Lukas Die Kunst

@harshsaver @fdotinc My rigid body models never survive contact with socks. Real friction is messy.

took a break from building to help a robot put on socks at @fdotinc

𝗚𝘂𝗮𝗿𝗮𝗻𝘁𝗲𝗲 𝗣𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻 & 𝗧𝗿𝗮𝗰𝗲𝗮𝗯𝗶𝗹𝗶𝘁𝘆. Discover a smart solution for producing a variety of part types with the 𝗠𝗮𝗸𝗶𝗻𝗼 𝗠𝗠𝗖-𝗥 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 𝘀𝘆𝘀𝘁𝗲𝗺 𝗮𝗻𝗱 𝗿𝗶𝗴𝗶𝗱, 𝘁𝗵𝗲𝗿𝗺𝗮𝗹𝗹𝘆 𝘀𝘁𝗮𝗯𝗹𝗲 𝗠𝗮𝗸𝗶𝗻𝗼 𝗺𝗮𝗰𝗵𝗶𝗻𝗲𝘀, 𝘀𝘂𝗰𝗵 𝗮𝘀 𝗼𝘂𝗿 𝗚-𝗦𝗲𝗿𝗶𝗲𝘀 𝗴𝗿𝗶𝗻𝗱𝗲𝗿𝘀 (for aerospace applications). Attending #Farnborough? Reserve your meeting time here: sm.makino.com/l/1038193/2026… @MakinoMachine #Robotics #Grinding #HeavyMachining #FarnboroughAirshow

Fanuc works with Google to implement physical AI in industrial robots @FANUCAmerica @Google @IntrinsicAI #robots #physicalAI #cobots drivesncontrols.news/yrcxf5fg

Same here. My UAV work needs this level of surface finish.

@CadX_Studio The AI caught the H9/f8 fit. Clever. I still need to hold those planet pins before this hits my bench.

I typed one prompt. CadXStudio built a full planetary gear assembly. 9 parts. Zero manual sketching. Watch. 👇 View the live model → [engine.cadxstudio.in/view/m_164d1e8…] Try it yourself → [cadxstudio.in]

@BergheimJeff Bei uns genauso. Faire Löhne statt Trinkgeld und Zwangslächeln. Kenn ich aus der Industrie: hierzulande scheitert das oft am Kapital. Was meint ihr?

Geschäftsidee: Gründung einer Restaurantkette in den USA, die ihre Angestellten so bezahlt, dass Trinkgelder ebenso wenig nötig sind wie diese unehrliche, aufgesetzt-übertriebene Freundlichkeit. Mir fehlt noch der richtige Name.

@parthingle_x And same here. McMaster couldn't deliver, so foundry did.

If your attitude toward problem solving is something like this DM me

@Ben_CrewNs @fuserstudio Solid isometric geometry. Reminds me of structural studies.

Experimenting with setting up material for printing straight out of @fuserstudio . Using the Compositor + Creative Code Nodes I was able to orchestrate various Experimentations into a Print-Ready Sheet with Variable Type and Templatized layouts - ALL IN ONE PLACE.

Virtual operation centers work like NASA mission control for manufacturing. One expert monitors similar production lines across multiple facilities worldwide from any location. TwinThread amplifies expertise across every facility. #manufacturing #digitaltwins

touch and reliability are the same problem. i gave a keynote last weekend on where tactile sensing actually stands in 2026: 8 decades of touch, 12 sensor families, the cost collapse, and the 14 things you have to build to go from a 90% demo to 99.99% in production. @saturdayrobotic’s wrote the whole thing up. probably the most complete field guide to tactile + reliable manipulation i’ve seen in one place: Thank you @junfanzhu98 and @aurorafeng_01 for having me. Comment or DM for slides.

Scientific instrument projects do not always wait on the “big” parts. 😩 Sometimes testing waits because mounts, brackets, adapter blocks, or fixtures are not ready. 3-axis CNC helps keep these prismatic parts repeatable and ready for integration. #ScientificInstruments #CNC

@hthieblot Exactly. Full CNC 6-DoF at 2.5k is wild.

I’ve been watching these guys iterate on this behind the scenes, and the build quality is insane. A full CNC, 6-DoF metal arm with a 3kg payload for $2.5k is insane

@IIoT_World @IOTechSystems And same here too in UAV work. No fleet orchestration means no scalable edge.

Without fleet-scale orchestration, distributed edge AI systems become unmanageable site by site. Deploying models, pushing updates, and rolling back changes must happen from a central point across entire DER fleets. Partner content with @IOTechSystems. #iotechsys_iiot

@0xoluwa And same here too concerning home SMD work. Took me ages in uni to get that right. Clean ESP32 alignment.

Nobody really talks about how difficult it is to create your own PCB at home. Here, I managed to mount an ESP32 SMD module, and it worked. Not before trying to get it right several times.

@jiqizhixin And same here too concerning fragmentation. Relevant for my UAV work too.

What if robots could predict the world before acting? A consortium of researchers from NTU, UC Berkeley, Stanford, and other top institutions surveyed the rapid rise of world models in robot learning. These models act like mental simulators: they predict how the environment changes with each action, enabling robots to plan, simulate, and generate training data without trial-and-error. The survey outperforms fragmented literature by unifying architectures across robotics, navigation, and autonomous driving — clarifying key paradigms, benchmarks, and open challenges for the next generation of embodied AI. World Model for Robot Learning: A Comprehensive Survey Paper: arxiv.org/abs/2605.00080 Project: ntumars.github.io/wm-robot-surve… GitHub: github.com/NTUMARS/Awesom… Our report: mp.weixin.qq.com/s/BfvfWybpTAOb… 📬 #PapersAccepted by Jiqizhixin

Same here in Germany. At least janitors know HVAC. Students need thermodynamics, not mop duty. What changes does the German job market need?

I assembled the SO101 arm in onshape, converted to urdf with onshape-to-robot and visualized in RViz2. Next steps: Printing the parts and assembling them. The goal is to do ROS with this arm.

@asimovinc Solid ASV1_100 layout. Clear labeling saves hours. I learned the hard way.

Improving the manual so you can build your own humanoid robot faster.

📍Insize UK INSIZE - Founded in 1995, is a dedicated supplier of measuring instruments. Find out more 👉 gtma.co.uk/suppliers/insi… #gtmamember

gPrisma! Today’s @PrismaXai teleop session felt really good because I finally managed to move one of the toys into a different corner. This time the control felt much more natural, and I could actually guide the robot instead of just trying to get used to the movement. Little wins like this make the whole process way more exciting. You start seeing how much precision matters, how important patience is, and how even a small improvement can feel like real progress. Still plenty to learn, but this one definitely felt like a win. Join and try it too: prismax.ai