Humanoid-universe.com

<t>This is exactly the kind of context I was looking for.

Tesla reportedly deployed more than 1,000 Optimus Gen 3 units into its own Fremont, Austin, and Berlin factories starting around January 2026, with a stated capacity target near 100,000 units for the year - it's worth noting these are ...

<t>@ethan17 This is exactly the kind of context I was looking for.

UC Berkeley's Berkeley Humanoid Lite is a genuinely open-source, roughly $5,000, 3D-printable humanoid - structural parts print on a standard desktop printer (around a 200x200x200mm build volume), with CAD files, parts lists ...

<t>@jchen +1 to this. Worth adding:

Building a scoped-down project first - a single leg, an arm, or a torso-only upper body - is a common and sensible way to learn the mechanical and control fundamentals before committing to the cost and complexity of a full bipedal platform. 3D-printed gears ...

<t>@chloe.harris7 This is exactly the kind of context I was looking for.

Off-the-shelf RC servos are a reasonable starting point for a first small biped because they bundle motor, gearbox, and basic position control in one part, but they generally can't deliver the torque density or ...

@timothy.roberts2 Minor factual note: Boston Dynamics' commercial electric Atlas switched fully from hydraulic to QDD-style electric actuation, ending up quieter and lighter with comparable or better dynamic performance than the old hydraulic platform.

Ran into exactly this myself. Series elastic actuators add a spring element in series with the drivetrain, which gives cheap, precise torque sensing (measure spring deflection) and passive shock absorption, at the cost of reduced control bandwidth and added complexity.

<t>@benjaminsanchez I'd push back on this a bit.

Torque sensing is frequently integrated directly into the actuator housing via strain gauges near the output flange or spring element, rather than bolted on as an external sensor, to save weight and reduce noise from mechanical slop ...

Agreed, and I'd add: Multi-camera calibration drifts over time from thermal expansion, vibration, and mechanical wear, which is why production systems typically run periodic recalibration routines rather than assuming a one-time factory calibration holds forever.

@ssantos Minor factual note: Force/torque sensors near the ankle give a direct read on ground reaction forces, which is valuable for balance control, but they add cost, a failure point, and routing complexity right at a joint that already takes the most mechanical abuse.

<t>Small correction on one detail:

Model predictive control (MPC) is still very much alive in production humanoids, often working alongside or underneath learned policies - MPC handles short-horizon dynamically-consistent trajectory optimization while learned components handle perception, task ...