ResearchResearch paperRobotics & Embodied AI1 source · Oct 20, 2026

SceneRig: An Agentic System for Simulation-Ready 3D Scene Reconstruction from Single Images

We present SceneRig, an agentic system that turns a single image into an editable, simulation-ready scene.

Key points

  • Simulation is a powerful tool for robotics, but building diverse, simulation-ready scenes at scale remains difficult.
  • Reconstructing scenes from single images is a promising alternative, but visual similarity alone is insufficient: objects must also be recovered, accurately placed, and physically grounded.
  • SceneRig combines geometric, visual, and physical tools to reconstruct a faithful 3D scene.
  • Furthermore, SceneRig is effective for downstream robotics: its scenes predict real-robot policy outcomes more accurately than the baseline method and more reliably support replay of recorded manipulations without trajectory adaptation.

Sources (1)

Extractive summary: sentences quoted from the sources.

Before this

  1. Oct 20, 2026How to train MOSS-Image LoRAs locally (8 GB VRAM): settings, tips and workflow
  2. Oct 19, 2026I built a free, local AI tool that works with PhotoCraft
  3. Oct 19, 2026Structured outputs for Anthropic Claude models on Amazon Bedrock
  4. Oct 19, 2026How to continue a GitHub Copilot app session in VS Code
  5. Oct 19, 2026Push-based KV transfer to improve TTFT and TPOT for disaggregated inference in vLLM
  6. Oct 19, 2026OV-PEAR: Observation-Verified Pose-Scale Estimation and Reconstruction for Unknown Industrial Manipulation Objects

Related