NVIDIA Pushes "Physical AI" Onto the Real World, Sidestepping Data Bottlenecks

NVIDIA's GTC 2026 focuses on 'Physical AI' for robots, using new tools like Cosmos and Isaac GR00T to train them with data from computers, not just the real world.

#PhysicalAI, #NVIDIAGTC, #Robotics, #AI, #EmbodiedAI

https://newsletter.tf/nvidia-physical-ai-robots-gtc-2026/

NVIDIA is making robots smarter by training them with computer-generated data, a new way to help them work in the real world.

#PhysicalAI, #NVIDIAGTC, #Robotics, #AI, #EmbodiedAI
https://newsletter.tf/nvidia-physical-ai-robots-gtc-2026/

NVIDIA GTC 2026: Physical AI Moves Robots From Labs to Real World

NVIDIA's GTC 2026 focuses on 'Physical AI' for robots, using new tools like Cosmos and Isaac GR00T to train them with data from computers, not just the real world.

NewsletterTF

NVIDIA Pushes "Physical AI" Onto the Real World, Sidestepping Data Bottlenecks

NVIDIA's GTC 2026 focuses on 'Physical AI' for robots, using new tools like Cosmos and Isaac GR00T to train them with data from computers, not just the real world.

#PhysicalAI, #NVIDIAGTC, #Robotics, #AI, #EmbodiedAI

https://newsletter.tf/nvidia-physical-ai-robots-gtc-2026/

NVIDIA is making robots smarter by training them with computer-generated data, a new way to help them work in the real world.

#PhysicalAI, #NVIDIAGTC, #Robotics, #AI, #EmbodiedAI
https://newsletter.tf/nvidia-physical-ai-robots-gtc-2026/

NVIDIA GTC 2026: Physical AI Moves Robots From Labs to Real World

NVIDIA's GTC 2026 focuses on 'Physical AI' for robots, using new tools like Cosmos and Isaac GR00T to train them with data from computers, not just the real world.

NewsletterTF
SAP and ANYbotics drive industrial adoption of physical AI

ANYbotics’ four-legged autonomous robots will be connected straight into SAP’s backend enterprise resource planning software.

AI News

Microsoft Research (@MSFTResearch)

AsgardBench는 시각 관찰을 바탕으로 임베디드 에이전트가 작업 중 계획을 수정할 수 있는지 평가하는 벤치마크다. 지각 기반 계획 능력에 초점을 맞춰 에이전트의 한계를 드러내고, 신뢰성 향상에 필요한 개선 방향을 제시한다.

https://x.com/MSFTResearch/status/2037244033475453210

#ai #benchmark #agents #embodiedai #planning

Microsoft Research (@MSFTResearch) on X

AsgardBench evaluates whether embodied agents can revise their plans based on visual observations as tasks unfold. By focusing on perception-driven planning, it exposes key limitations and guides improvements in agent reliability. https://t.co/6jAXzgCLvH

X (formerly Twitter)

田中義弘 | taziku CEO / AI × Creative (@taziku_co)

Roadrunner는 횡렬/인라인 바퀴와 스테핑을 모두 다루는 멀티모달 이동 로봇으로, 단일 정책으로 다양한 주행 방식을 학습했다. 넘어짐 복구와 한쪽 바퀴 균형까지 실기기에서 제로샷으로 동작해 범용 로봇 이동 제어의 가능성을 보여줬다.

https://x.com/taziku_co/status/2036373975698465266

#robotics #multimodal #reinforcementlearning #zeroshot #embodiedai

田中義弘 | taziku CEO / AI × Creative (@taziku_co) on X

マルチモーダル移動を実現 「Roadrunner(@rai_inst)」は、横並び車輪、インライン車輪、ステッピングを使いこなす車輪融合ロボット。 横並び車輪とインライン走行などを単一ポリシーで学習し、 起き上がりや片輪バランスまでゼロショットで実機展開。

X (formerly Twitter)

Danfei Xu (@danfei_xu)

사람의 1인칭 시점 데이터를 활용해 로봇을 학습시키는 생태계 EgoVerse가 소개됐다. 4개 연구소와 3개 산업 파트너가 참여했으며, 1300시간 이상·240개 장면·2000개 이상 과제를 포함하는 대규모 데이터셋과 연구 결과를 제공한다.

https://x.com/danfei_xu/status/2036108953017368960

#robotics #dataset #embodiedai #machinelearning #research

Danfei Xu (@danfei_xu) on X

Introducing EgoVerse: an ecosystem for robot learning from egocentric human data. Built and tested by 4 research labs + 3 industry partners, EgoVerse enables both science and scaling 1300+ hrs, 240 scenes, 2000+ tasks, and growing Dataset design, findings, and ecosystem 🧵

X (formerly Twitter)

fly51fly (@fly51fly)

강화학습을 활용해 로봇이 언제 생각해야 하는지 자원 인지형 추론을 학습하는 ‘Resource-Aware Reasoning’ 연구가 소개됐다. 임베디드 로보틱 의사결정에서 계산 자원을 절약하면서도 효율적인 추론을 목표로 하는 새로운 로봇 AI 접근이다.

https://x.com/fly51fly/status/2034385797588418905

#robotics #reinforcementlearning #embodiedai #reasoning #arxiv

fly51fly (@fly51fly) on X

[RO] When Should a Robot Think? Resource-Aware Reasoning via Reinforcement Learning for Embodied Robotic Decision-Making J Liu, P Zhao, Z Kong, X Shen… [CMU & Northeastern University & Harvard University] (2026) https://t.co/NdTpWenMq2

X (formerly Twitter)
How the Eon Team Produced a Virtual Embodied Fly

A technical deep dive into integrating the adult fly connectome, connectome-constrained brain models, and neuromechanical body simulations to build the first virtual embodied fly.