Our-PhysicalAI-Space Transistor (https://transistor.fm) https://feeds.transistor.fm/our-physicalai-space Welcome to my channel ;) I'll be sharing what I'm learning about physical AI. The format will be an audio (podcasts generated by notebooklm) + 1-line or so summaries + sources Ubundi 5e2657c4-5d38-56f9-96c1-200cbedf93ae yes en Tue, 05 May 2026 21:48:40 +0200 Tue, 05 May 2026 21:49:08 +0200 https://img.transistorcdn.com/g5QRHnEvWTsgrKg1Z0mEZ78JWAWaYd9nh0JrssWoIb4/rs:fill:0:0:1/w:1400/h:1400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS80ZDRh/ZGI1MTU2NjM0YjBi/ZDFjZmYwNTc3NWJi/MWUyMS5wbmc.jpg Our-PhysicalAI-Space episodic Furaha Welcome to my channel ;) I'll be sharing what I'm learning about physical AI. The format will be an audio (podcasts generated by notebooklm) + 1-line or so summaries + sources Welcome to my channel ;) I'll be sharing what I'm learning about physical AI. technology, Robotics, Data, Infrastructure Furaha No No The Hidden Roadblock to PhysicalAI : Surgical Autonomy via Cross-Embodiment Data 5 5 The Hidden Roadblock to PhysicalAI : Surgical Autonomy via Cross-Embodiment Data full 84e75719-70a4-453b-93f2-33367169bd0a https://share.transistor.fm/s/fc4b588b This episode explores the "structural data bottleneck" in medical robotics and the landmark project designed to break it: Open-H-Embodiment. We discuss why the world’s most advanced general-purpose robot models, which can fold laundry and pick up toys, scored a 0% success rate on basic surgical suturing. We dive into the "Open X" playbook for surgery, the creation of a 770-hour dataset spanning 20 different robotic platforms, and the rise of GR00T-H—the first foundation model to achieve autonomous end-to-end surgical task completion

]]> This episode explores the "structural data bottleneck" in medical robotics and the landmark project designed to break it: Open-H-Embodiment. We discuss why the world’s most advanced general-purpose robot models, which can fold laundry and pick up toys, scored a 0% success rate on basic surgical suturing. We dive into the "Open X" playbook for surgery, the creation of a 770-hour dataset spanning 20 different robotic platforms, and the rise of GR00T-H—the first foundation model to achieve autonomous end-to-end surgical task completion

]]> Tue, 05 May 2026 21:35:19 +0200 Furaha Furaha 1333 This episode explores the "structural data bottleneck" in medical robotics and the landmark project designed to break it: Open-H-Embodiment. We discuss why the world’s most advanced general-purpose robot models, which can fold laundry and pick up toys, scored a 0% success rate on basic surgical suturing. We dive into the "Open X" playbook for surgery, the creation of a 770-hour dataset spanning 20 different robotic platforms, and the rise of GR00T-H—the first foundation model to achieve autonomous end-to-end surgical task completion

]]> technology, Robotics, Data, Infrastructure No Industrial Physical AI and the Data Factory 4 4 Industrial Physical AI and the Data Factory full 795e089d-2c78-4f6d-9aa5-698ce421d662 https://share.transistor.fm/s/2788042e This episode explores the massive transition in robotics from rigid, step-by-step programming to Physical AI which is a neural networks that "output movement instead of pixels". We discuss why the lack of a "scrappable internet" for robot trajectories has created a critical data bottleneck and how new systems like the UR AI Trainer and NVIDIA’s Data Factory Blueprint are solving it. By using imitation learning on production-grade hardware, the industry is finally bridging the gap between isolated lab experiments and reliable factory-floor deployment

]]> This episode explores the massive transition in robotics from rigid, step-by-step programming to Physical AI which is a neural networks that "output movement instead of pixels". We discuss why the lack of a "scrappable internet" for robot trajectories has created a critical data bottleneck and how new systems like the UR AI Trainer and NVIDIA’s Data Factory Blueprint are solving it. By using imitation learning on production-grade hardware, the industry is finally bridging the gap between isolated lab experiments and reliable factory-floor deployment

]]> Tue, 05 May 2026 21:17:02 +0200 Furaha Furaha 1376 This episode explores the massive transition in robotics from rigid, step-by-step programming to Physical AI which is a neural networks that "output movement instead of pixels". We discuss why the lack of a "scrappable internet" for robot trajectories has created a critical data bottleneck and how new systems like the UR AI Trainer and NVIDIA’s Data Factory Blueprint are solving it. By using imitation learning on production-grade hardware, the industry is finally bridging the gap between isolated lab experiments and reliable factory-floor deployment

]]> technology, Robotics, Data, Infrastructure No Building the Infrastructure for Surgical and General Autonomy 3 3 Building the Infrastructure for Surgical and General Autonomy full 1c905062-d521-4df8-ab2a-1f79ab1bcf01 https://share.transistor.fm/s/9b19686b In this episode we dive into the Open X-Embodiment project, which created a "GPT-1 moment" for robotics by pooling data from 22 different robotic platforms. We then look at Open-H-Embodiment, which is a much more recent project and currently the largest open dataset for medical robotics, which is enabling autonomous surgical tasks like suturing

]]> In this episode we dive into the Open X-Embodiment project, which created a "GPT-1 moment" for robotics by pooling data from 22 different robotic platforms. We then look at Open-H-Embodiment, which is a much more recent project and currently the largest open dataset for medical robotics, which is enabling autonomous surgical tasks like suturing

]]> Tue, 05 May 2026 10:04:31 +0200 Furaha Furaha 1284 In this episode we dive into the Open X-Embodiment project, which created a "GPT-1 moment" for robotics by pooling data from 22 different robotic platforms. We then look at Open-H-Embodiment, which is a much more recent project and currently the largest open dataset for medical robotics, which is enabling autonomous surgical tasks like suturing

]]> technology, Robotics, Data, Infrastructure No Scaling Physical AI through Superhuman Sensing 2 2 Scaling Physical AI through Superhuman Sensing full cbec6f48-09aa-4ce7-bac5-795a692348c4 https://share.transistor.fm/s/c23bec34 In this episode we dive Meta’s Digit 360, an artificial fingertip that provides robots with superhuman touch, digitizing modalities like force, heat, and even scent to bridge the gap between digital intelligence and physical contact.

]]> In this episode we dive Meta’s Digit 360, an artificial fingertip that provides robots with superhuman touch, digitizing modalities like force, heat, and even scent to bridge the gap between digital intelligence and physical contact.

]]> Tue, 05 May 2026 09:15:53 +0200 Furaha Furaha 1278 In this episode we dive Meta’s Digit 360, an artificial fingertip that provides robots with superhuman touch, digitizing modalities like force, heat, and even scent to bridge the gap between digital intelligence and physical contact.

]]> technology, Robotics, Data, Infrastructure No What First Motive is actually betting on ? 1 1 What First Motive is actually betting on ? full d02f286c-5595-4043-939b-069295361a4a https://share.transistor.fm/s/23921377 What is the most defensible barrier to entry in the Physical AI race as model architectures become commoditised?

One-line answer: Access to verified, real-world "ground truth" through a structural supply of 1.7 million domain experts that competitors cannot scrape from the internet or perfectly replicate in simulation
Sources:

  • Physical AI Pre-Read v3 
  • Healthcare POC Plan v1.0
]]> What is the most defensible barrier to entry in the Physical AI race as model architectures become commoditised?

One-line answer: Access to verified, real-world "ground truth" through a structural supply of 1.7 million domain experts that competitors cannot scrape from the internet or perfectly replicate in simulation
Sources:

  • Physical AI Pre-Read v3 
  • Healthcare POC Plan v1.0
]]> Thu, 30 Apr 2026 15:05:00 +0200 Furaha Furaha 1327 What is the most defensible barrier to entry in the Physical AI race as model architectures become commoditised?

One-line answer: Access to verified, real-world "ground truth" through a structural supply of 1.7 million domain experts that competitors cannot scrape from the internet or perfectly replicate in simulation
Sources:

  • Physical AI Pre-Read v3 
  • Healthcare POC Plan v1.0
]]> technology No