Daily arXiv paper reports focused on 3D vision, video generation, world models, robotics, Physical AI, Physical 3D, and intelligent/autonomous driving.
The latest daily arXiv batch must include every clearly relevant paper on:
- Video generation, including text-to-video, image-to-video, video diffusion, video synthesis/editing, controllable video generation, and generative video world models
- Physical AI, including embodied agents, robotics, manipulation, navigation, physical reasoning, physics-aware learning, and simulation
- Physical 3D, including physics-aware or physically grounded 3D reconstruction/generation, 4D dynamics, interaction, simulation, and 3D representations of the physical world
- Intelligent and autonomous driving across the full stack, including perception and sensor fusion, mapping/localization, scene and occupancy representation, motion/behavior prediction, planning and control, end-to-end and VLA driving, driving world models, cooperative driving/V2X, datasets and simulation, closed-loop evaluation, safety, robustness, and deployment
On the website, conventional automotive papers covering perception and sensor fusion, BEV/occupancy, mapping/localization, prediction, planning/control, V2X, simulation/evaluation, safety, or deployment are grouped into a collapsed Automotive Collection. Driving VLA and driving world-model papers remain visible in the main paper list, even when they also cover one of the grouped topics.
These are required coverage areas, not optional priority boosts. A paper in the latest eligible batch must not be omitted merely because it is outside a narrower interpretation of 3D vision. Apply the same publication-date verification and report format used for all other entries.
In addition to the topic-based filter above, pay extra attention to papers and project releases from the following authors and groups:
- Kaiming He
- Anpei Chen
- Shangzhe Wu
- Qianqian Wang
- Saining Xie
- Oxford Visual Geometry Group (VGG)
When these authors or groups appear in the author list, affiliations, project pages, or related lab releases, raise the reading priority even if the paper is slightly outside the usual daily themes. Still prefer work with a clear connection to 3D vision, video generation, generative/reconstruction systems, world models, embodied AI, robotics, Physical AI, Physical 3D, or intelligent/autonomous driving.