i second this! At least for those working on robot simulation there’s a lot of small details that go into making sim fast, rl+sim fast, making sim working in the first place, sim2real and real2sim!
I think we need an AMA series for Robotics / Embodied AI with an optional anonymous setting. Will be both fun and informative to new community members to absorb folk knowledge.
Does everyone in your community agree on some folk knowledge that isn’t published anywhere? Put it in a paper! It’s a pretty valuable contribution
November 28, 2024 at 6:28 AM
i second this! At least for those working on robot simulation there’s a lot of small details that go into making sim fast, rl+sim fast, making sim working in the first place, sim2real and real2sim!
They even note that there is a skill library that they provide to the system beforehand and the reconstruction requires "Human-recorded video" (see FAQ "What is required for accurate Real2Sim reconstruction?").
September 27, 2026 at 5:10 PM
They even note that there is a skill library that they provide to the system beforehand and the reconstruction requires "Human-recorded video" (see FAQ "What is required for accurate Real2Sim reconstruction?").
🤖 Li Feifei's Team Shifts from Sim2Real to Real2Sim for Robot Training
Li Feifei's team and NVIDIA GEAR have started using Real2Sim approach with their new SimFoundry system, moving away from traditional Sim2Real methods for robot...
#Robotics #GenerativeAI #DataEfficiency #AI #AIPulse
Li Feifei's team and NVIDIA GEAR have started using Real2Sim approach with their new SimFoundry system, moving away from traditional Sim2Real methods for robot...
#Robotics #GenerativeAI #DataEfficiency #AI #AIPulse
Read the full article →
www.synestesia.uk
July 5, 2026 at 12:40 PM
🤖 Li Feifei's Team Shifts from Sim2Real to Real2Sim for Robot Training
Li Feifei's team and NVIDIA GEAR have started using Real2Sim approach with their new SimFoundry system, moving away from traditional Sim2Real methods for robot...
#Robotics #GenerativeAI #DataEfficiency #AI #AIPulse
Li Feifei's team and NVIDIA GEAR have started using Real2Sim approach with their new SimFoundry system, moving away from traditional Sim2Real methods for robot...
#Robotics #GenerativeAI #DataEfficiency #AI #AIPulse
We have zero safety here (the same thing that killed the massive circa 2016 hype around self-driving cars).
Homebody's own researchers noted that the "fingers" on the Unitree G1 robot began overheating over extended operational periods.
The impressiveness of the software part is moot as it stands.
Homebody's own researchers noted that the "fingers" on the Unitree G1 robot began overheating over extended operational periods.
The impressiveness of the software part is moot as it stands.
September 27, 2026 at 4:37 PM
We have zero safety here (the same thing that killed the massive circa 2016 hype around self-driving cars).
Homebody's own researchers noted that the "fingers" on the Unitree G1 robot began overheating over extended operational periods.
The impressiveness of the software part is moot as it stands.
Homebody's own researchers noted that the "fingers" on the Unitree G1 robot began overheating over extended operational periods.
The impressiveness of the software part is moot as it stands.
And thanks to all the organizers for working together to make this a successful workshop! Including
@egavves.bsky.social @cgumbsch.bsky.social @yunzhu-li.bsky.social @zadaianchuk.bsky.social
+ many others!
#CoRL2025 #sim2real #real2sim #robotworldmodels #robotics
@egavves.bsky.social @cgumbsch.bsky.social @yunzhu-li.bsky.social @zadaianchuk.bsky.social
+ many others!
#CoRL2025 #sim2real #real2sim #robotworldmodels #robotics
August 1, 2025 at 6:24 PM
And thanks to all the organizers for working together to make this a successful workshop! Including
@egavves.bsky.social @cgumbsch.bsky.social @yunzhu-li.bsky.social @zadaianchuk.bsky.social
+ many others!
#CoRL2025 #sim2real #real2sim #robotworldmodels #robotics
@egavves.bsky.social @cgumbsch.bsky.social @yunzhu-li.bsky.social @zadaianchuk.bsky.social
+ many others!
#CoRL2025 #sim2real #real2sim #robotworldmodels #robotics
CRISP: Contact-Guided Real2Sim from Monocular Video with Planar Scene Primitives
Deva Ramanan, Jeff Tan et al.
Paper
Details
#ComputerVision #MonocularVideo #ScenePrimitives
Deva Ramanan, Jeff Tan et al.
Paper
Details
#ComputerVision #MonocularVideo #ScenePrimitives
December 17, 2025 at 9:01 AM
CRISP: Contact-Guided Real2Sim from Monocular Video with Planar Scene Primitives
Deva Ramanan, Jeff Tan et al.
Paper
Details
#ComputerVision #MonocularVideo #ScenePrimitives
Deva Ramanan, Jeff Tan et al.
Paper
Details
#ComputerVision #MonocularVideo #ScenePrimitives
K-Scale has a few awesome videos breaking down their hardware and real2sim methods -- recommend checking them out. All their stuff is open source ❤️
www.kscale.dev/kbot
www.kscale.dev/kbot
October 15, 2025 at 8:02 PM
K-Scale has a few awesome videos breaking down their hardware and real2sim methods -- recommend checking them out. All their stuff is open source ❤️
www.kscale.dev/kbot
www.kscale.dev/kbot
RoboGSim: A Real2Sim2Real Robotic Gaussian Splatting Simulator
Authors: Xinhai Li, Jialin Li et al.
pre-print -> arxiv.org/abs/2411.11839
website -> robogsim.github.io
#robotics #data_generation #sim2real #real2sim #real2sim2real
Authors: Xinhai Li, Jialin Li et al.
pre-print -> arxiv.org/abs/2411.11839
website -> robogsim.github.io
#robotics #data_generation #sim2real #real2sim #real2sim2real
November 19, 2024 at 12:23 PM
RoboGSim: A Real2Sim2Real Robotic Gaussian Splatting Simulator
Authors: Xinhai Li, Jialin Li et al.
pre-print -> arxiv.org/abs/2411.11839
website -> robogsim.github.io
#robotics #data_generation #sim2real #real2sim #real2sim2real
Authors: Xinhai Li, Jialin Li et al.
pre-print -> arxiv.org/abs/2411.11839
website -> robogsim.github.io
#robotics #data_generation #sim2real #real2sim #real2sim2real
"Unlike vision and language, data for learning is not available passively[...]. This makes applying the same recipes we did in vision and language challenging"
#real2sim is a strong emergent tendency in robotics this year.
See previously shared articles
🧵
#real2sim is a strong emergent tendency in robotics this year.
See previously shared articles
🧵
December 3, 2024 at 3:07 PM
"Unlike vision and language, data for learning is not available passively[...]. This makes applying the same recipes we did in vision and language challenging"
#real2sim is a strong emergent tendency in robotics this year.
See previously shared articles
🧵
#real2sim is a strong emergent tendency in robotics this year.
See previously shared articles
🧵
hmm, i think there is a lot of interest in better real2sim via view synthesis (dr robot, robot see robot do, d3 fields were all oral talks). some posters about learning object-centric "flow" from video (notably pi0 but there are several more posters proposing similar work).
November 16, 2024 at 6:42 PM
hmm, i think there is a lot of interest in better real2sim via view synthesis (dr robot, robot see robot do, d3 fields were all oral talks). some posters about learning object-centric "flow" from video (notably pi0 but there are several more posters proposing similar work).
Davide Bargellini, Alex Pasquali, Andrea Govoni, Riccardo Zanella, Gianluca Palli
Enhancing Sim2Real Transfer for Torque-Controlled Robots through Real2Sim Dynamics Estimation and Reinforcement Learning
https://arxiv.org/abs/2608.22629
Enhancing Sim2Real Transfer for Torque-Controlled Robots through Real2Sim Dynamics Estimation and Reinforcement Learning
https://arxiv.org/abs/2608.22629
August 25, 2026 at 7:02 PM
Davide Bargellini, Alex Pasquali, Andrea Govoni, Riccardo Zanella, Gianluca Palli
Enhancing Sim2Real Transfer for Torque-Controlled Robots through Real2Sim Dynamics Estimation and Reinforcement Learning
https://arxiv.org/abs/2608.22629
Enhancing Sim2Real Transfer for Torque-Controlled Robots through Real2Sim Dynamics Estimation and Reinforcement Learning
https://arxiv.org/abs/2608.22629
Bargellini, Pasquali, Govoni, Zanella, Palli: Enhancing Sim2Real Transfer for Torque-Controlled Robots through Real2Sim Dynamics Estimation and Reinforcement Learning https://arxiv.org/abs/2608.22629 https://arxiv.org/pdf/2608.22629 https://arxiv.org/html/2608.22629
August 25, 2026 at 6:45 AM
Bargellini, Pasquali, Govoni, Zanella, Palli: Enhancing Sim2Real Transfer for Torque-Controlled Robots through Real2Sim Dynamics Estimation and Reinforcement Learning https://arxiv.org/abs/2608.22629 https://arxiv.org/pdf/2608.22629 https://arxiv.org/html/2608.22629
Guanxiong Chen, et al.: Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents https://arxiv.org/abs/2607.19190 https://arxiv.org/pdf/2607.19190 https://arxiv.org/html/2607.19190
July 22, 2026 at 6:44 AM
Guanxiong Chen, et al.: Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents https://arxiv.org/abs/2607.19190 https://arxiv.org/pdf/2607.19190 https://arxiv.org/html/2607.19190
Guanxiong Chen, Qianjun Xia, Jiawei Peng, Heng Zhang, Bole Ma, Justin Qian, Ziyi Jiao, Bingyang Zhou, Luoxin Ye, Kaifeng Zhang, Kunyi Wang, Weijia Zeng, Yunuo Chen, ...
Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents
https://arxiv.org/abs/2607.19190
Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents
https://arxiv.org/abs/2607.19190
July 22, 2026 at 8:38 AM
Guanxiong Chen, Qianjun Xia, Jiawei Peng, Heng Zhang, Bole Ma, Justin Qian, Ziyi Jiao, Bingyang Zhou, Luoxin Ye, Kaifeng Zhang, Kunyi Wang, Weijia Zeng, Yunuo Chen, ...
Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents
https://arxiv.org/abs/2607.19190
Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents
https://arxiv.org/abs/2607.19190
Yang, Wang, Tan, Chan, Wang, Yan, Li, Ang: Beyond Heuristics: A Standardized Real2Sim Pipeline for Physical Human Robot Interaction in Human-in-the-Loop Simulation https://arxiv.org/abs/2607.03017 https://arxiv.org/pdf/2607.03017 https://arxiv.org/html/2607.03017
July 7, 2026 at 6:44 AM
Yang, Wang, Tan, Chan, Wang, Yan, Li, Ang: Beyond Heuristics: A Standardized Real2Sim Pipeline for Physical Human Robot Interaction in Human-in-the-Loop Simulation https://arxiv.org/abs/2607.03017 https://arxiv.org/pdf/2607.03017 https://arxiv.org/html/2607.03017
Real2Sim: A Revolutionary Replacement for Sim2Real by NVIDIA and Fei-Fei Li's New System
#NVIDIA #SimFoundry #Real2Sim
https://singulism.com/en/2026-07-05-sim2real-simfoundry-robotics-ai
#NVIDIA #SimFoundry #Real2Sim
https://singulism.com/en/2026-07-05-sim2real-simfoundry-robotics-ai
Real2Sim: A Revolutionary Replacement for Sim2Real by NVIDIA and Fei-Fei Li's New System
NVIDIA GEAR and Fei-Fei Li's team have developed SimFoundry, which can automatically generate robot simulation environments from real-world videos, enabling nearly infinite training data through objec
singulism.com
July 5, 2026 at 4:32 PM
Real2Sim: A Revolutionary Replacement for Sim2Real by NVIDIA and Fei-Fei Li's New System
#NVIDIA #SimFoundry #Real2Sim
https://singulism.com/en/2026-07-05-sim2real-simfoundry-robotics-ai
#NVIDIA #SimFoundry #Real2Sim
https://singulism.com/en/2026-07-05-sim2real-simfoundry-robotics-ai
💼 The Real2Sim approach could disrupt traditional robotics training by drastically cutting costs and time, potentially eroding the data collection moat that has protected some players in the industry.
July 5, 2026 at 12:40 PM
💼 The Real2Sim approach could disrupt traditional robotics training by drastically cutting costs and time, potentially eroding the data collection moat that has protected some players in the industry.
Super exciting work, Pr. Gupta. Regarding the real2sim approach, the paper states, "Digital twins are obtained
directly from real-world videos [..] using photogrammetry methods such as Gaussian splatting and neural radiance fields". Do you use both? Only these 2? Why not settle for only one?
directly from real-world videos [..] using photogrammetry methods such as Gaussian splatting and neural radiance fields". Do you use both? Only these 2? Why not settle for only one?
December 5, 2024 at 7:58 AM
Super exciting work, Pr. Gupta. Regarding the real2sim approach, the paper states, "Digital twins are obtained
directly from real-world videos [..] using photogrammetry methods such as Gaussian splatting and neural radiance fields". Do you use both? Only these 2? Why not settle for only one?
directly from real-world videos [..] using photogrammetry methods such as Gaussian splatting and neural radiance fields". Do you use both? Only these 2? Why not settle for only one?
Alessandro Adami, Sebastian Zudaire, Ruggero Carli, Pietro Falco
Real2Sim based on Active Perception with automatically VLM-generated Behavior Trees
https://arxiv.org/abs/2601.08454
Real2Sim based on Active Perception with automatically VLM-generated Behavior Trees
https://arxiv.org/abs/2601.08454
January 14, 2026 at 2:44 PM
Alessandro Adami, Sebastian Zudaire, Ruggero Carli, Pietro Falco
Real2Sim based on Active Perception with automatically VLM-generated Behavior Trees
https://arxiv.org/abs/2601.08454
Real2Sim based on Active Perception with automatically VLM-generated Behavior Trees
https://arxiv.org/abs/2601.08454
Nicholas Pfaff, Evelyn Fu, Jeremy Binagia, Phillip Isola, Russ Tedrake
Scalable Real2Sim: Physics-Aware Asset Generation Via Robotic Pick-and-Place Setups
https://arxiv.org/abs/2503.00370
Scalable Real2Sim: Physics-Aware Asset Generation Via Robotic Pick-and-Place Setups
https://arxiv.org/abs/2503.00370
March 4, 2025 at 11:55 AM
Nicholas Pfaff, Evelyn Fu, Jeremy Binagia, Phillip Isola, Russ Tedrake
Scalable Real2Sim: Physics-Aware Asset Generation Via Robotic Pick-and-Place Setups
https://arxiv.org/abs/2503.00370
Scalable Real2Sim: Physics-Aware Asset Generation Via Robotic Pick-and-Place Setups
https://arxiv.org/abs/2503.00370
Yubo Zhao, Yujin Chai, Yunao Dong, Chengfeng Zhao, Zijiao Zeng, Yuan Liu, Chi-Keung Tang: Real2Sim in HOI: Toward Physically Plausible HOI Reconstruction from Monocular Videos https://arxiv.org/abs/2605.14462 https://arxiv.org/pdf/2605.14462 https://arxiv.org/html/2605.14462
May 15, 2026 at 6:40 AM
Yubo Zhao, Yujin Chai, Yunao Dong, Chengfeng Zhao, Zijiao Zeng, Yuan Liu, Chi-Keung Tang: Real2Sim in HOI: Toward Physically Plausible HOI Reconstruction from Monocular Videos https://arxiv.org/abs/2605.14462 https://arxiv.org/pdf/2605.14462 https://arxiv.org/html/2605.14462
Kaicong Huang, Talha Azfar, Weisong Shi, Ruimin Ke: Real2Sim: A Physics-driven and Editable Gaussian Splatting Framework for Autonomous Driving Scenes https://arxiv.org/abs/2605.13591 https://arxiv.org/pdf/2605.13591 https://arxiv.org/html/2605.13591
May 14, 2026 at 6:41 AM
Kaicong Huang, Talha Azfar, Weisong Shi, Ruimin Ke: Real2Sim: A Physics-driven and Editable Gaussian Splatting Framework for Autonomous Driving Scenes https://arxiv.org/abs/2605.13591 https://arxiv.org/pdf/2605.13591 https://arxiv.org/html/2605.13591
Zihan Wang, Jiashun Wang, Jeff Tan, Yiwen Zhao, Jessica Hodgins, Shubham Tulsiani, Deva Ramanan: CRISP: Contact-Guided Real2Sim from Monocular Video with Planar Scene Primitives https://arxiv.org/abs/2512.14696 https://arxiv.org/pdf/2512.14696 https://arxiv.org/html/2512.14696
December 17, 2025 at 6:31 AM
Zihan Wang, Jiashun Wang, Jeff Tan, Yiwen Zhao, Jessica Hodgins, Shubham Tulsiani, Deva Ramanan: CRISP: Contact-Guided Real2Sim from Monocular Video with Planar Scene Primitives https://arxiv.org/abs/2512.14696 https://arxiv.org/pdf/2512.14696 https://arxiv.org/html/2512.14696
arXiv:2503.04082v1 Announce Type: new
Abstract: Real2Sim is becoming increasingly important with the rapid development of surgical artificial intelligence (AI) and autonomy. In this work, we propose a novel Real2Sim methodology, [1/6 of https://arxiv.org/abs/2503.04082v1]
Abstract: Real2Sim is becoming increasingly important with the rapid development of surgical artificial intelligence (AI) and autonomy. In this work, we propose a novel Real2Sim methodology, [1/6 of https://arxiv.org/abs/2503.04082v1]
March 7, 2025 at 6:28 AM
arXiv:2503.04082v1 Announce Type: new
Abstract: Real2Sim is becoming increasingly important with the rapid development of surgical artificial intelligence (AI) and autonomy. In this work, we propose a novel Real2Sim methodology, [1/6 of https://arxiv.org/abs/2503.04082v1]
Abstract: Real2Sim is becoming increasingly important with the rapid development of surgical artificial intelligence (AI) and autonomy. In this work, we propose a novel Real2Sim methodology, [1/6 of https://arxiv.org/abs/2503.04082v1]