fast Plans, Faithful Actions: Closing the Planning-Execution Gap in Hierarchical Vision-Language-Action Models

Hierarchical vision-language-action (VLA) systems consist of a high-level vision-language planner and a low-level action expert that generates continuous actions. This hierarchical design has practical value only if the planner can generate plans fast enough to meet real-time control requirements, and the resulting pla...

Chuan-Liang Xie, Bo-Yu Ma, Gen Li et al. · 0 citations
Preprint Sep 2026

Towards Neuro-Symbolic Procedural Reasoning for Long-Horizon Vision-Language-Action Manipulation

A neuro-symbolic framework that combines learned VLA control with explicit task graphs and multimodal procedural memory is investigated, which positions structured symbolic reasoning and demonstration-derived visual guidance as complementary mechanisms for reliable long-horizon VLA manipulation.

Vivek Chavan, Ya-Huan Shi, O. Heimann et al. · 0 citations
Preprint Sep 2026

Safe Task Planning with Long-Term Graph Memory for Embodied Agents

A novel safe task-planning framework, SafeMem, which constructs and maintains a long-term semantic graph memory of the open and dynamic environment, and substantially improves safe success rates compared to state-of-the-art VLM-driven task planners.

Si-Yuan Li, Tai-Yan Lang, Ao Yan et al. · 1 citation
Conference Open access Aug 2026

Reasoning-Distilled Planning: Transferring Vision–Language Guidance to Real-Time End-to-End Driving

Reasoning-Distilled Planner (RDP) is introduced, which uses a vision–language model only while constructing training signals for a sparse planner and returns schema-constrained scene reasoning and an auxiliary trajectory.

Ling Duan, Jia-Xin Lin, Liang Chen et al. · 0 citations
Conference Open access Aug 2026

PRISM: Planning with Belief Under Incomplete World Models

PRISM, Planning with Belief under Incomplete World Models, a lightweight inference-time framework that augments VLM planners with an explicit probabilistic belief over candidate object locations provides a transparent uncertainty representation that exposes the agent's location hypotheses at each step, supporting inter...

M. S. Yasar, Murugan Sankaradas, Srimat T. Chakradhar · 0 citations
Conference Aug 2026

Structured-VLA: An Explicit Structural Interface for Vision-Language-Action Control

Vision-Language-Action (VLA) models have shown strong promise for general-purpose robotic manipulation, but existing approaches often condition control on entangled multimodal features that do not explicitly preserve task-relevant structure during action generation. This limitation is especially pronounced in long-hori...

Pei-Sen Huang, Jian-Hua Hu, Jie-Ren Deng et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.