Great summary! I also really like @DrJimFan
’s ideas about digital cousins. The challenge is the sim2real gap: it is surmountable for flying and locomotion but still an open problem for manipulation due in part to the nuances of dynamics, friction, and deformation.
@ken_goldberg
-
Sim2Real Gap Challenges in Robotic Manipulation and Digital Cousins
By
–
-
Generalist VLA Models and Specialist Models for Data Collection
By
–
I agree. Generalist VLA models can be much more versatile—given enough data. I believe specialist models can be useful for collecting this data.
-
Robot Hand Capable of Taking Human Pulse: Details Needed
By
–
I like simple grippers but hats off to the first robot hand that can take a human pulse….please share details on precision, weight and cost. https://t.co/H5JAOGl2sX
— Ken Goldberg (@Ken_Goldberg) 15 septembre 2025I like simple grippers but hats off to the first robot hand that can take a human pulse….please share details on precision, weight and cost.
-
Pi0.5 bin picking experiment and robotic picking throughput advances
By
–
Thank you Junyao for running this experiment so quickly! It’s helpful to see how Pi0.5 picks from bins. Dex-Net used a CNN inside a kinematic control loop; newer methods @AmbiRobotics operate at 500 PPH. https://t.co/FF1W2IwPX5
— Ken Goldberg (@Ken_Goldberg) 14 septembre 2025Thank you Junyao for running this experiment so quickly! It’s helpful to see how Pi0.5 picks from bins. Dex-Net used a CNN inside a kinematic control loop; newer methods @AmbiRobotics operate at 500 PPH.
-
Comparing robotic picking performance and Dex-Net evolution
By
–
Thanks for sharing this! It’s worth watching all the way through. Lots of retries, so roughly 100 picks per hour. Does it also pick objects out of the bin? Interesting to compare with Dex-Net from 2018:https://t.co/xp8yDCDaN1 https://t.co/ebpB6LShHR
— Ken Goldberg (@Ken_Goldberg) 11 septembre 2025Thanks for sharing this! It’s worth watching all the way through. Lots of retries, so roughly 100 picks per hour. Does it also pick objects out of the bin? Interesting to compare with Dex-Net from 2018: https://
youtu.be/pcp6kroGOVQ -
Large Models Opacity: We Become Audience Instead of Magician
By
–
Important insight about the opacity of large models: "…the issue with wizards: We're getting something magical, but we're also becoming the audience rather than the magician."
-
Robotics Progress Demands Realistic Expectations and Promises
By
–
So many exciting developments in robotics lately! But let's not overpromise:
-
Physics counting argument for high performance robotics
By
–
Thx for your comments. Re: “Conceptual robotics can largely be solved using data, but high performance robotics cannot….There are reasons and theory behind this insight, that I will return to in some other text.” I’m curious: is there a counting argument from physics?
-
Humanoid Robots Arrive Soon But Adoption Remains Distant
By
–
Humanoid robots are coming, but don't hold your breath waiting for them:
-
Engineering Solutions to Bridge Robotics Data Gap
By
–
Good Old-Fashioned Engineering Can Close the 100,000-Year “Data Gap” in Robotics. Science Robotics Editorial: https://
science.org/doi/10.1126/sc
irobotics.aea7390
…