SVLL: Staged Vision-Language Learning for Physically Grounded Embodied Task Planning

• URL: http://arxiv.org/abs/2603.11563v1
• Date: Thu, 12 Mar 2026 05:35:29 GMT
• Title: SVLL: Staged Vision-Language Learning for Physically Grounded Embodied Task Planning
• Abstract summary: We present Staged Vision-Language Learning (SVLL), a unified three-stage framework for robust, physically-grounded embodied planning.<n>In the first two stages, SVLL decouples spatial grounding from temporal reasoning, establishing robust visual dependency before introducing sequential action history.<n>In the final stage, we identify a key limitation of standard Direct Preference Optimization (DPO), its purely relative nature -- optimizing only the preference gap between winning and losing trajectories.
• Score: 21.113678610046453
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Embodied task planning demands vision-language models to generate action sequences that are both visually grounded and causally coherent over time. However, existing training paradigms face a critical trade-off: joint end-to-end training often leads to premature temporal binding, while standard reinforcement learning methods suffer from optimization instability. To bridge this gap, we present Staged Vision-Language Learning (SVLL), a unified three-stage framework for robust, physically-grounded embodied planning. In the first two stages, SVLL decouples spatial grounding from temporal reasoning, establishing robust visual dependency before introducing sequential action history. In the final stage, we identify a key limitation of standard Direct Preference Optimization (DPO), its purely relative nature -- optimizing only the preference gap between winning and losing trajectories while neglecting absolute likelihood constraints on optimal path, often yields unsafe or hallucinated behaviors. To address this, we further introduce Bias-DPO, a novel alignment objective that injects an inductive bias toward expert trajectories by explicitly maximizing likelihood on ground-truth actions while penalizing overconfident hallucinations. By anchoring the policy to the expert manifold and mitigating causal misalignment, SVLL, powered by Bias-DPO, ensures strict adherence to environmental affordances and effectively suppresses physically impossible shortcuts. Finally, extensive experiments on the interactive AI2-THOR benchmark and real-world robotic deployments demonstrate that SVLL outperforms both state-of-the-art open-source (e.g., Qwen2.5-VL-7B) and closed-source models (e.g., GPT-4o, Gemini-2.0-flash) in task success rate, while significantly reducing physical constraint violations.

Related papers

• CofactVLA: Deconfounding Vision-Language-Action Models via Counterfactual Intervention [48.20065816234311]
  Vision-Language-Action (VLA) models have driven significant progress in robotic manipulation, yet they struggle with the vision-override phenomenon.<n>We propose CofactVLA, a novel causal intervention framework.<n>We show that CofactVLA establishes a new state-of-the-art across diverse simulation benchmarks.
  arXiv  Detail & Related papers  (2026-08-05T02:58:41Z)
• ARP: Enhancing Quantized Skill Abstractions via Visual Alignment and Iterative Refinement for Robotic Manipulation [12.782646427089967]
  Aligned Refinement Policy (ARP) couples semantic grounding with execution-level refinement.<n>ARP achieves state-of-the-art performance on the LIBERO and Meta-World benchmarks.
  arXiv  Detail & Related papers  (2026-06-21T12:49:46Z)
• Think Less, Act Early: Reinforced Latent Reasoning with Early Exit in Vision-Language-Action Models [3.4519796338615225]
  We propose a novel Vision-Language-Action (VLA) framework that models reasoning as a sequence of unobservable latent variables.<n>We introduce a Reinforcement Learning-based Denoising mechanism that treats latent state generation as a sequential decision process.<n>Experiments on embodied decision benchmarks demonstrate that AVA-VLA achieves a 6x inference speedup over explicit CoT methods.
  arXiv  Detail & Related papers  (2026-06-13T04:16:18Z)
• Bridging the Morphology Gap: Adapting VLA Models to Dexterous Manipulation via Intent-Conditioned Fine-Tuning [7.018869512628755]
  We present InDex, a novel, data-efficient adaptation framework rooted in cross-morphology semantic inheritance.<n>InDex effectively masters intricate skills with minimal demonstration data, substantially outperforming monolithic baselines.
  arXiv  Detail & Related papers  (2026-06-10T14:03:52Z)
• PRTS: A Primitive Reasoning and Tasking System via Contrastive Representations [66.94988600664574]
  We present textbfPRTS (textbfPrimitive textbfReasoning and textbfTasking textbfSystem), a VLA foundation model that reformulates pretraining through Goal-Conditioned Reinforcement Learning.
  arXiv  Detail & Related papers  (2026-04-30T06:14:02Z)
• DIAL: Decoupling Intent and Action via Latent World Modeling for End-to-End VLA [72.9197085473598]
  DIAL is a framework bridging high-level decision making and low-level motor execution through a differentiable latent intent bottleneck.<n>A VLM-based System-2 performs latent world modeling by latent visual foresight within the VLM's native feature space.<n>A lightweight System-1 policy then decodes this predicted intent together with the current observation into precise robot actions.
  arXiv  Detail & Related papers  (2026-03-31T15:02:27Z)
• Dual-End Consistency Model [41.982957134224904]
  Slow iterative sampling is a major bottleneck for the practical deployment of diffusion and flow-based generative models.<n>We propose a Dual-End Consistency Model (DE-CM) that selects vital sub-trajectory clusters to achieve stable and effective training.<n>Our method achieves a state-of-the-art FID score of 1.70 in one-step generation on the ImageNet 256x256 dataset, outperforming existing CM-based one-step approaches.
  arXiv  Detail & Related papers  (2026-02-11T11:51:01Z)
• OmniVL-Guard: Towards Unified Vision-Language Forgery Detection and Grounding via Balanced RL [63.388513841293616]
  Existing forgery detection methods fail to handle the interleaved text, images, and videos prevalent in real-world misinformation.<n>To bridge this gap, this paper targets to develop a unified framework for omnibus vision-language forgery detection and grounding.<n>We propose textbf OmniVL-Guard, a balanced reinforcement learning framework for omnibus vision-language forgery detection and grounding.
  arXiv  Detail & Related papers  (2026-02-11T09:41:36Z)
• SynCast: Synergizing Contradictions in Precipitation Nowcasting via Diffusion Sequential Preference Optimization [62.958457694151384]
  We introduce preference optimization into precipitation nowcasting for the first time, motivated by the success of reinforcement learning from human feedback in large language models.<n>In the first stage, the framework focuses on reducing FAR, training the model to effectively suppress false alarms.
  arXiv  Detail & Related papers  (2025-10-22T16:11:22Z)
• IntentionVLA: Generalizable and Efficient Embodied Intention Reasoning for Human-Robot Interaction [51.130510883952546]
  Vision-Language-Action (VLA) models leverage pretrained vision-language models (VLMs) to couple perception with robotic control.<n>We propose textbfIntentionVLA, a VLA framework with a curriculum training paradigm and an efficient inference mechanism.<n>Our proposed method first leverages carefully designed reasoning data that combine intention inference, spatial grounding, and compact embodied reasoning.
  arXiv  Detail & Related papers  (2025-10-09T04:49:46Z)
• TD-JEPA: Latent-predictive Representations for Zero-Shot Reinforcement Learning [63.73629127832652]
  We introduce TD-JEPA, which leverages TD-based latent-predictive representations into unsupervised RL.<n> TD-JEPA trains explicit state and task encoders, a policy-conditioned multi-step predictor, and a set of parameterized policies directly in latent space.<n> Empirically, TD-JEPA matches or outperforms state-of-the-art baselines on locomotion, navigation, and manipulation tasks across 13 datasets.
  arXiv  Detail & Related papers  (2025-10-01T10:21:18Z)
• STITCH-OPE: Trajectory Stitching with Guided Diffusion for Off-Policy Evaluation [18.55356623615343]
  Off-policy evaluation (OPE) estimates the performance of a target policy using offline data collected from a behavior policy.<n>Existing OPE methods are ineffective for high-dimensional, long-horizon problems.<n>We propose STITCH-OPE, a model-based generative framework that leverages denoising diffusion for long-horizon OPE.
  arXiv  Detail & Related papers  (2025-05-27T06:39:26Z)
• Plan-R1: Safe and Feasible Trajectory Planning as Language Modeling [74.41886258801209]
  We propose a two-stage trajectory planning framework that decouples principle alignment from behavior learning.<n>Plan-R1 significantly improves planning safety and feasibility, achieving state-of-the-art performance.
  arXiv  Detail & Related papers  (2025-05-23T09:22:19Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.