Physical Reality, Inc. (@physicreality) header
Physical Reality, Inc. (@physicreality) profile picture

Amplifying and extending human will.

USA
Joined May 2026
Photo by Physical Reality, Inc. (@physicreality)
A Third Industrial Revolution is coming. We need salesmen, engineers, and scientists. Talent is welcome. Will is enough. Join us. [email protected]
3
8
1
22
188,971
Robot durability, affordability, and usability constraints cannot be adequately solved through guessing. We'll soon release an update on the future of application-informed robot design. Stay tuned.
1
5
8
549
[1/2] “The Task Selects the Hand” Today we introduce a method for synthesizing families of gripper morphologies using reinforcement learning. Starting from a single first-person video of an everyday task, we extract the hand, object, and contact trace; reconstruct the scene into a physics-ready simulation environment; and compose and train a gripper for that task, rather than forcing the task onto an arbitrary gripper. Every candidate that succeeds in simulation is then subjected to static finite element analysis under the working load, and any hand that cannot carry it is vetoed. The method closes with a printable CAD file describing the resulting gripper. The four finger hand produced in this blog, PR-HAND-F4J2, did not exist before the video. Link below 🧵
85
[1/2] “The Task Selects the Hand” Today we introduce a method for synthesizing families of gripper morphologies using reinforcement learning. Starting from a single first-person video of an everyday task, we extract the hand, object, and contact trace; reconstruct the scene into a physics-ready simulation environment; and compose and train a gripper for that task, rather than forcing the task onto an arbitrary gripper. Every candidate that succeeds in simulation is then subjected to static finite element analysis under the working load, and any hand that cannot carry it is vetoed. The method closes with a printable CAD file describing the resulting gripper. The four finger hand produced in this blog, PR-HAND-F4J2, did not exist before the video. Link below 🧵
7
6
4
69
283,375
Physical Reality was founded on the belief that we are not beasts of burden; that our hands were made for more than toil; that they were made to build a world better than this one; that the advent of physical agents is the single greatest extension of human will the world has ever seen. We are teaching machines to work, and putting them to work everywhere work should be done.
2
6
506
Physical Reality, Inc. retweeted
Can an embodied reasoning model do a brake job? We ran 9 VLMs through an expert mechanic's exam on his own repair, with @physicreality. In theory they know what to do. Noticing and adapting on their own is where they fall short. This is where deployment would fail.
8
9
4
38
182,295
[1/2] In collaboration with @ContractedAI: "Knowing, but not Noticing" - We show that frontier ER models (which robot foundation models dominantly initialize from) consistently fail to know 'what to do next' in a chained task; - how to measure that failure quantitatively using data sourced from a novel branching/ 'counterfactual'-style ego data collection regime; and - propose a means to produce post-training data to fill in those gaps, based on what context ablations via harnessing reveal the model already knows, but simply fails to conjure reflexively on its own. Link in thread 🧵
3
8
3
36
91,772