Miron's Portfolio

1 · Introduction

“Replacing a missing hand exposes the true complexity of the human body.” — project report, Rev 1.41

This project explores a myoelectric prosthetic arm controlled by surface EMG. The goal is to approach natural hand function while keeping the material cost under $500, producing a platform that’s practical for research and affordable for real users. Motivation & scope (from Rev 1.41):

Physical design: printable, robust mechanics that mimic human hand kinematics.
Control scheme: intuitive EMG control for basic tasks (grasp, release, pinch).
Practicality: useful in day-to-day tasks, not just lab demos.
Affordability: consumer prostheses often cost $20k–$40k; this build targets sub-$500 materials.

2 · Quick specs (current state)

Subsystem	Spec
Actuation	Tendon-driven fingers (artificial ligaments via cord)
Transmission	Bowden/cord routing with captive pulleys
EMG sensing	2–4 channel surface EMG (skin electrodes), analog front-end → ADC
Control	Microcontroller reading EMG envelopes → grip state machine
Grips	Open/close, pinch, tripod, power grasp (configurable)
Hand DOF	Coupled flexion per finger; thumb opposition (mechanical option)
Materials	PLA/PETG rigid + TPU compliant pads/sleeves
Target cost	< $500 (materials)

Note: EMG pipeline uses band-pass (~20–450 Hz), notch (50/60 Hz), rectification, and moving RMS/MA envelope before classification.

3 · Media

3.1 · CAD render — tendon routing

CAD render: tendon network to finger phalanges

3.2 · Engineering drawing (CAD)

3.3 · Finger sequence demo (video)

4 · Design overview

4.1 · Tendon-driven mechanics

Why tendons? Compact, printable, tolerant to misalignment, easy to service.
Links & joints: printed phalanges with captive pins and low-friction bushings.
Pads & contact: TPU grip pads for friction; swappable liners for wear.

4.2 · Control & EMG pipeline

Acquire: surface electrodes → EMG preamp (gain, BP/Notch).
Condition: rectify → moving RMS / moving average envelope.
Map: thresholds / simple classifier → grip state machine.
Command: target tendon displacements (position/velocity) with soft limits.

EMG_raw → BP(20–450Hz) → Notch(50/60Hz) → 
Rectify → RMS(τ≈100–200ms) → Thresholds/Features → Grips

4.3 · Grips & control logic (v1)

Power grasp, pinch, tripod, open.
Long-press / co-contraction to cycle grips; envelope magnitude → speed scaling.
Safety: timeout watchdog, stall detection (current/pos error), soft endstops.

5 · Literature snapshot (context)

Commercial hands (e.g., predefined-grip systems) often couple finger joints, limiting fine control. Research prostheses push mechanics/control but can suffer in robustness and cost. This build aims at a middle ground: printable mechanics + simple, reliable EMG control suitable for real-world trials.

6 · Manufacturing notes

Prints: 0.2–0.28 mm layers; PETG for high-stress parts, TPU for compliant pads.
Tendons: low-creep cord (e.g., UHMWPE/Dyneema class), crimped or knotted with printed retainers.
Assembly: captive pins, heat-set inserts where service is expected.
Serviceability: tendon swap and pad replacement in minutes.

7 · Early results

Sequential finger actuation confirmed (see video above).
EMG grip switching stable with simple envelope thresholds.
Printable mechanics tolerate repeated open/close cycles; pads improve grip.

This is an R&D prototype intended for research/education. It is not a medical device.

8 · Downloads

📄 Project slides (PDF) — Download
📘 Full report Rev 1.41 (PDF) — Download

9 · What’s next?

Add adaptive thresholds per user calibration (on-device).
Evaluate multi-channel EMG for more reliable grip inference.
Long-term durability testing (cord wear, joint play, pad abrasion).
Explore haptic feedback (vibration) for closed-loop grip force cues.

TL;DR A printable, tendon-driven, myoelectric hand under $500 in materials: EMG → simple grip logic → reliable everyday grasps. Built to be hacked on, serviced, and actually used.

Bionic Arm — 3D-Printed Myoelectric Prosthesis