RT @RussTedrake: TRI's latest Large Behavior Model (LBM) paper landed on arxiv last night! Check out our project website: .

Excited to share our new work on multi-object scene completion and grasp pose estimation from a single RGB-D image!. Kudos to @s1wase and the incredible team from @ToyotaResearch, @WbyT_Tech, and @CarnegieMellon. Come chat with us at #CVPR2025 to learn more.

RT @NicholasEPfaff: Want to scale robot data with simulation, but don’t know how to get large numbers of realistic, diverse, and task-relev….

RT @s1wase: Check out our #ECCV2024 paper on "Zero-Shot Multi-Object Scene Completion"! Drop by poster 313 this morning if you're intereste….

RT @stephentian_: Learned visuomotor robot policies are sensitive to observation viewpoint shifts, which happen all the time. Can visual pr….

RT @mzubairirshad: Thrilled to share that our 𝐍𝐞𝐑𝐅-𝐌𝐀𝐄 paper has been accepted to #ECCV2024! ✨. How to effectively utilize large-scale NeRF….

Excited to introduce our paper, ReFiNe, at #SIGGRAPH2024 this Thursday! Learn how we encode multiple assets as continuous neural fields with high precision & low memory usage by exploiting object self-similarity. @RaresAmbrus @robo_kat @adnothing.Webpage:

RT @vannguyen_ng: 📢 Benchmark for 6D Object Pose Estimation 📢. BOP challenge 24 has been opened!. Results to be pr….

Check out our new exciting collaboration with the Robotics team at @Woven_Toyota on managing symmetry and uncertainty in 6D pose estimation using diffusion models! .

RT @sedrickkeh2: 📢 Releasing TRI's open-source Mamba-7B trained on 1.2T tokens of RefinedWeb!. Mamba-7B is the largest fully recurrent Mamb….

RT @basilevanh: Happy to share our #ICCV2023 paper on 3D reconstruction from a single image!. In Zero-1-to-3, we teach diffusion models to….

RT @MercatJean: How do you make safer plans without changing your planner? Change your predictions to be risk-aware! We present our new mod….

We show that a single network can decode over one thousand objects given a single latent vector per object, resulting in over 99% compression. (5/n)

Our latent space analysis shows that at higher LoDs similar areas of the projected latent space are increasingly shared by the different classes, suggesting that our approach efficiently encodes object geometry by learning geometric primitives common in the dataset. (4/n)

ROAD uses a lightweight representation that encodes models hierarchically such that the fidelity of the reconstruction increases as we traverse down an octree. The same network is called recursively as the level of detail increases. (3/n)

ROAD is a novel recursive implicit representation that can accurately encode scenes and large datasets of complex 3D shapes. Previous works either captured single scenes with high fidelity, or encoded multiple objects but struggled to reconstruct high frequency details. (2/n).

Excited to announce that our paper, ROAD: Learning an Implicit Recursive Octree Auto-Decoder to Efficiently Encode 3D Shapes, will be presented at #CORL2022 this Sunday! @RaresAmbrus @robo_kat @adnothing . Paper: Webpage: ⬇️(1/n)

RT @vincesitzmann: Considering a PhD and interested in differentiable rendering, self-supervised representation learning in vision, and 3D….

RT @mzubairirshad: 📢📢 Code release for "𝐒𝐡𝐀𝐏𝐎🎩: 𝐈𝐦𝐩𝐥𝐢𝐜𝐢𝐭 𝐑𝐞𝐩𝐫𝐞𝐬𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧𝐬 𝐟𝐨𝐫 𝐌𝐮𝐥𝐭𝐢 𝐎𝐛𝐣𝐞𝐜𝐭 𝐒𝐡𝐚𝐩𝐞 𝐀𝐩𝐩𝐞𝐚𝐫𝐚𝐧𝐜𝐞 𝐚𝐧𝐝 𝐏𝐨𝐬𝐞 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐚𝐭𝐢𝐨𝐧". Check it….

Excited to announce our new work PNDR: A novel approach towards sim-to-real adaptation by means of a neural ray tracer approximator with randomizable materials. #ECCV2022 @ToyotaResearch @WovenPlanet_JP @RaresAmbrus @vitorguizilini @wadimkehl @adnothing