Bio-Inspired Tendon-Driven Robotic Hand (Underactuated & Parametric)

Bio-Inspired Tendon-Driven Robotic Hand (Underactuated & Parametric)

This is a high-performance, bio-inspired robotic hand designed for researchers, hobbyists, and prosthetic developers. Inspired by the sophisticated musculoskeletal structure of the human hand, this design prioritizes dexterity, durability, and affordability.

The hand utilizes a tendon-driven actuation system, mimicking how forearm muscles transmit force through tendons to move biological joints. It features an underactuated design, meaning 15 kinematic degrees of freedom are controlled by only 5 actuators. This allows the fingers to mechanically "conform" and wrap around objects of various shapes without complex sensors or inverse kinematics.

Key Engineering Innovations

  • Flared Tendon Channels: Conical "trumpet" geometry (14° transition) reduces friction and contact stress on cables by 8x, significantly extending the life of your tendon lines.
  • Mechanical Hyperextension Stops: Integrated ridges prevent joints from "flopping" backward, maintaining control stability during extension.
  • Parametric OpenSCAD Workflow: The entire model is driven by global variables. Need a smaller child-sized hand or a larger one? Change a single line of code in the provided .scad file to resize the geometry instantly.
  • Optimized for FDM Printing: All parts are designed to be printed flat on the bed. This aligns the filament strands with the primary load path, preventing the delamination common in vertical 3D prints.
  • Dual-Channel Antagonistic Control: Features both Dorsal and Volar channels for active flexion and extension, providing precise bidirectional control.

Applications

  1. Affordable Prosthetics: A functional alternative to expensive commercial myoelectric hands. Total Bill of Materials is approximately £200 compared to £25,000+ systems.
  2. Teleoperation: Ideal for hazardous material handling using a master-slave configuration with a data glove.
  3. Research: A robust platform for testing EMG control, neural networks, and adaptive grasping.

Bill of Materials (BOM)

CategoryItemQuantityPurpose
3D PartsPLA or PETG Filament~150gPrinting the palm and fingers.
ActuatorsMG996R Servo Motors5High-torque (10kg/cm) drivers.
TendonsDyneema or Fishing Line1 SpoolHigh-modulus fiber for actuation.
FastenersM3 BoltsAs neededStandard pivot pins for joints.
ControlEMG Sensors/Arduino1 SetOptional for gesture-based control.

Assembly Instructions

1. Printing the Components

  • Orientation: Ensure all phalanges are printed flat (rotated -90°). The provided OpenSCAD script includes logic to handle this automatically.
  • Tolerance: The design includes a 0.5mm parametric gap in the tongue-and-groove joints to ensure smooth movement straight off the printer.

2. Joint Assembly

  • Modular Chains: Assemble the fingers using the standard segments (Proximal: 35mm, Medial: 25mm, Distal: 22mm).
  • Mechanical Stops: When joining segments, ensure the interference ridges on the dorsal side are aligned to prevent hyperextension.
  • Pivot Pins: Insert M3 bolts through the joint holes. For rapid prototyping, pieces of 3D printing filament can be used as pins.

3. Tendon Routing

  • Dual Channels: Route two lines through each finger—one through the Dorsal channel for extension and one through the Volar channel for flexion.
  • Friction Reduction: Ensure the tendons pass through the flared entries to minimize fraying.
  • Palm Hub: Guide all 10 tendons through the internal palm pathways toward the wrist exit.

4. Actuation Setup

  • Dual-Spool Pulley: Connect each finger's two tendons to a single MG996R motor using a dual-spool pulley. This enables active "pull-pull" control.
  • Underactuation: Note that a single motor drives all three joints of a finger simultaneously; the joint with the least resistance moves first.

Troubleshooting & Limitations

  • Grip Force: The estimated grip force is approximately 10N. This is suitable for light tasks but may fall short for heavy-duty lifting compared to commercial benchmarks.
  • Sensory Feedback: This version is open-loop and lacks native haptic or force sensors.
  • Friction: While flared channels reduce wear, tendon friction still consumes a portion of the servo's stall torque.
  • Compliance: Tendon systems are inherently compliant; if the hand accidentally impacts an object, the tendons may slacken or stretch to prevent damage.

SEO Tags / Keywords

Robotic Hand Bionic Hand Prosthetic OpenSCAD Tendon-Driven Bio-inspired Underactuated Robotics 3D Printed Hand Anthropomorphic EMG Control Parametric Design Cybernetics DIY Prosthetic Mechanical Hand