← Robotics Academy

Mechanical & system design

Designing a humanoid-class system is not “more servos” — it is load paths, vibration, thermal headroom, wiring serviceability, and failure modes. This module frames how mechanical choices lock hands with software and operations.

Mechanical integration, structure, and load paths

Learning outcomes

  • Trade off stiffness, mass, and thermal paths for one subsystem.
  • Explain why service access changes total cost of ownership.
  • Link mechanical choices to integration risk in a pilot.
Learn

Loads, materials, power/thermal, human factors for service.

Practice

Red-team a CAD or photo: list three service nightmares.

Challenge

One-page design review memo; post summary on the Forum.

Facilitation: Use real robot photos — students mark fasteners and cable strain points.

Self-check

Arms vs legs — what differs?
Repeatable contact vs impact tolerance; different failure signatures.
What is “integration debt”?
Shortcuts in cabling, cooling, or access that slow every future change.

STEM alignment: engineering design process, constraints & trade studies, communication of design intent.

Why this chapter matters

This is where “the robot works on paper” turns into “the robot can actually be built, serviced, and operated without constant surprises.”

Carry forward from earlier pages

Bring in assumptions from Software, Communication, and the platform chapters so your CAD, wiring, and access decisions match the stack you really intend to run.

Design lenses

  • Stiffness vs mass: arms need repeatable contact; legs need impact tolerance — often opposing goals.
  • Power & cooling: sustained walking + manipulation draws real watts; budget thermals early.
  • Service loops: can a field tech replace a joint in < 30 minutes? If not, downtime dominates.
  • Integration points: align CAD, electrical pinout, and software frame IDs — one naming scheme end-to-end.
  • Humanoids specifically: coordinate G1/H1-class constraints with morphology tradeoffs.
SVRC angle: We see many teams succeed faster when they treat integration as a product discipline — the same mindset behind our Approach and Data Services engagements.

Best next move: continue into industry if you want to translate integration tradeoffs into pilot constraints, or revisit the humanoid chapters if you are still deciding what form factor makes sense.

← Software stack · Next: Industry applications →