dronlemon retweeted
“To get into robotics, you no longer need a Ph.D.” - Spencer Huang
Great read from @IEEESpectrum on how open‑source AI, NVIDIA’s robotics stack, and @HuggingFace’s LeRobot are opening the field to a much wider community.
📖 nvda.ws/4dZs6aN
Great read from @IEEESpectrum on how open‑source AI, NVIDIA’s robotics stack, and @HuggingFace’s LeRobot are opening the field to a much wider community.
📖 nvda.ws/4dZs6aN
13
54
294
13,959
dronlemon retweeted
Conference Room Mess Cleanup Test: Unitree WVLA 2.0 Model🎉
This video was recorded in a single take.
Multi-task, fully autonomous operation. Strong external interference present.
This video was recorded in a single take.
Multi-task, fully autonomous operation. Strong external interference present.
112
171
1,859
7,645,716
dronlemon retweeted
Unitree Unveils: GD01, A Manned Transformable Mecha, from $650,000 👏
The world's first production-ready manned mecha. It can transform. It's a civilian vehicle. It weighs ~500kg with you inside.
Please everyone be sure to use the robot in a Friendly and Safe manner.
The world's first production-ready manned mecha. It can transform. It's a civilian vehicle. It weighs ~500kg with you inside.
Please everyone be sure to use the robot in a Friendly and Safe manner.
1,111
3,149
16,929
9,155,904
dronlemon retweeted
Physical AI is taking humanoid robots beyond scripts and into the real world. 🌍
Join Spencer Huang at #NVIDIAGTC Taipei for a technical look at end-to-end workflows for building general-purpose humanoid robots, from data collection all the way to deployment on hardware.
6 | 10:30 AM CST
📍TICC
Add to calendar: nvda.ws/4fdrAqG
Join Spencer Huang at #NVIDIAGTC Taipei for a technical look at end-to-end workflows for building general-purpose humanoid robots, from data collection all the way to deployment on hardware.
6 | 10:30 AM CST
📍TICC
Add to calendar: nvda.ws/4fdrAqG
7
23
147
12,524
dronlemon retweeted
We have always believed that #AGIBOT is not a robotics company in the traditional sense, but an embodied intelligence foundation model company.
In the era of embodied intelligence, #AI models should not remain confined to data, text, or screens. They need a physical carrier that can enter the real world, perceive its surroundings, and execute tasks. This is why AGIBOT has chosen to define robotic embodiment through AI, building robotics hardware, perception and decision-making systems, and real-world deployment capabilities around its core foundation model capabilities.
What we are building is not simply a robot that can move or operate, but an embodied intelligence system that can continuously learn, understand, and perform tasks in the real world. From AI models and system architecture to robotic hardware and scenario-based deployment, our goal is to enable AI to truly operate in physical environments and create real-world value.
Over the past year, we have advanced systematic deployment across multiple real-world scenarios, covering industrial operations, service collaboration, and multi-robot coordination. Through this process, one shift has become increasingly clear: real value does not lie in isolated capability breakthroughs, but in whether an AI system can operate reliably and sustainably in real environments, while continuing to evolve through ongoing data feedback.
We chose #HongKong because it connects capital markets, international industry, and innovation ecosystems. On May 12, we will share more about our judgment and practice on the next phase of AI and embodied intelligence in Hong Kong. The real inflection point is now moving from “model capability” to “sustainable real-world operation.”
In the era of embodied intelligence, #AI models should not remain confined to data, text, or screens. They need a physical carrier that can enter the real world, perceive its surroundings, and execute tasks. This is why AGIBOT has chosen to define robotic embodiment through AI, building robotics hardware, perception and decision-making systems, and real-world deployment capabilities around its core foundation model capabilities.
What we are building is not simply a robot that can move or operate, but an embodied intelligence system that can continuously learn, understand, and perform tasks in the real world. From AI models and system architecture to robotic hardware and scenario-based deployment, our goal is to enable AI to truly operate in physical environments and create real-world value.
Over the past year, we have advanced systematic deployment across multiple real-world scenarios, covering industrial operations, service collaboration, and multi-robot coordination. Through this process, one shift has become increasingly clear: real value does not lie in isolated capability breakthroughs, but in whether an AI system can operate reliably and sustainably in real environments, while continuing to evolve through ongoing data feedback.
We chose #HongKong because it connects capital markets, international industry, and innovation ecosystems. On May 12, we will share more about our judgment and practice on the next phase of AI and embodied intelligence in Hong Kong. The real inflection point is now moving from “model capability” to “sustainable real-world operation.”
AI is shifting from a model capability race to real-world validation.
For years, progress has been measured by parameters, benchmarks, and reasoning performance. These metrics remain important, but they no longer define the full picture. As models converge, a new threshold is n operate reliably in the real world.
Real-world capability is not about generating responses or passing tests. It is about performing consistently in complex, unstructured, and dynamic environments—across industrial, logistics, and service scenarios. The question is shifting from “how intelligent it looks” to “how reliably it works.”
This shift is redefining embodied AI and robotics. Models are becoming components within larger systems. Data becomes a continuous feedback loop. And deployment is now as critical as model performance.
The real divide is no longer benchmark performance, but real-world stability and scalability.
This is the core question we will explore at the first Hong Kong Embodied AI Industry Summit and AGIBOT Partner Conference (APC 2026) in Hong Kong on May 12.
#APC2026HK #AGIBOT #EmbodiedAI #Robotics #Innovation #HongKong
1
2
12
2,004
dronlemon retweeted
Ecosystem collaboration is a critical step forward for robotics.
Delighted to partner with @openmind_agi.
Delighted to partner with @openmind_agi.
OpenMind App Store Partner: @boosterobotics
Booster Robotics is a humanoid company providing developers with stable and user-friendly humanoid robots and efficient development tools.
Booster's "KidSize" humanoids have managed to stand out in the highly competitive humanoid its outstanding characteristics of being lightweight, flexible, durable, and resistant to falls.
10
12
95
16,063
dronlemon retweeted
We're launching the agentic robotics app store today. Let's democratize AI robotics for all!
300+ apps shipped. 10,000 robots in the wild. It used to take weeks from a robotics engineer to build apps, now everyone can do it in hours with ML intern or your favorite neighborhood achy mini app was built by Joel, a 78yo marketing exec who'd never coded in his life. Personally, I built an office receptionist in two hours last week.
More info to start building here: huggingface.co/blog/clem/reac…
300+ apps shipped. 10,000 robots in the wild. It used to take weeks from a robotics engineer to build apps, now everyone can do it in hours with ML intern or your favorite neighborhood achy mini app was built by Joel, a 78yo marketing exec who'd never coded in his life. Personally, I built an office receptionist in two hours last week.
More info to start building here: huggingface.co/blog/clem/reac…
46
94
684
126,263
dronlemon retweeted
.@huggingface's agentic robotics app store for Reachy Mini is a big step toward more accessible physical AI. 🙌
Excited to see NVIDIA Isaac GR00T N integrated with Hugging Face LeRobot, helping developers post-train, evaluate, and deploy robot foundation models with open robotics workflows.
Our latest blog 📖 huggingface.co/blog/nvidia/nv…
Github 💻 huggingface.co/nvidia/GR00T-N…
Excited to see NVIDIA Isaac GR00T N integrated with Hugging Face LeRobot, helping developers post-train, evaluate, and deploy robot foundation models with open robotics workflows.
Our latest blog 📖 huggingface.co/blog/nvidia/nv…
Github 💻 huggingface.co/nvidia/GR00T-N…
We're launching the agentic robotics app store today. Let's democratize AI robotics for all!
300+ apps shipped. 10,000 robots in the wild. It used to take weeks from a robotics engineer to build apps, now everyone can do it in hours with ML intern or your favorite neighborhood achy mini app was built by Joel, a 78yo marketing exec who'd never coded in his life. Personally, I built an office receptionist in two hours last week.
More info to start building here: huggingface.co/blog/clem/reac…
11
29
262
53,434
dronlemon retweeted
Physical AI Days are back at #NVIDIAGTC, this time in Taipei. ✈️
Check out our curated set of sessions on all things physical AI - from robots and industrial AI to digital twins and simulation - built to help you turn ideas into real-world deployments.
📅 June 3 (Wednesday) – 📍 TICC (Taipei International Convention Center)
Add it to your schedule: nvda.ws/49excgB
Can’t make it in person? Sessions will be available on demand.
Check out our curated set of sessions on all things physical AI - from robots and industrial AI to digital twins and simulation - built to help you turn ideas into real-world deployments.
📅 June 3 (Wednesday) – 📍 TICC (Taipei International Convention Center)
Add it to your schedule: nvda.ws/49excgB
Can’t make it in person? Sessions will be available on demand.
14
70
397
40,931
dronlemon retweeted
UniStore — Unitree Official Shared Application Platform Now Fully Open to All👏
The world's first humanoid robot task and action app store.
A new era from robots of functions to robots of intelligence has begun.
Develop and download new robot applications as easily as mobile possibilities for your robot.
International version coming soon!
The world's first humanoid robot task and action app store.
A new era from robots of functions to robots of intelligence has begun.
Develop and download new robot applications as easily as mobile possibilities for your robot.
International version coming soon!
25
81
432
70,467
dronlemon retweeted
Balancing commercial goals and robotics research can be tricky, but with Atlas we're making it work.
331
1,242
7,461
530,142
dronlemon retweeted
Unitree Launch | Dual‑Arm Humanoid Robot, from $4290🤩
Powered by high‑performance motors and sensors, with integrated binocular vision and voice interaction. Lightweight, compact, and ready for smarter task development.
Powered by high‑performance motors and sensors, with integrated binocular vision and voice interaction. Lightweight, compact, and ready for smarter task development.
73
184
1,463
11,997,160
dronlemon retweeted
The bottleneck for most people who want to build a robot isn't motivation
it's not knowing where to start
what parts. what order. what software stack.
Tnkr solves exactly that — open source robot projects with step-by-step assembly, CAD, firmware, everything
U
it's not knowing where to start
what parts. what order. what software stack.
Tnkr solves exactly that — open source robot projects with step-by-step assembly, CAD, firmware, everything
U
34
283
2,014
158,142
Princeton just made its Intro to Robotics course public.
Free access to:
→ lecture videos
→ notes
→ slides
→ assignments
You’ll learn the core foundations behind real robotic systems:
→ feedback control
→ motion planning
→ state estimation
→ localization + mapping
→ computer vision
→ deep learning
→ reinforcement learning
→ robotics law, ethics, and economics
The coolest part: the final project has students program drones for vision-based navigation using real-time camera feeds.
Basically, Princeton dropped a free robotics roadmap.
Perfect timing if you’re building with robots or AI agents.
Link 👇
Free access to:
→ lecture videos
→ notes
→ slides
→ assignments
You’ll learn the core foundations behind real robotic systems:
→ feedback control
→ motion planning
→ state estimation
→ localization + mapping
→ computer vision
→ deep learning
→ reinforcement learning
→ robotics law, ethics, and economics
The coolest part: the final project has students program drones for vision-based navigation using real-time camera feeds.
Basically, Princeton dropped a free robotics roadmap.
Perfect timing if you’re building with robots or AI agents.
Link 👇
1
1
7
365
dronlemon retweeted
Training humanoid robots?
You need motion data. Real, high-fidelity, human motion data. And until now - there was no open dataset purpose-built for humanoid robotics.
For 5 years, we've been building the largest enterprise-grade human motion and behavior datasets for embodied eakthrough SONIC research.
Today, at GTC, with @NVIDIARobotics, we're opening a piece of it to the world.
BONES-SEED:
→ 142,200 motion capture animations
→ Up to 6 natural language descriptions per motion
→ Temporal segmentation of every action
→ Curated for humanoid robotics
→ In NVIDIA SOMA and Unitree G1 (MuJoCo) formats
From text to action. Now yours.
Go build → bones.studio/datasets/seed
#NVIDIAGTC
You need motion data. Real, high-fidelity, human motion data. And until now - there was no open dataset purpose-built for humanoid robotics.
For 5 years, we've been building the largest enterprise-grade human motion and behavior datasets for embodied eakthrough SONIC research.
Today, at GTC, with @NVIDIARobotics, we're opening a piece of it to the world.
BONES-SEED:
→ 142,200 motion capture animations
→ Up to 6 natural language descriptions per motion
→ Temporal segmentation of every action
→ Curated for humanoid robotics
→ In NVIDIA SOMA and Unitree G1 (MuJoCo) formats
From text to action. Now yours.
Go build → bones.studio/datasets/seed
#NVIDIAGTC
5
44
190
50,238
