RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions Paper • 2610.34356 • Published 22 days ago • 11
SynthVerse Collection A large-scale synthetic dataset and benchmark for point tracking across diverse domains, dynamic motions, and embodied interactions. • 3 items • Updated 26 days ago • 1
ForcingBooster: Adaptive Patch-wise Denoising for Fast Autoregressive Video Diffusion Paper • 2610.12625 • Published Oct 8 • 32
ACE-Data-0: Human-Centric Ambient Capture as Embodied Data Engine Paper • 2607.28625 • Published Jul 30 • 44
RynnWorld-Teleop: An Action-Conditioned World Model for Digital Teleoperation Paper • 2607.06558 • Published Jul 7 • 23
RynnWorld-4D: 4D Embodied World Models for Robotic Manipulation Paper • 2607.06559 • Published Jul 7 • 37
Code-as-Room: Generating 3D Rooms from Top-Down View Images via Agentic Code Synthesis Paper • 2605.18451 • Published May 18 • 41
ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning Paper • 2604.24300 • Published Apr 27 • 67
CoMoVi: Co-Generation of 3D Human Motions and Realistic Videos Paper • 2601.10632 • Published Jan 15 • 8
SynthVerse: A Large-Scale Diverse Synthetic Dataset for Point Tracking Paper • 2602.04441 • Published Feb 4 • 2
PLANING: A Loosely Coupled Triangle-Gaussian Framework for Streaming 3D Reconstruction Paper • 2601.22046 • Published Jan 29 • 21
MesaTask: Towards Task-Driven Tabletop Scene Generation via 3D Spatial Reasoning Paper • 2509.22281 • Published Sep 26, 2025 • 32