# RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics

arXiv: https://arxiv.org/abs/2506.04308 · HTML: https://mlsanity.com/paper/2506.04308/

## Key facts

- Cited by 127 papers in the map (data through 14 Oct 2026).
- Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning.
- Builds on 20 of its 159 references that are in the map.
- First posted on arXiv 4 Jun 2025.

## Placed in

- LLM reasoning / [Spatial and embodied reasoning](https://mlsanity.com/lr/spatial-and-embodied-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 20 of 20)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [WorldSimBench: Towards Video Generation Models as World Simulators](https://mlsanity.com/paper/2410.18072/) (2024)
- [VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model](https://mlsanity.com/paper/2504.07615/) (2025)
- [RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete](https://mlsanity.com/paper/2502.21257/) (2025)
- [Code-as-Monitor: Constraint-aware Visual Programming for Reactive and Proactive Robotic Failure Detection](https://mlsanity.com/paper/2412.04455/) (2024)
- [LaMP: Language-Motion Pretraining for Motion Generation, Retrieval, and Captioning](https://mlsanity.com/paper/2410.07093/) (2024)
- [MineDreamer: Learning to Follow Instructions via Chain-of-Imagination for Simulated-World Control](https://mlsanity.com/paper/2403.12037/) (2024)
- [T2ISafety: Benchmark for Assessing Fairness, Toxicity, and Privacy in Image Generation](https://mlsanity.com/paper/2501.12612/) (2025)
- [TrackVLA: Embodied Visual Tracking in the Wild](https://mlsanity.com/paper/2505.23189/) (2025)
- [SSR: Enhancing Depth Perception in Vision-Language Models via Rationale-Guided Spatial Reasoning](https://mlsanity.com/paper/2505.12448/) (2025)
- [Cubify Anything: Scaling Indoor 3D Object Detection](https://mlsanity.com/paper/2412.04458/) (2024)
- [ReSo: A Reward-driven Self-organizing LLM-based Multi-Agent System for Reasoning Tasks](https://mlsanity.com/paper/2503.02390/) (2025)
- [RoboFactory: Exploring Embodied Agent Collaboration with Compositional Constraints](https://mlsanity.com/paper/2503.16408/) (2025)
- [RoboOS: A Hierarchical Embodied Framework for Cross-Embodiment and Multi-Agent Collaboration](https://mlsanity.com/paper/2505.03673/) (2025)
- [ManipLVM-R1: Reinforcement Learning for Reasoning in Embodied Manipulation with Large Vision-Language Models](https://mlsanity.com/paper/2505.16517/) (2025)
- [RH20T-P: A Primitive-Level Robotic Dataset Towards Composable Generalization Agents](https://mlsanity.com/paper/2403.19622/) (2024)
- [LabUtopia: High-Fidelity Simulation and Hierarchical Benchmark for Scientific Embodied Agents](https://mlsanity.com/paper/2505.22634/) (2025)
- [NavigateDiff: Visual Predictors are Zero-Shot Navigation Assistants](https://mlsanity.com/paper/2502.13894/) (2025)
- [Story3D-Agent: Exploring 3D Storytelling Visualization with Large Language Models](https://mlsanity.com/paper/2408.11801/) (2024)
- [MulSMo: Multimodal Stylized Motion Generation by Bidirectional Control Flow](https://mlsanity.com/paper/2412.09901/) (2024)

## Built on by (most cited first, 30 of 127)

- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [A Survey of Reinforcement Learning for Large Reasoning Models](https://mlsanity.com/paper/2509.08827/) (2025)
- [InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot Policy](https://mlsanity.com/paper/2510.13778/) (2025)
- [Detect Anything via Next Point Prediction](https://mlsanity.com/paper/2510.12798/) (2025)
- [Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalization](https://mlsanity.com/paper/2601.12993/) (2026)
- [Seed1.8 Model Card: Towards Generalized Real-World Agency](https://mlsanity.com/paper/2603.20633/) (2026)
- [MolmoAct2: Action Reasoning Models for Real-world Deployment](https://mlsanity.com/paper/2605.02881/) (2026)
- [Qwen-RobotManip Technical Report: Alignment Unlocks Scale for Robotic Manipulation Foundation Models](https://mlsanity.com/paper/2606.17846/) (2026)
- [How Far are VLMs from Visual Spatial Intelligence? A Benchmark-Driven Perspective](https://mlsanity.com/paper/2509.18905/) (2025)
- [Robo-Dopamine: General Process Reward Modeling for High-Precision Robotic Manipulation](https://mlsanity.com/paper/2512.23703/) (2025)
- [VLM4VLA: Revisiting Vision-Language-Models in Vision-Language-Action Models](https://mlsanity.com/paper/2601.03309/) (2026)
- [LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence](https://mlsanity.com/paper/2605.25979/) (2026)
- [Embodied-R1: Reinforced Embodied Reasoning for General Robotic Manipulation](https://mlsanity.com/paper/2508.13998/) (2025)
- [Holistic Evaluation of Multimodal LLMs on Spatial Intelligence](https://mlsanity.com/paper/2508.13142/) (2025)
- [TIGeR: Tool-Integrated Geometric Reasoning in Vision-Language Models for Robotics](https://mlsanity.com/paper/2510.07181/) (2025)
- [Robobench: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models as Embodied Brain](https://mlsanity.com/paper/2510.17801/) (2025)
- [Think3D: Thinking with Space for Spatial Reasoning](https://mlsanity.com/paper/2601.13029/) (2026)
- [Vlaser: Vision-Language-Action Model with Synergistic Embodied Reasoning](https://mlsanity.com/paper/2510.11027/) (2025)
- [Spatial Reasoning in Multimodal Large Language Models: A Survey of Tasks, Benchmarks and Methods](https://mlsanity.com/paper/2511.15722/) (2025)
- [A Systematic Study of Data Modalities and Strategies for Co-training Large Behavior Models for Robot Manipulation](https://mlsanity.com/paper/2602.01067/) (2026)
- [SpaceTools: Tool-Augmented Spatial Reasoning via Double Interactive RL](https://mlsanity.com/paper/2512.04069/) (2025)
- [Seeing Across Views: Benchmarking Spatial Reasoning of Vision-Language Models in Robotic Scenes](https://mlsanity.com/paper/2510.19400/) (2025)
- [Reasoning in Space via Grounding in the World](https://mlsanity.com/paper/2510.13800/) (2025)
- [Theory of Space: Can Foundation Models Construct Spatial Beliefs through Active Exploration?](https://mlsanity.com/paper/2602.07055/) (2026)
- [VIKI-R: Coordinating Embodied Multi-Agent Cooperation via Reinforcement Learning](https://mlsanity.com/paper/2506.09049/) (2025)
- [From Language to Locomotion: Retargeting-free Humanoid Control via Motion Latent Guidance](https://mlsanity.com/paper/2510.14952/) (2025)
- [Thinking with Geometry: Active Geometry Integration for Spatial Reasoning](https://mlsanity.com/paper/2602.06037/) (2026)
- [HY-Embodied-0.5: Embodied Foundation Models for Real-World Agents](https://mlsanity.com/paper/2604.07430/) (2026)
- [DrivePI: Spatial-aware 4D MLLM for Unified Autonomous Driving Understanding, Perception, Prediction and Planning](https://mlsanity.com/paper/2512.12799/) (2025)
- [ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning](https://mlsanity.com/paper/2604.24300/) (2026)

## Abstract (arXiv)

Spatial referring is a fundamental capability of embodied robots to interact with the 3D physical world. However, even with the powerful pretrained vision language models (VLMs), recent approaches are still not qualified to accurately understand the complex 3D scenes and dynamically reason about the instruction-indicated locations for interaction. To this end, we propose RoboRefer, a 3D-aware VLM that can first achieve precise spatial understanding by integrating a disentangled but dedicated depth encoder via supervised fine-tuning (SFT). Moreover, RoboRefer advances generalized multi-step spatial reasoning via reinforcement fine-tuning (RFT), with metric-sensitive process reward functions tailored for spatial referring tasks. To support SFT and RFT training, we introduce RefSpatial, a large-scale dataset of 20M QA pairs (2x prior), covering 31 spatial relations (vs. 15 prior) and supporting complex reasoning processes (up to 5 steps). In addition, we introduce RefSpatial-Bench, a challenging benchmark filling the gap in evaluating spatial referring with multi-step reasoning. Experiments show that SFT-trained RoboRefer achieves state-of-the-art spatial understanding, with an average success rate of 89.6%. RFT-trained RoboRefer further outperforms all other baselines by a large margin, even surpassing Gemini-2.5-Pro by 17.4% in average accuracy on RefSpatial-Bench. Notably, RoboRefer can be integrated with various control policies to execute long-horizon, dynamic tasks across diverse robots (e,g., UR5, G1 humanoid) in cluttered real-world scenes.