Spatial Memory Intelligence: Endowing World Models with Understanding-Driven Long-Term Memory
cs.CV
Submitted: 2026-10-01
Updated: 2026-10-08
Comments: 32 pages. Project page: https://spatial-memory-intelligence.github.io/
Project page: https://spatial-memory-intelligence.github.io
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Cosmos 3: Omnimodal World Models for Physical AI
- Qwen3-VL Technical Report
- Qwen2.5-VL Technical Report
- Mixture of Contexts for Long Video Generation
- ABot-PhysWorld: Interactive World Foundation Model for Robotic Manipulation with Physics Alignment
- Rethinking Attention with Performers
- Emerging Properties in Unified Multimodal Pretraining
- Infinite Worlds with Versatile Interactions
- LVSA: Training-Free Sparse Attention for Long Video Diffusion
- World Models
- LTX-Video: Realtime Video Latent Diffusion
- Mastering Diverse Domains through World Models
- RELIC: Interactive Video World Model with Long-Horizon Memory
- Multiplayer Interactive World Models with Representation Autoencoders
- Memory Forcing: Spatio-Temporal Memory for Consistent Scene Generation on Minecraft
- MemFlow: Flowing Adaptive Memory for Consistent and Efficient Long Video Narratives
- Retrieve What's Missing: Coverage-Maximizing Retrieval for Consistent Long Video Generation
- Echo-Memory: A Controlled Study of Memory in Action World Models
- Hunyuan-GameCraft: High-dynamic Interactive Game Video Generation with Hybrid History Condition
- Causal World Modeling for Robot Control
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models