Humanoid-universe.com

<t>@timothy.roberts2 One nitpick -

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. SLAM in a working ...

This lines up with my experience. 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.

<t>Just to be precise about one thing:

Balance-recovery controllers are usually evaluated with push-recovery tests (a known, repeatable lateral push) in demos, but real-world robustness also depends on recovering from unstructured events like uneven flooring, unexpected contact, or a dropped ...

<t>I dealt with almost this exact situation.

Vision-based grasp confidence estimation (predicting success before attempting a grasp) and tactile-based confirmation (confirming after contact) are complementary rather than competing - vision helps you choose a grasp, tactile tells you if it ...

<t>@diego.moore6 Side note that might be relevant:

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 ...

<t>Small correction on one detail:

In-hand reorientation (repositioning a grasped object without setting it down) is one of the more advanced manipulation skills, requiring either a highly dexterous hand with enough DoF or clever use of gravity and controlled slipping - it's an active research ...

<t>This matches what I've seen too.

A first successful unassisted walk, from a standing start on a from-scratch project, commonly takes many months of part-time weekend work even for someone with a solid mechanical and software background - it's a genuinely hard milestone, not a weekend project ...

<t>Can I ask a dumb follow-up -

Self-reported deployment and uptime numbers from humanoid companies are inherently hard to independently verify, since there's no standardized, third-party reporting requirement yet - a healthy dose of skepticism toward company press releases is reasonable until ...

<t>@dorothy_kuma I'd frame this differently.

How much ML background a controls role actually requires varies a lot by company and team - some whole-body control positions remain heavily classical-optimization-focused, while others increasingly expect familiarity with reinforcement learning or ...

<t>This matches what I've seen too.

Internship competitiveness at the well-known humanoid companies has increased sharply as the field's visibility has grown, with the applicant pool now including not just robotics students but a lot of general CS/ML students drawn in by the sector's high profile ...