Hand: open-source Robot Hand

Naming, Form Factor, and Cost

• Project is called “AmazingHand”; some note the generic “Hand” title hurts searchability.
• Design is praised as approachable, printable, and “cartoon-style” (three fingers + thumb).
• Uses off‑the‑shelf servos and 3D‑printed parts; ~$135 BOM is seen as impressively low.
• Four fingers likely chosen because servo width would make a five‑finger hand uncomfortably wide.

Tendons, Servos, and Control Complexity

• Question raised about tendon-driven hands to move actuator mass into the arm.
• Replies highlight tendon elasticity causing calibration drift, friction changes, and breakage, requiring proprioceptive sensing and continual learning.
• Some argue neural networks in the control loop are needed; others suggest faster, more specialized function approximators.
• Suggestions include optical tracking of tendon motion, external vision-based finger tracking, and absolute encoders where space allows.
• For grasping, several say force sensing is often more useful than precise joint sensing, but elasticity and gravity still matter for delicate objects.

Human-Like Hand vs Alternative Grippers

• Debate on whether a human hand is actually “best” for robots.
• Strong argument: the built environment and objects are designed for human hands, so human-like grippers maximize compatibility and are easier to teleoperate or pretrain with human motion.
• For single, well-defined tasks, simpler dedicated grippers (parallel jaws, chucks, suction, magnets) are cheaper, stronger, and more reliable.
• Some extend this to locomotion (feet vs wheels) and raise ethical questions about designing future environments for human vs machine capabilities.

Strength, Materials, and Manufacturing

• People ask about payload, grip, and failure forces; hand strength ultimately depends on the attached arm.
• PLA parts marked “needs to be strong” are seen as too weak for serious work; suggestions include polycarbonate, glass‑filled nylon, CNC‑machined aluminum, or stamped metal.
• Discussion of potential injection molding or pressed parts notes that tooling and mold design remain a barrier for hobbyists, though small upgrade kits are conceivable.

Sensing and Capabilities

• Several note that to rival human hands, widespread tactile sensing (at least pressure, ideally also temperature) is needed across the surface.
• Adding skins like AnySkin could help but increases weight, cabling, and sensor fusion complexity, potentially limiting real‑world usefulness at this price point.

Use Cases, Safety, and Trajectory

• Many see it as an educational/hobby platform or Halloween prop rather than an industrial tool.
• Some imagine household helpers (wall- or rail-mounted arms for kitchens or laundry) but others worry about safety (e.g., “knife-flinging arms”).
• One thread contrasts older industrial robots optimized for repeatable motions with an emerging vision of general-purpose robots where adaptability and error correction matter more than exact repeatability.
• Overall sentiment is enthusiastic about open-source hardware done this way—fully documented CAD, commodity parts, and room for community-driven iteration—while acknowledging that serious applications would need stronger materials and richer sensing.