Robotic Hand

Goals

Robotic hands are fascinating because they sit at the meeting point of mechanics, control, and anatomy. I wanted to build a low-cost platform that was not merely functional, but useful as a thinking tool: a hand that could be reproduced, modified, and studied as different actuation methods and design ideas were tested against the stubborn realities of physical movement.

Prototype Progression

Prototype One

The first prototype established key baselines, including the relative success of servo-driven tendon routing and how well inexpensive parts could hold alignment under load.

Prototype Two

The second prototype was lighter and lower profile. I switched to rolling contact joints, using 0.2 mm fishing line as the contact surface instead of bulky axles. This reduced friction and simplified the assembly.

Prototype Three

The third prototype was the most ambitious and also the easiest to assemble. I scanned a human hand skeleton and connected each segment to mimic biological joints. This created a more realistic imitation of the forces involved, with the goal of inheriting the advantages built into human anatomy.

Design Themes

Rolling-Contact / Non-Traditional Finger Joints

I moved beyond simple pin joints to achieve smoother, more human-like motion and force transmission. The rolling-contact joints explored here draw from biological joints and precision mechanisms, emphasizing cam surfaces, rolling interfaces, and constrained motion over software compensation.

• Reduced backlash and wear versus revolute joints.
• Smoother torque transfer and more natural finger trajectories.
• Mechanical intelligence before software intelligence.

Tendon-Driven Finger Actuation Concepts

I investigated tendon / cable-driven fingers to move motors closer to the palm or forearm, lowering distal mass and increasing dexterity. This work focused on biomechanics-inspired routing and control coupling, rather than a servo-per-joint approach.

• Elasticity, hysteresis, and tension management in tendon routing.
• Tension control versus position control for stable grasping.
• Tradeoffs between underactuated and fully actuated fingers.

Force, Compliance, and Tactile Interaction

The goal was not just motion, but safe and expressive interaction. I prioritized compliance both in materials and in control to avoid brittle, rigid grasps.

• Passive compliance through joint design and material selection.
• Active compliance through impedance and admittance control.
• Hands are for interaction, not just positioning.

Manufacturing-Constrained Design (Real-World Buildability)

Every design decision was filtered through buildability. The project emphasized simple assemblies and serviceable parts that can be printed, tuned, and repaired without custom machining.

• 3D-printable joint geometries with realistic tolerances.
• Off-the-shelf bearings, tendons, and fasteners.
• Elegance includes manufacturability.

Prototype Videos

Early videos of the robotic hand in motion.