PanoEnv: Exploring 3D Spatial Intelligence in Panoramic Environments with Reinforcement Learning

• URL: http://arxiv.org/abs/2602.21992v1
• Date: Wed, 25 Feb 2026 15:12:17 GMT
• Title: PanoEnv: Exploring 3D Spatial Intelligence in Panoramic Environments with Reinforcement Learning
• Authors: Zekai Lin, Xu Zheng,
• Abstract summary: 360 panoramic images are increasingly used in virtual reality, autonomous driving, and robotics for holistic scene understanding.<n>Current Vision-Language Models (VLMs) struggle with 3D spatial reasoning on Equirectangular Projection (ERP) images due to geometric distortion and limited 3D supervision.<n>We introduce PanoEnv, a large-scale VQA benchmark built from synthetic 3D environments.<n>Our 7B model achieves new state-of-the-art performance, improving overall accuracy to 52.93% (+3.59%) and open-ended accuracy to 14.83% while maintaining structured-task performance.
• Score: 5.308328605042682
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: 360 panoramic images are increasingly used in virtual reality, autonomous driving, and robotics for holistic scene understanding. However, current Vision-Language Models (VLMs) struggle with 3D spatial reasoning on Equirectangular Projection (ERP) images due to geometric distortion and limited 3D supervision. We introduce PanoEnv, a large-scale VQA benchmark built from synthetic 3D environments, containing 14.8K questions across five categories (e.g., relative position, volume comparison) grounded in accurate 3D annotations including depth, segmentation, and bounding boxes. Benchmarking 14 state-of-the-art VLMs reveals limited 3D understanding, achieving only 49.34% overall accuracy and 8.36% on open-ended (OE) questions. To enhance 3D reasoning, we propose a reinforcement learning post-training framework based on Group Relative Policy Optimization (GRPO) with a ground-truth-guided reward that incorporates five geometry-aware strategies such as distance tolerance and spatial consistency. A two-stage curriculum further mitigates catastrophic forgetting: Stage 1 trains on structured tasks (true/false and multiple choice), and Stage 2 fine-tunes on mixed open-ended data to improve generalization. Our 7B model achieves new state-of-the-art performance, improving overall accuracy to 52.93% (+3.59%) and open-ended accuracy to 14.83% while maintaining structured-task performance. It also achieves top semantic evaluation scores (Q-Score 6.24, P-Score 5.95), surpassing 32B models. These results demonstrate that PanoEnv-QA and our curriculum-based RL framework effectively instill 3D spatial intelligence in VLMs for omnidirectional perception.

Related papers

• Spa3R: Predictive Spatial Field Modeling for 3D Visual Reasoning [43.746951848993035]
  spatial intelligence can emerge from 2D vision alone, rather than being imposed via explicit spatial instruction tuning.<n>We introduce Spa3R, a self-supervised framework that learns a unified view-in spatial representation directly from unposed multi-view images.<n>Experiments demonstrate Spa3-VLM achieves state-of-the-art accuracy of 58.6% on 3D VQA, significantly outperforming prior methods.
  arXiv  Detail & Related papers  (2026-02-24T18:37:34Z)
• Video Spatial Reasoning with Object-Centric 3D Rollout [58.12446467377404]
  We propose Object-Centric 3D Rollout (OCR) to enable robust video spatial reasoning.<n>OCR introduces structured perturbations to the 3D geometry of selected objects during training.<n>OCR compels the model to reason holistically across the entire scene.
  arXiv  Detail & Related papers  (2025-11-17T09:53:41Z)
• Abstract 3D Perception for Spatial Intelligence in Vision-Language Models [100.13033631690114]
  Vision-language models (VLMs) struggle with 3D-related tasks such as spatial cognition and physical understanding.<n>We introduce SandboxVLM, a framework that leverages abstract bounding boxes to encode geometric structure and physical kinematics for VLM.<n>Our approach consistently improves spatial intelligence, achieving an 8.3% gain on SAT Real compared with baseline methods.
  arXiv  Detail & Related papers  (2025-11-14T04:16:09Z)
• ManiVID-3D: Generalizable View-Invariant Reinforcement Learning for Robotic Manipulation via Disentangled 3D Representations [19.02933938928656]
  ManiVID-3D is a novel 3D visual reinforcement learning architecture for robotic manipulation.<n>It learns view-invariant representations through self-supervised disentangled feature learning.<n>It achieves a 44.7% higher success rate than state-of-the-art methods under viewpoint variations.
  arXiv  Detail & Related papers  (2025-09-14T06:31:04Z)
• Move to Understand a 3D Scene: Bridging Visual Grounding and Exploration for Efficient and Versatile Embodied Navigation [54.04601077224252]
  Embodied scene understanding requires not only comprehending visual-spatial information but also determining where to explore next in the 3D physical world.<n>underlinetextbf3D vision-language learning enables embodied agents to effectively explore and understand their environment.<n>model's versatility enables navigation using diverse input modalities, including categories, language descriptions, and reference images.
  arXiv  Detail & Related papers  (2025-07-05T14:15:52Z)
• Zero-shot Inexact CAD Model Alignment from a Single Image [53.37898107159792]
  A practical approach to infer 3D scene structure from a single image is to retrieve a closely matching 3D model from a database and align it with the object in the image.<n>Existing methods rely on supervised training with images and pose annotations, which limits them to a narrow set of object categories.<n>We propose a weakly supervised 9-DoF alignment method for inexact 3D models that requires no pose annotations and generalizes to unseen categories.
  arXiv  Detail & Related papers  (2025-07-04T04:46:59Z)
• E3D-Bench: A Benchmark for End-to-End 3D Geometric Foundation Models [78.1674905950243]
  We present the first comprehensive benchmark for 3D geometric foundation models (GFMs)<n>GFMs directly predict dense 3D representations in a single feed-forward pass, eliminating the need for slow or unavailable precomputed camera parameters.<n>We evaluate 16 state-of-the-art GFMs, revealing their strengths and limitations across tasks and domains.<n>All code, evaluation scripts, and processed data will be publicly released to accelerate research in 3D spatial intelligence.
  arXiv  Detail & Related papers  (2025-06-02T17:53:09Z)
• VLM-3R: Vision-Language Models Augmented with Instruction-Aligned 3D Reconstruction [86.82819259860186]
  We introduce VLM-3R, a unified framework for Vision-Language Models (VLMs) that incorporates 3D Reconstructive instruction tuning.<n>VLM-3R processes monocular video frames by employing a geometry encoder to derive implicit 3D tokens that represent spatial understanding.
  arXiv  Detail & Related papers  (2025-05-26T17:56:30Z)
• SURDS: Benchmarking Spatial Understanding and Reasoning in Driving Scenarios with Vision Language Models [15.50826328938879]
  We introduce SURDS, a benchmark designed to evaluate the spatial reasoning capabilities of vision language models (VLMs)<n>Built on the nuScenes dataset, SURDS comprises 41,080 vision-question-answer training instances and 9,250 evaluation samples.<n>We propose a reinforcement learning-based alignment scheme leveraging spatially grounded reward signals.
  arXiv  Detail & Related papers  (2024-11-20T08:14:01Z)
• On Deep Learning for Geometric and Semantic Scene Understanding Using On-Vehicle 3D LiDAR [4.606106768645647]
  3D LiDAR point cloud data is crucial for scene perception in computer vision, robotics, and autonomous driving. We present DurLAR, the first high-fidelity 128-channel 3D LiDAR dataset featuring panoramic ambient (near infrared) and reflectivity imagery. To improve the segmentation accuracy, we introduce Range-Aware Pointwise Distance Distribution (RAPiD) features and the associated RAPiD-Seg architecture.
  arXiv  Detail & Related papers  (2024-11-01T14:01:54Z)

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.