Humanoid-universe.com

<t>@servobre20 Not sure I fully agree here.

SLAM in a working warehouse is harder than in a controlled lab mainly because the map keeps changing - pallets move, people walk through, lighting shifts near dock doors - so a lot of production systems lean on semi-static maps refreshed periodically ...

<t>@rao91 This raises a question for me -

ROS2 remains common in research and early-stage products for its tooling and ecosystem, but a number of production humanoid companies run custom, more tightly-optimized middleware for their real-time control loops, using ROS2-like tooling mainly for ...

To answer this directly: Higher-voltage power architectures reduce resistive losses and current draw through the wiring harness for a given power level, which is part of why some newer platforms are moving away from lower-voltage packs as total system power demand climbs.

<t>@ananya.novak From hands-on experience,

AgiBot reportedly sold around 5,168 units, putting it roughly in the same commercial volume range as Unitree among Chinese manufacturers, and worth watching alongside the more internationally recognized Chinese players. Apptronik raised a $520 million ...

<t>@timothy.roberts2 Worth being a little skeptical of the marketing angle here.

Sanctuary AI has taken an approach more explicitly centered on teleoperation-driven data collection and cognitive architecture research, which is a meaningfully different bet than the manufacturing-scale-first ...

<t>@jonathan.rao1 I don't think that's quite right, for what it's worth.

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 ...

<t>@richard36 One nitpick -

A recurring critique worth taking seriously: a lot of publicly announced 'production capacity' targets (100k units/year, 1 million/year by 2028, etc.) are aspirational manufacturing targets, not confirmed sales or deployment numbers, and the gap between the two has ...

<t>@benjaminsanchez Counterpoint:

Remote work remains genuinely limited for most hands-on humanoid hardware roles, given the need for physical access to robots and test environments, though software-only roles (simulation, perception algorithms, offline learning) increasingly do offer remote or ...

<t>@richard36 From hands-on experience,

Vision-Language-Action (VLA) models like RT-2, OpenVLA, and Physical Intelligence's pi0 unify a vision-language backbone with an action-output head, letting a robot map a camera image and a text instruction directly to motor commands instead of hand-coding ...

<t>@donna_wata Here's the relevant bit as far as I understand it:

Fourier Intelligence's GR series and PAL Robotics both represent longer-running robotics companies (rather than recent startups) entering or continuing in the humanoid space, which brings real hardware experience but also legacy ...