analiza aiqbench
This research introduces GaussianFactory, a system that generates high-fidelity robotic demonstrations using 3D Gaussian Splatting from a single video scan, eliminating the need for expensive manual data collection or heavy physics simulators. By reconstructing scenes as editable 3D replicas and planning kinematic trajectories, the method enables diffusion policies to achieve over 84% success rates in real-world deployment with minimal human input. The approach represents a significant step forward in reducing the data overhead required for training visuomotor policies while maintaining visual fidelity essential for generalization.
Demonstration Synthesis from a Single Scan via Gaussian Splatting for Visuomotor Policy Learning
arXiv:2609.21112v1 Announce Type: new Abstract: Training a visuomotor policy calls for abundant demonstrations that closely match the target environment, yet collecting them anew remains expensive. Existing demonstration synthesis methods reduce this cost but remain constrained by high manual effort, limited visual fidelity, or heavy reliance on physics simulators. This paper introduces GaussianFactory, a high-fidelity data engine that mass-produces demonstrations with a single video scan as its only human input and no physics engine in the generation loop. Specifically, GaussianFactory reconstructs the scene as an editable 3D Gaussian Splatting (3DGS) replica and samples from the object-combination tasks the scene affords. For each task, it plans grasps and trajectories purely kinematically on the geometric reconstruction, rendering photorealistic demonstrations that visually match the target environment. Physical dynamics enter the pipeline only where contact force interactions dictate the outcome---during grasp formation, via a learned contact model pretrained once on an interaction dataset. To evaluate the downstream utility of the synthesized demonstrations, we implement an end-to-end scan-to-deployment workflow in two setups: a simulated scene that stands in for the real world to enable reproducibility, and a real-world workspace with a physical UR10e robot. In each setup, a standard diffusion policy trained solely on the synthesized demonstrations achieves 95.1% and 84.2% success rates, respectively.
Abstract:Training a visuomotor policy calls for abundant demonstrations that closely match the target environment, yet collecting them anew remains expensive. Existing demonstration synthesis methods reduce this cost but remain constrained by high manual effort, limited visual fidelity, or heavy reliance on physics simulators. This paper introduces GaussianFactory, a high-fidelity data engine that mass-produces demonstrations with a single video scan as its only human input and no physics engine in the generation loop. Specifically, GaussianFactory reconstructs the scene as an editable 3D Gaussian Splatting (3DGS) replica and samples from the object-combination tasks the scene affords. For each task, it plans grasps and trajectories purely kinematically on the geometric reconstruction, rendering photorealistic demonstrations that visually match the target environment. Physical dynamics enter the pipeline only where contact force interactions dictate the outcome---during grasp formation, via a learned contact model pretrained once on an interaction dataset. To evaluate the downstream utility of the synthesized demonstrations, we implement an end-to-end scan-to-deployment workflow in two setups: a simulated scene that stands in for the real world to enable reproducibility, and a real-world workspace with a physical UR10e robot. In each setup, a standard diffusion policy trained solely on the synthesized demonstrations achieves 95.1% and 84.2% success rates, respectively.
Robotics (cs.RO)
arXiv:2609.21112 [cs.RO]
(or arXiv:2609.21112v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.21112 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 296 cuvinte
Subiecte conexe
[AINews] not much happened today
Latent Space · 18 sept. 2026
Cross-Platform Benchmark of Neural 3D Reconstruction for Autonomous Laboratory Robots
arXiv · 28 aug. 2026
FAST-GS: Frequency Aware Space-time Gaussian Splatting for Photorealistic Dynamic Novel View Synthesis
arXiv · 5 aug. 2026
Learning Geometrically-Grounded Amodal 3D Representations for View-Generalizable Robotic Manipulation
arXiv · 29 sept. 2026