PanoWorld: Towards Spatial Supersensing in 360$^\circ$ Panorama World

TL;DR

This paper introduces PanoWorld, a model designed for ERP panorama understanding, leveraging geometry-aware supervision and spherical attention to improve spatial reasoning in 360-degree environments.

Contribution

It develops a new pano-native understanding framework with a geometry-aware training pipeline and a spherical spatial attention mechanism, advancing panoramic spatial reasoning.

Findings

PanoWorld outperforms baselines on multiple spatial reasoning benchmarks.

Geometry-aware supervision significantly improves ERP-native spatial understanding.

Spherical spatial cross-attention enhances model reasoning over panoramic data.

Abstract

Multimodal large laboratory models (MLLMs) still struggle with spatial understanding under the dominant perspective-image paradigm, which inherits the narrow field of view of human-like perception. For navigation, robotic search, and 3D scene understanding, 360-degree panoramic sensing offers a form of supersensing by capturing the entire surrounding environment at once. However, existing MLLM pipelines typically decompose panoramas into multiple perspective views, leaving the spherical structure of equirectangular projection (ERP) largely implicit. In this paper, we study pano-native understanding, which requires an MLLM to reason over an ERP panorama as a continuous, observer-centered space. To this end, we first define the key abilities for pano-native understanding, including semantic anchoring, spherical localization, reference-frame transformation, and depth-aware 3D spatial reasoning. We then build a large-scale metadata construction pipeline that converts mixed-source ERP panoramas into geometry-aware, language-grounded, and depth-aware supervision, and instantiate these signals as capability-aligned instruction tuning data. On the model side, we introduce PanoWorld with Spherical Spatial Cross-Attention, which injects spherical geometry into the visual stream. We further construct PanoSpace-Bench, a diagnostic benchmark for evaluating ERP-native spatial reasoning. Experiments show that PanoWorld substantially outperforms both proprietary and open-source baselines on PanoSpace-Bench, H* Bench, and R2R-CE Val-Unseen benchmarks. These results demonstrate that robust panoramic reasoning requires dedicated pano-native supervision and geometry-aware model adaptation. All source code and proposed data will be publicly released.