NVIDIA7Skild AI2ADLINK1Apple1Applied Intuition1articulated assets1Astribot1Axis Robotics1Boston Dynamics1development timeline of Moby.1development timeline reduction.1examples for adapting to new robotic embodiments.1Fudan University1Google DeepMind1Hugging Face1Lightwheel1LimX Dynamics1LM Studio1manipulation tasks1Moby deployment in a semiconductor facility.1Noble Machines1Ollama1Perplexity1post-training duration1SAGE-10K dataset1Scale AI1Scale AI data delivery1Schaeffler1Seeed1Solomon1Stretch1success rate of the post-trained policy.1tasks available in RoboLab.1The University of Hong Kong1Training Data1training hours for LDA-1B model1training needed for current VLA models to match S1's accuracy1TranscEngram1TRON 21validated run uses 64 nodes of 4× GB200 for 60K iterations, roughly 68 hours (17.4K GB200-hours).1Zhejiang University1ZiNovaLabs1
Who captures
NVIDIA7Skild AI2Stretch1Astribot1@smsehy1Perplexity1https://x.com/VaderResearch1ZiNovaLabs1The University of Hong Kong1https://x.com/RoboPapers1Axis Robotics1Google DeepMind1
Sourced numbers
manipulation tasks120 units“across 120 language-conditioned manipulation tasks”@NVIDIARobotics on X
The harness problem is getting its own framework category. Tardigrade ships today: an agent harness built as typed state machine components over an immutable event log. The framing is React for harness. Each component knows its previous states. The whole harness is a pure function of the log. I keep seeing teams reach this conclusion from two directions. One: they try to build a stateful agent loop and hit the problem of debugging mid-run failures, resuming interrupted sessions, and auditing what the agent actually did and when. Two: they want to add AI behavior to existing Effect/functional…
3:03 PM ET
ASAstribot
3:03 PM ETASAstribot
@XRoboHub on X
Robots can’t stop while the model thinks. In a high-speed throw, even one pause can kill the momentum. Astribot released SmoothRL for online RL during async inference. S1 keeps moving as the model computes the next action chunk. Most actions in a chunk never execute. Train on the full chunk, and RL credits or blames moves that never happened. SmoothRL learns only from executed actions, matching real deployment timing. After 250 rollouts, tossing jumped 39%→94%, pen capping 8%→83%, and box opening 30%→90%. One autonomous toss cut acceleration RMS by 52% and jerk RMS by 47%. The model keeps…
5:17 AM ET
@smsehy
5:17 AM ET@smsehy
@smsehy on X
Writing a custom robotics runtime in Rust eliminates garbage collection latency spikes on the factory floor, but maintaining sub-millimeter trajectory repeatability across varying ambient thermal conditions remains the core challenge in production welding. https://t.co/q21icpHIJu
7:34 PM ET
PEPerplexity
7:34 PM ETPEPerplexity
@stretchcloud on X
The hybrid compute pattern is getting serious. Perplexity just open-sourced Lily, the local inference engine powering their Mac hybrid compute feature, and the architecture choices are worth understanding.Lily runs Qwen3.6-35B-A3B on Apple silicon using a Rust runtime with custom Metal kernels. The optimization split is separate paths for prefill and decode. That matters because the bottleneck on local AI models shifts depending on whether you're processing a long input or generating output tokens. Lily tunes both separately to match what the hardware can actually do.The use case is cleaner…
3:03 PM ET
https://x.com/VaderResearch
3:03 PM EThttps://x.com/VaderResearch
mete on X: "Buying a quadruped? Consider this:
✸ Wheels vs legs change where it can go
✸ Payload on paper ≠ payload in motion
✸ Runtime changes with load and terrain
✸ Top speed rarely means useful speed" / X
Post Log in Sign up Post mete on X: "Buying a quadruped? Consider this: ✸ Wheels vs legs change where it can go ✸ Payload on paper ≠ payload in motion ✸ Runtime changes with load and terrain ✸ Top speed rarely means useful speed" - mete @VaderResearch Buying a quadruped? Consider this: ✸ Wheels vs legs change where it can go ✸ Payload on paper ≠ payload in motion ✸ Runtime changes with load and terrain ✸ Top speed rarely means useful speed View media 3:03 PM · Sep 2, 2026 Reply
12:00 PM ET
NVIDIA
$NVDA
12:00 PM ETNVIDIA
@NVIDIARobotics on X
🤖 A robot policy can be compact enough to run onboard and fast enough to keep the arm moving continuously. Post-trained NVIDIA Cosmos 3 Edge on NVIDIA Jetson AGX Thor T5000 generates roughly 2.13 seconds of robot motion in about 1.53 seconds, then replans from fresh camera and robot-state observations. Developers can test the same policy in closed-loop RoboLab simulation across 120 language-conditioned manipulation tasks before evaluating on hardware.
10:31 PM ET
ZIZiNovaLabs
10:31 PM ETZIZiNovaLabs
LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows.
@ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, includi… / X
Post Log in Sign up Post LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows. @ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, including formwork assembly, multi-layer rebar placement and tying. TRON 2 serves as a modular and extensible embodied robotic platform for multi-tool, multi-step tasks across large workspaces. Its dual arms handle construction tools and materials across different orientations and…
7:00 PM ET
Nvidia
$NVDA
7:00 PM ETNvidia
Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA
Manufacturing \| Robotics Noble Machines Accelerates General Purpose Industrial Robot Development With NVIDIA Isaac GR00T Learn More Objective Noble Machines is advancing industrial robot development with the Moby general-purpose industrial robot and its supporting software stack, including a proprietary AI driven whole-body-control system. Moby is designed for complex, hazardous, and labor-intensive work across factories, logistics centers, construction sites, and semiconductor facilities, where operating reliably requires advanced perception, reasoning, adaptation, and physical interaction.…
2:09 PM ET
Nvidia
$NVDA
2:09 PM ETNvidia
Seeed reBot Arm B601-RS: Physical AI & VLA Model Course | NVIDIA DLI Series
Warehouse China Warehouse US Warehouse Germany Warehouse The store will not work correctly in the case when cookies are disabled. US Warehouse: Enjoy FREE UNIUNI shipping on orders under $50! (Excludes XIAO & Raspberry Pi series) \\ \\ the AI Hardware Partner For Industry Products Local Warehouse Documents Customization Solution Software Support Open Claw Quick Order Account Get more with a SeeedStudio accountclear \\ \\ 1 on 1 Support \\ \\ Local Delivery \\ \\ 30-Day DOA Guarantee \\ \\ 30-Day Returns Sign In Create Account My Orders\\ Track, change, cancel Account Information\\ Update…
5:00 PM ET
TUThe University of Hong Kong
5:00 PM ETTUThe University of Hong Kong
The Imitator Game — Benchmarking Robot Imitative Ability Beyond Action Prediction
Intent imitation, L0-L320,000+ paired episodesHuman evaluation Arena The Imitator Game BenchmarkingRobotImitativeAbilityBeyondActionPrediction Xunzhe Zhou 1,2,\,† · Yiyang Cai 2,3,\ · Fengyi Wang 2,3,\ · Ran Ju 1,2,\ · Hanxiang Ren 2,4 · Ruizhe Liu 1 Yu Zhang 1 · Qian Luo 1,2 · Feng Chen 1 · Pei Zhou 1,2 · Yi Ma 1,2 · Yanchao Yang 1,2,‡ 1The University of Hong Kong · 2TranscEngram · 3Fudan University · 4Zhejiang University \\ Equal contribution · † Project lead · ‡ Corresponding author Can a robot imitate what a human intends — not just what they do? We widen the gap between the demonstration…
4:05 PM ET
NVIDIA
$NVDA
4:05 PM ETNVIDIA
How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents | NVIDIA Technical Blog
Technical Blog Subscribe Related Resources Robotics English한국어中文 How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents Aug 26, 2026 By Yan Chang, Mihir Acharya, Wei Liu, Katie Washabaugh and Aishwarya Singh +9 Like Discuss (0) - L - T - F - R - E AI-Generated Summary - COMPASS adapts the pretrained NVIDIA X-Mobility policy into residual specialists for specific robots and environments using reinforcement learning. - An agent-driven workflow with human approval gates automates environment validation, scene preparation, smoke testing, residual training, checkpoint evaluation,…
1:19 PM ET
Skild AI
1:19 PM ETSkild AI
Skild AI on X: "We found that to match the accuracy that S1 can achieve with just one example of prompting, current VLA models would need to be post-trained with 50-100 hours of data collection followed by fine-tuning!" / X
Post Log in Sign up Post Skild AI on X: "We found that to match the accuracy that S1 can achieve with just one example of prompting, current VLA models would need to be post-trained with 50-100 hours of data collection followed by fine-tuning!" - Skild AI @SkildAI Aug 25 Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 427 1K 7.1K 3.4M - Skild AI @SkildAI Aug 25 It can be taught…
1:19 PM ET
Skild AI
1:19 PM ETSkild AI
Skild AI on X: "Introducing S1, our new foundation model that learns from one example.
It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.
Watch S1 operate in real-time via in-context learning:" / X
Post Log in Sign up Post Skild AI on X: "Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning:" - Skild AI @SkildAI Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 1:19 PM · Aug 25, 20263.4MViews 427…
9:15 AM ET
https://x.com/RoboPapers
9:15 AM EThttps://x.com/RoboPapers
RoboPapers on X: "Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data.
Part of the secret is that LDA-1B jointly learns forward … / X
Post Log in Sign up Post RoboPapers on X: "Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data. Part of the secret is that LDA-1B jointly learns forward dynamics, action prediction, and visual forecasting, all in a structured DINO latent space which avoids the pitfalls of redundant pixel-level prediction which isn’t necessarily aligned robot action. This approach works on both dexterous hands and simple robot grippers; it also…
4:00 AM ET
ARAxis Robotics
4:00 AM ETARAxis Robotics
@axisrobotics on X
Axis Weekly Last week, we closed the remaining replay and runtime gaps between the browser, policy server, and physics stack — then scaled the coverage of our task generation engine (TaskGen). An articulated asset library is now in the generation path, the full RoboCasa scene grid is online, and a cleaner long- versus short-horizon task split is ready for training and distillation. Key updates: • Replay & runtime: Policy and human-trajectory replay now match the frontend path, with web runtime reaching full replay fidelity. Scene loads faster, and simulation pauses automatically when idle. •…
12:00 PM ET
NVIDIA
$NVDA
12:00 PM ETNVIDIA
Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control | NVIDIA Technical Blog
Technical Blog Subscribe Related Resources Robotics Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control Aug 19, 2026 By Saeed Babamohamadi +12 Like Discuss (0) - L - T - F - R - E AI-Generated Summary - NVIDIA Jetson Thor can run the 4B Cosmos 3 Edge omni-model natively for on-device robot manipulation policies. - Post-training Cosmos 3 Edge on the Cosmos3-DROID dataset produces a policy that generates action chunks in about 1.53 seconds on Jetson AGX Thor T5000, enabling continuous real-time control. - In closed-loop RoboLab evaluation, the post-trained Edge policy reaches 22.9%…
4:01 PM ET
NVIDIA
$NVDA
4:01 PM ETNVIDIA
5 Physical AI infrastructure platforms shaping robotics in 2026 - The Robot Report
Facebook X LinkedIn Reddit Pinterest Share From accelerated computing and simulation to data operations, open-source tooling, validation engineering, and continuous learning, these five platforms represent distinct control points in the emerging physical AI stack. For most of the modern AI boom, infrastructure had a single center of gravity: compute. Models grew larger, training runs consumed more GPUs, and the industry organized itself around accelerators, cloud clusters, training frameworks, and developer software. That stack was sufficient when AI’s outputs were text, images, video, or…
12:00 PM ET
Google DeepMind
$GOOGL
12:00 PM ETGoogle DeepMind
Gemini Robotics 2 brings whole body intelligence to robots — Google DeepMind
Skip to main content July 30, 2026 Models Gemini Robotics 2 brings whole body intelligence to robots Carolina Parada Share Your browser does not support the video tag. From feet to fingertips — we are teaching robots intelligent whole-body control, fine dexterity, and teamwork to complete a broad range of complex tasks For decades, we’ve dreamed of robots that can seamlessly step into our world and lend a hand. Now, that vision takes a significant stride forward. Most robots are pre-programmed or teleoperated for narrow, repetitive task sequences. They lack the ability to truly learn for…
NVIDIA7Skild AI2ADLINK1Apple1Applied Intuition1articulated assets1Astribot1Axis Robotics1Boston Dynamics1development timeline of Moby.1development timeline reduction.1examples for adapting to new robotic embodiments.1Fudan University1Google DeepMind1Hugging Face1Lightwheel1LimX Dynamics1LM Studio1manipulation tasks1Moby deployment in a semiconductor facility.1Noble Machines1Ollama1Perplexity1post-training duration1SAGE-10K dataset1Scale AI1Scale AI data delivery1Schaeffler1Seeed1Solomon1Stretch1success rate of the post-trained policy.1tasks available in RoboLab.1The University of Hong Kong1Training Data1training hours for LDA-1B model1training needed for current VLA models to match S1's accuracy1TranscEngram1TRON 21validated run uses 64 nodes of 4× GB200 for 60K iterations, roughly 68 hours (17.4K GB200-hours).1Zhejiang University1ZiNovaLabs1
Who captures
NVIDIA7Skild AI2Stretch1Astribot1@smsehy1Perplexity1https://x.com/VaderResearch1ZiNovaLabs1The University of Hong Kong1https://x.com/RoboPapers1Axis Robotics1Google DeepMind1
Sourced numbers
manipulation tasks120 units“across 120 language-conditioned manipulation tasks”@NVIDIARobotics on X
The harness problem is getting its own framework category. Tardigrade ships today: an agent harness built as typed state machine components over an immutable event log. The framing is React for harness. Each component knows its previous states. The whole harness is a pure function of the log. I keep seeing teams reach this conclusion from two directions. One: they try to build a stateful agent loop and hit the problem of debugging mid-run failures, resuming interrupted sessions, and auditing what the agent actually did and when. Two: they want to add AI behavior to existing Effect/functional…
3:03 PM ET
ASAstribot
3:03 PM ETASAstribot
@XRoboHub on X
Robots can’t stop while the model thinks. In a high-speed throw, even one pause can kill the momentum. Astribot released SmoothRL for online RL during async inference. S1 keeps moving as the model computes the next action chunk. Most actions in a chunk never execute. Train on the full chunk, and RL credits or blames moves that never happened. SmoothRL learns only from executed actions, matching real deployment timing. After 250 rollouts, tossing jumped 39%→94%, pen capping 8%→83%, and box opening 30%→90%. One autonomous toss cut acceleration RMS by 52% and jerk RMS by 47%. The model keeps…
5:17 AM ET
@smsehy
5:17 AM ET@smsehy
@smsehy on X
Writing a custom robotics runtime in Rust eliminates garbage collection latency spikes on the factory floor, but maintaining sub-millimeter trajectory repeatability across varying ambient thermal conditions remains the core challenge in production welding. https://t.co/q21icpHIJu
7:34 PM ET
PEPerplexity
7:34 PM ETPEPerplexity
@stretchcloud on X
The hybrid compute pattern is getting serious. Perplexity just open-sourced Lily, the local inference engine powering their Mac hybrid compute feature, and the architecture choices are worth understanding.Lily runs Qwen3.6-35B-A3B on Apple silicon using a Rust runtime with custom Metal kernels. The optimization split is separate paths for prefill and decode. That matters because the bottleneck on local AI models shifts depending on whether you're processing a long input or generating output tokens. Lily tunes both separately to match what the hardware can actually do.The use case is cleaner…
3:03 PM ET
https://x.com/VaderResearch
3:03 PM EThttps://x.com/VaderResearch
mete on X: "Buying a quadruped? Consider this:
✸ Wheels vs legs change where it can go
✸ Payload on paper ≠ payload in motion
✸ Runtime changes with load and terrain
✸ Top speed rarely means useful speed" / X
Post Log in Sign up Post mete on X: "Buying a quadruped? Consider this: ✸ Wheels vs legs change where it can go ✸ Payload on paper ≠ payload in motion ✸ Runtime changes with load and terrain ✸ Top speed rarely means useful speed" - mete @VaderResearch Buying a quadruped? Consider this: ✸ Wheels vs legs change where it can go ✸ Payload on paper ≠ payload in motion ✸ Runtime changes with load and terrain ✸ Top speed rarely means useful speed View media 3:03 PM · Sep 2, 2026 Reply
12:00 PM ET
NVIDIA
$NVDA
12:00 PM ETNVIDIA
@NVIDIARobotics on X
🤖 A robot policy can be compact enough to run onboard and fast enough to keep the arm moving continuously. Post-trained NVIDIA Cosmos 3 Edge on NVIDIA Jetson AGX Thor T5000 generates roughly 2.13 seconds of robot motion in about 1.53 seconds, then replans from fresh camera and robot-state observations. Developers can test the same policy in closed-loop RoboLab simulation across 120 language-conditioned manipulation tasks before evaluating on hardware.
10:31 PM ET
ZIZiNovaLabs
10:31 PM ETZIZiNovaLabs
LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows.
@ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, includi… / X
Post Log in Sign up Post LimX Dynamics on X: "Together with ZINOVA's Tool Intelligence, TRON 2 takes on increasingly complex construction workflows. @ZiNovaLabs builds on TRON 2 to explore an innovative robotic configuration for construction, demonstrating key tasks in a scaled-down tilt-up construction workflow, including formwork assembly, multi-layer rebar placement and tying. TRON 2 serves as a modular and extensible embodied robotic platform for multi-tool, multi-step tasks across large workspaces. Its dual arms handle construction tools and materials across different orientations and…
7:00 PM ET
Nvidia
$NVDA
7:00 PM ETNvidia
Noble Machines Accelerates Humanoid Robot Development 3X | NVIDIA
Manufacturing \| Robotics Noble Machines Accelerates General Purpose Industrial Robot Development With NVIDIA Isaac GR00T Learn More Objective Noble Machines is advancing industrial robot development with the Moby general-purpose industrial robot and its supporting software stack, including a proprietary AI driven whole-body-control system. Moby is designed for complex, hazardous, and labor-intensive work across factories, logistics centers, construction sites, and semiconductor facilities, where operating reliably requires advanced perception, reasoning, adaptation, and physical interaction.…
2:09 PM ET
Nvidia
$NVDA
2:09 PM ETNvidia
Seeed reBot Arm B601-RS: Physical AI & VLA Model Course | NVIDIA DLI Series
Warehouse China Warehouse US Warehouse Germany Warehouse The store will not work correctly in the case when cookies are disabled. US Warehouse: Enjoy FREE UNIUNI shipping on orders under $50! (Excludes XIAO & Raspberry Pi series) \\ \\ the AI Hardware Partner For Industry Products Local Warehouse Documents Customization Solution Software Support Open Claw Quick Order Account Get more with a SeeedStudio accountclear \\ \\ 1 on 1 Support \\ \\ Local Delivery \\ \\ 30-Day DOA Guarantee \\ \\ 30-Day Returns Sign In Create Account My Orders\\ Track, change, cancel Account Information\\ Update…
5:00 PM ET
TUThe University of Hong Kong
5:00 PM ETTUThe University of Hong Kong
The Imitator Game — Benchmarking Robot Imitative Ability Beyond Action Prediction
Intent imitation, L0-L320,000+ paired episodesHuman evaluation Arena The Imitator Game BenchmarkingRobotImitativeAbilityBeyondActionPrediction Xunzhe Zhou 1,2,\,† · Yiyang Cai 2,3,\ · Fengyi Wang 2,3,\ · Ran Ju 1,2,\ · Hanxiang Ren 2,4 · Ruizhe Liu 1 Yu Zhang 1 · Qian Luo 1,2 · Feng Chen 1 · Pei Zhou 1,2 · Yi Ma 1,2 · Yanchao Yang 1,2,‡ 1The University of Hong Kong · 2TranscEngram · 3Fudan University · 4Zhejiang University \\ Equal contribution · † Project lead · ‡ Corresponding author Can a robot imitate what a human intends — not just what they do? We widen the gap between the demonstration…
4:05 PM ET
NVIDIA
$NVDA
4:05 PM ETNVIDIA
How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents | NVIDIA Technical Blog
Technical Blog Subscribe Related Resources Robotics English한국어中文 How to Train a Cross-Embodiment Robot Navigation Policy with AI Agents Aug 26, 2026 By Yan Chang, Mihir Acharya, Wei Liu, Katie Washabaugh and Aishwarya Singh +9 Like Discuss (0) - L - T - F - R - E AI-Generated Summary - COMPASS adapts the pretrained NVIDIA X-Mobility policy into residual specialists for specific robots and environments using reinforcement learning. - An agent-driven workflow with human approval gates automates environment validation, scene preparation, smoke testing, residual training, checkpoint evaluation,…
1:19 PM ET
Skild AI
1:19 PM ETSkild AI
Skild AI on X: "We found that to match the accuracy that S1 can achieve with just one example of prompting, current VLA models would need to be post-trained with 50-100 hours of data collection followed by fine-tuning!" / X
Post Log in Sign up Post Skild AI on X: "We found that to match the accuracy that S1 can achieve with just one example of prompting, current VLA models would need to be post-trained with 50-100 hours of data collection followed by fine-tuning!" - Skild AI @SkildAI Aug 25 Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 427 1K 7.1K 3.4M - Skild AI @SkildAI Aug 25 It can be taught…
1:19 PM ET
Skild AI
1:19 PM ETSkild AI
Skild AI on X: "Introducing S1, our new foundation model that learns from one example.
It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning.
Watch S1 operate in real-time via in-context learning:" / X
Post Log in Sign up Post Skild AI on X: "Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning:" - Skild AI @SkildAI Introducing S1, our new foundation model that learns from one example. It can be taught 10-minute long tasks that it has never seen before, from one video prompt without any fine-tuning. Watch S1 operate in real-time via in-context learning: 00:00 View media 1:19 PM · Aug 25, 20263.4MViews 427…
9:15 AM ET
https://x.com/RoboPapers
9:15 AM EThttps://x.com/RoboPapers
RoboPapers on X: "Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data.
Part of the secret is that LDA-1B jointly learns forward … / X
Post Log in Sign up Post RoboPapers on X: "Instead of choosing between training a world model and training a language conditioned robot policy, why not do both? LDA-1B s a new foundation model that is trained on 30,000 hours of human and robot interaction data. Part of the secret is that LDA-1B jointly learns forward dynamics, action prediction, and visual forecasting, all in a structured DINO latent space which avoids the pitfalls of redundant pixel-level prediction which isn’t necessarily aligned robot action. This approach works on both dexterous hands and simple robot grippers; it also…
4:00 AM ET
ARAxis Robotics
4:00 AM ETARAxis Robotics
@axisrobotics on X
Axis Weekly Last week, we closed the remaining replay and runtime gaps between the browser, policy server, and physics stack — then scaled the coverage of our task generation engine (TaskGen). An articulated asset library is now in the generation path, the full RoboCasa scene grid is online, and a cleaner long- versus short-horizon task split is ready for training and distillation. Key updates: • Replay & runtime: Policy and human-trajectory replay now match the frontend path, with web runtime reaching full replay fidelity. Scene loads faster, and simulation pauses automatically when idle. •…
12:00 PM ET
NVIDIA
$NVDA
12:00 PM ETNVIDIA
Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control | NVIDIA Technical Blog
Technical Blog Subscribe Related Resources Robotics Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control Aug 19, 2026 By Saeed Babamohamadi +12 Like Discuss (0) - L - T - F - R - E AI-Generated Summary - NVIDIA Jetson Thor can run the 4B Cosmos 3 Edge omni-model natively for on-device robot manipulation policies. - Post-training Cosmos 3 Edge on the Cosmos3-DROID dataset produces a policy that generates action chunks in about 1.53 seconds on Jetson AGX Thor T5000, enabling continuous real-time control. - In closed-loop RoboLab evaluation, the post-trained Edge policy reaches 22.9%…
4:01 PM ET
NVIDIA
$NVDA
4:01 PM ETNVIDIA
5 Physical AI infrastructure platforms shaping robotics in 2026 - The Robot Report
Facebook X LinkedIn Reddit Pinterest Share From accelerated computing and simulation to data operations, open-source tooling, validation engineering, and continuous learning, these five platforms represent distinct control points in the emerging physical AI stack. For most of the modern AI boom, infrastructure had a single center of gravity: compute. Models grew larger, training runs consumed more GPUs, and the industry organized itself around accelerators, cloud clusters, training frameworks, and developer software. That stack was sufficient when AI’s outputs were text, images, video, or…
12:00 PM ET
Google DeepMind
$GOOGL
12:00 PM ETGoogle DeepMind
Gemini Robotics 2 brings whole body intelligence to robots — Google DeepMind
Skip to main content July 30, 2026 Models Gemini Robotics 2 brings whole body intelligence to robots Carolina Parada Share Your browser does not support the video tag. From feet to fingertips — we are teaching robots intelligent whole-body control, fine dexterity, and teamwork to complete a broad range of complex tasks For decades, we’ve dreamed of robots that can seamlessly step into our world and lend a hand. Now, that vision takes a significant stride forward. Most robots are pre-programmed or teleoperated for narrow, repetitive task sequences. They lack the ability to truly learn for…