Humanoid-universe.com

<t>Small correction on one detail:

OpenVLA is a notable open-source VLA model - roughly 7 billion parameters, trained on hundreds of thousands of real-world robot demonstrations - and has been shown to outperform much larger closed models on some manipulation benchmarks, which says a lot about ...

<t>Appreciate the detailed answer.

A lot of the manipulation shown in production demos still leans heavily on teleoperation, particularly for anything involving fine force control or novel objects - autonomous grasp success rates on genuinely unstructured, previously-unseen clutter are still well ...

<t>@kwameivanov Agreed, and I'd add:

Imitation learning from human demonstration video (without robot teleoperation data) is an appealing way to scale up training data cheaply, but it runs into the embodiment gap - human hand kinematics and force profiles don't map directly onto a robot hand's ...

<t>I'll believe the stronger version of that claim when it's independently verified.

XPeng's IRON humanoid debuted with a solid-state battery, part of a broader trend of Chinese entrants (alongside GAC's GoMate and EngineAI's T800) pushing energy-density and runtime claims that Western ...

<t>@rtorres Appreciate the detailed answer.

Starting in simulation (even a simplified physics sim of just the leg or the target gait) before touching real hardware is one of the more reliable ways to avoid destroying expensive parts while debugging balance and walking control from scratch ...

<t>@emilyperez Appreciate the detailed answer.

3D-printed gears wearing out is one of the most common early hobbyist failure points - infill percentage, layer orientation relative to load direction, and filament choice (nylon or reinforced composites tend to outlast standard PLA/PETG under ...

<t>@servoken70 Appreciate the detailed answer.

A lot of hobbyists underestimate how much of the real difficulty in a DIY humanoid build is software and tuning rather than mechanical assembly - getting a robot to physically stand up is a much smaller fraction of the total project time than getting ...

<t>@matthew.watanabe This lines up with my experience.

Automotive OEMs becoming the leading early-adopter vertical isn't really about cars specifically - it's about those companies already having structured facilities, existing automation budgets, and risk tolerance for piloting new equipment at ...

<t>@jwang Worth being a little skeptical of the marketing angle here.

Estimating joint torque from motor current draw is cheap and requires no extra sensor, but it's less accurate than a dedicated torque sensor because it doesn't capture friction losses through the gearbox - good enough for ...

<t>@greta.carter Ran into exactly this myself.

3D-printed gears wearing out is one of the most common early hobbyist failure points - infill percentage, layer orientation relative to load direction, and filament choice (nylon or reinforced composites tend to outlast standard PLA/PETG under ...