Amar Malhi

My Role & Contributions

Personally owned mechanical architecture, kinematic analysis, housing/attachment design, 3D-print fabrication strategy, and the SolidWorks drawing package.

Directed a 60+ person program across mechanical, electrical, software, ergonomics, physiological, and safety subteams (leadership role in addition to the mechanical ownership above).

Built PLA fit-check parts then a PETG functional model with 0.3–0.5 mm joint clearances for as-printed motion.

First-authored the multi-author technical paper on sEMG thresholding, mechanical design, and subsystem integration.

Project Overview

Status: first-semester prototype completed May 2026 (this page documents that delivered build).

Sub-$500 unit cost vs. clinical myoelectric devices that often exceed $20,000.

sEMG from brachioradialis and flexor groups with ≤100 ms system latency design target and EMA filtering (α = 0.1).

19-part tendon-driven hand; MG996R servos rated 12 kg·cm vs. ≥2.52 kg·cm torque requirement.

<500 g assembly mass budget; 10–15 N ADL grip-force requirement; ~1.2 kgf measured on hanging-scale grip tests.

Technical Milestones

#	Objective	Design Target
1	sEMG capture (brachioradialis & flexor groups)	≤ 100 ms total system latency
2	DSP layer — exponential moving average filter	Stabilize noisy MUAPs; reduce EMI-driven accidental triggers
3	Modular strap-based suspension	Accommodate edema and limb volume changes
4	Weight-optimized 3D-printed chassis	< 500 g assembly · 10–15 N ADL grip

Cost Comparison vs. Market

Device Class	Cost Range	Control	Materials
High-end clinical (e.g. Össur i-Limb)	$30,000 – $100,000	Multi-channel sEMG / TMR	Carbon fiber / titanium
Mid-range commercial (e.g. TASKA)	$15,000 – $30,000	Pattern recognition	High-impact plastic / alloy
Open-source / hobbyist (e.g. e-NABLE)	$50 – $200	Mechanical / passive	PLA / PETG
This prototype	$350 – $500	Single-channel sEMG w/ EMA	PLA / PETG (+ industrial TPU/resin options)

Mechanical Design Specifications

Parameter	Value	Notes
Hand assembly	19 parts	Hinge joints (PIP/DIP analog); no MCP rotation
Grip force requirement	10–15 N	Normal ADL loads
Experimental grip force	~1.2 kgf (~12 N)	Hanging-scale validation
Required servo torque	≥ 2.52 kg·cm	3-joint tendon path + friction margin
Servo rating	12 kg·cm	Hosyond MG996R × 3
Joint print clearance	0.3 – 0.5 mm	Smooth rotation without post-print sanding
Prototype material	PLA	Iterative fit checks
Functional material	PETG	Higher durability under operational stress
Sizing basis	95th percentile	Adjustable base for smaller users

Control & Electronics Architecture

Component	Specification	Role
MyoWare 2.0	sEMG sensors	Brachioradialis & flexor muscle groups
ESP32	Microcontroller	Analog input, servo control, signal processing
EMA algorithm	Smoothing α = 0.1	Stabilize raw voltage; threshold-based actuation
System latency	≤ 100 ms	Neurological integration design target
PCB	Custom board	Sensor input, motor drive, power distribution
Battery	Top-mounted compartment	Weight distribution; ESP32 clearance

Fail-safe principle: localized fuse protection per subcircuit; passive vented cooling for ESP32 and battery pack. EMA/threshold logic was designed to reduce EMI-driven false triggers; quantified before/after rates were not published in the paper.

Subsystem Integration

Subteam	Primary Contribution	Mechanical Interface
Mechanical	Hand, housing, attachment, SolidWorks package	Tendon routing, servo clearance, structural plate
Electrical	PCB, wiring, power distribution	Drives housing geometry and component layout
Software	EMA filtering, threshold logic	Constrains sensor placement and sampling
Physiological	MyoWare activation thresholds	Defines muscle group targeting
Ergonomics	95th-percentile sizing, strap system	Housing dimensions and attachment geometry
Safety	Fusing, materials review, load limits	Reviews insulation, thermal, and structural revisions