Reference Library

Domain research in
biomedical engineering
and Physical AI.

An annotated bibliography of primary references assembled across fifteen years of engineering practice — spanning medical device design, surgical robotics, Physical AI, and precision manufacturing. Organized as a living research corpus and fabrication curriculum resource.

484 Curated Videos
231 Repositories
1.05M Transcript Words
18 Technical Clusters
YouTube Playlist 484 videos GitHub Repositories 231 repos
01

Surgical Robotics &
Medical Devices

This cluster anchors the intersection of regulatory engineering and clinical device translation. Resources range from clean room aseptic manufacturing standards to the full JHU da Vinci Research Kit software ecosystem. The emphasis is on understanding the complete device lifecycle — from clinical needs identification through OR shadowing, prototype verification, and ISO 13485 design control. Selected for direct application to electromechanical actuator design, implantable device architecture, and AI-assisted surgical systems.

Video Resources

Repository References

  • sawIntuitiveResearchKit ↗ jhu-dvrk / sawIntuitiveResearchKit

    The complete cisst/SAW software stack for the da Vinci Research Kit (dVRK). Covers kinematic control, ROS integration, teleoperation, and haptic feedback for surgical robotics research. Primary reference for the GI Surgical Actuator capstone.

    C++
  • i4h-workflows ↗ isaac-for-healthcare / i4h-workflows

    NVIDIA Isaac for Healthcare reference workflows. Bridges sim-to-real transfer for medical robotics — surgical instrument tracking, OR environment simulation, and AI-driven clinical workflow automation.

    Python
  • MONAI ↗ Project-MONAI / MONAI

    PyTorch-based AI toolkit purpose-built for healthcare imaging. Covers segmentation, classification, detection, and registration across CT, MRI, and pathology modalities. Reference for integrating imaging intelligence into device design pipelines.

    Python
  • cisst ↗ jhu-cisst / cisst

    Johns Hopkins ERC CISST Library — component framework underlying all dVRK software. Provides real-time data collection, state machines, and multi-threaded component architecture for surgical robotics systems.

    C++
  • pdf-compliance-validator ↗ tusha-p / pdf-compliance-validator

    Open-source tool for flagging potential 21 CFR Part 11 issues in PDFs. Validates audit trail completeness, timestamp formatting (ISO 8601), and signature requirements. Reference for QMS document validation and regulatory submission preparation.

    Java
  • ambf ↗ WPI-AIM / ambf

    Asynchronous Multi-Body Framework for surgical simulation. Enables real-time physics simulation of deformable tissues and rigid surgical instruments. Used as a reference for pre-clinical actuator testing environments.

    C++
02

Physical AI, Robotics
& Simulation

Physical AI — as distinct from language or generative AI — refers to intelligence embedded directly into physical systems: robots, manufacturing equipment, and medical devices that perceive, decide, and act. This cluster covers the simulation infrastructure (NVIDIA Isaac), sensor fusion pipelines, 3D scene understanding, and reinforcement learning frameworks that underpin next-generation autonomous manufacturing and surgical systems. Selected for their direct applicability to intelligent fabrication, computer vision quality control, and AI-driven clinical translation.

Video Resources

Repository References

  • IsaacSim ↗ isaac-sim / IsaacSim

    NVIDIA Isaac Sim — open-source robotics simulation on Omniverse. Photorealistic rendering, physics, and sensor simulation for developing and testing AI-driven robots. Primary reference for Physical AI manufacturing system development.

    Python
  • ToolOrchestra ↗ NVlabs / ToolOrchestra

    End-to-end RL training framework for orchestrating tools and agentic workflows. Reference for designing AI systems that coordinate multiple manufacturing processes and decision agents.

    Python
  • vomp ↗ nv-tlabs / VoMP

    Predicting volumetric mechanical property fields from geometry alone. Directly applicable to AI-assisted material property estimation in medical device design and FEA preprocessing.

    Python
  • FAST_LIO ↗ hku-mars / FAST_LIO

    Computationally efficient LiDAR-inertial odometry for robotic navigation in complex environments. Reference for sensor fusion in surgical robot localization and autonomous manufacturing cell navigation.

    C++
  • pycuvslam ↗ nvidia-isaac / cuVSLAM

    NVIDIA's GPU-accelerated visual SLAM system. High-accuracy simultaneous localization and mapping for robots operating in unstructured clinical and manufacturing environments.

    Python
03

Advanced Manufacturing &
Digital Fabrication

The largest cluster in the library, reflecting the breadth of fabrication methods required to translate a clinical need into a manufacturable device. Covers injection mold design (gates, ejectors, cooling, multi-cavity), five-axis additive manufacturing, robotic toolchanging, resin process control, and vacuum forming. Resources are selected for their direct applicability to ISO 13485-compliant prototype fabrication, DFM curriculum development, and the specific manufacturing methods used in the UNC BME Fabrication Lab and consulting practice.

Video Resources

Repository References

  • StealthChanger ↗ DraftShift / StealthChanger

    Open-source robotic toolchanging system for Voron-architecture printers. Reference for CAN-bus-networked multi-material and multi-process fabrication cells — directly applicable to automated bioprinting and multi-material device fabrication.

    Python
  • Open5x ↗ FreddieHong19 / Open5x

    Open-source 5-axis 3D printing hardware and software. Reference for continuous fiber orientation control in structural medical device components and patient-specific implant geometries that cannot be fabricated with 3-axis systems.

    GAP
  • Marlin ↗ MarlinFirmware / Marlin

    The dominant open-source firmware for FDM 3D printers. Reference for understanding low-level motion control, G-code execution, and temperature management in fabrication equipment used throughout the lab.

    C++
  • UVtools ↗ sn4k3 / UVtools

    MSLA/DLP resin 3D printer file analysis, calibration, repair, and manipulation toolkit. Reference for optimizing resin print parameters for biocompatible resins used in microfluidic device and dental device fabrication.

    C#
  • stlTexturizer ↗ CNCKitchen / stlTexturizer

    Bump-mesh 3D model texture engine for applying surface textures to STL geometry. Reference for osseointegration surface design on orthopedic implants and grip-texture features on handheld surgical instruments.

    JavaScript
04

PCB Design &
Embedded Systems

Electromechanical device design requires fluency at every level of the electronics stack — from schematic capture and PCB layout through firmware, real-time control loops, and communications protocols. This cluster covers KiCad and Eagle PCB workflows, CAN bus and RS485 industrial protocols, FOC BLDC motor control, PID tuning, dual 9-DOF IMU sensor fusion with quaternion mathematics, and the ESP32/STM32 ecosystems that underpin the lab's embedded electronics curriculum and device prototyping practice.

Video Resources

  • 23
    Metallization of Vias — CAN Bus Indicator Using STM32

    Via metallization for multilayer PCBs and STM32-based CAN bus implementation. Direct reference for the PCB fabrication teaching module and embedded communications design in medical device prototypes.

  • 47
    Beyond UART: RS485

    RS485 differential signaling for multi-drop industrial networks. Reference for long-cable embedded systems in medical equipment and manufacturing environments where EMI noise immunity is required.

  • 97
    How Drone ESCs Work

    FOC (Field-Oriented Control) BLDC motor control theory and ESC architecture. Foundation reference for understanding torque ripple, back-EMF, and the control mathematics underlying electromechanical actuators in surgical and industrial devices.

  • 117
    ESP32 Guide 2026

    Comprehensive current-state ESP32 ecosystem reference — covering ESP-IDF, Arduino framework, Bluetooth, Wi-Fi, power management, and OTA update strategies for connected medical and industrial devices.

  • 124
    How to Order Assembled PCB Boards Using KiCad

    End-to-end KiCad to JLCPCB PCBA workflow. Reference for the PCB fabrication curriculum and for transitioning device prototypes from breadboard to assembled PCB within an ISO 13485 design history file.

Repository References

  • esp-idf ↗ espressif / esp-idf

    Official Espressif IoT Development Framework. The production-grade alternative to the Arduino abstraction layer for ESP32 devices requiring deterministic timing, custom RTOS task management, or IEC 60601-compliant power behavior.

    C
  • candleLight_fw ↗ candle-usb / candleLight_fw

    gs_usb-compatible firmware for USB-to-CAN adapters. Reference for CAN bus development tooling — used alongside cangaroo for monitoring and debugging CAN networks in multi-axis electromechanical device prototypes.

    C
  • smartknob ↗ scottbez1 / smartknob

    Haptic input knob with software-defined endstops and virtual detents using FOC motor control. Reference for surgical instrument tactile feedback design and configurable haptic UI in medical device control panels.

    C++
  • BLHeli ↗ bitdump / BLHeli

    Open-source ESC firmware implementing FOC BLDC control. Primary reference for understanding motor commutation, quaternion-based orientation tracking, and the PID control loop mathematics underlying motion control in electromechanical devices.

    Assembly
  • photon ↗ photonfirmware / photon

    Open-source firmware for automated pick-and-place feeders. Reference for the lab's SMT assembly workflow and for understanding how industrial feeder hardware integrates with automated PCB assembly systems.

    C++
05

Materials, Finishing &
Precision Machining

Material selection is a regulatory and clinical decision as much as an engineering one. This cluster covers the engineering polymers central to medical device design (PEEK, PPS, PTFE, biocompatible silicone), surface treatment processes (anodizing, cerakote, vapor polishing, electroplating), and the precision machining methods required to achieve tolerance-critical geometries. Resources include GD&T application, tolerance stack-up, wire EDM, Swiss machining, and metrological techniques — all selected in the context of manufacturing components that must perform reliably in clinical environments.

Video Resources

  • 111
    Engineering Drawings: How to Make Prints a Machinist Will Love

    GD&T and engineering drawing best practices for manufacturing communication. Direct curriculum reference for the ASME Y14.5 content delivered in UNC BME Junior Design fabrication sections.

  • 151
    How Wire EDM Works

    Wire EDM principles, kerf compensation, and surface finish capabilities. Reference for understanding when EDM is the appropriate manufacturing method for high-hardness, complex-profile medical device tooling and components.

  • 159
    Silicone Explained

    Comprehensive reference on silicone elastomer chemistry, shore hardness, tear strength, biocompatibility, and processing methods. Essential for selecting silicone grades for wearable sensors, implantable seals, and soft-tissue-contacting device components.

  • 165
    Why Scratching Metal Creates Extreme Precision

    The mechanics of precision surface generation through lapping and scraping. Reference for understanding how machine tool accuracy is established and maintained — foundational context for metrology in device manufacturing environments.

  • 186
    Engineering Plastics for R&D

    Selection guide for engineering thermoplastics in research and development contexts — comparing machinability, chemical resistance, sterilizability, and regulatory status. Reference for material selection rationale in design history files.

Repository References

  • micromanipulatorstepper ↗ 0x23 / MicroManipulatorStepper

    Sub-micrometer 3D motion control platform. Reference for precision positioning system design in surgical robotics, optogenetics rigs, and any device requiring nanometer-scale displacement resolution and repeatability.

    C++
  • tubular-linear-motor ↗ stijnsprojects / tubular-linear-motor

    Open-source tubular linear synchronous motor (LSM) design. Reference for understanding direct-drive linear actuation — relevant to surgical tool actuation, biopsy needle drives, and other medical devices requiring force-transparent linear motion.

    Python
  • gcode-cmm ↗ XDGFX / gcode-cmm

    Adapts a USB-controllable 3D printer into a coordinate measuring machine (CMM). Reference for low-cost in-process measurement in a fabrication lab context — used to validate geometric tolerances on printed fixtures and jigs.

    Python
  • reremeter ↗ arminstr / reremeter

    Low-cost handheld NIR reflectance spectroscopy system for plastic type identification. Reference for non-destructive material verification in incoming inspection workflows — relevant to polymer identification in device component receiving.

    C
06

Agentic AI &
Intelligent Systems

This cluster addresses the infrastructure for embedding AI into physical engineering workflows — not as an interface layer, but as a functional component of the manufacturing and design system itself. Covers retrieval-augmented generation (RAG) architectures, local large language model deployment, agentic workflow orchestration, and multimodal vision-language models. Resources are selected specifically for their applicability to AI-assisted regulatory documentation, intelligent process automation in fabrication environments, and the design of Physical AI systems for medical device manufacturing.

Video Resources

  • 27
    The 7 Levels of Claude Code & RAG

    Progressive RAG architecture from basic document retrieval to agentic multi-step reasoning. Framework reference for designing AI systems that operate on engineering documentation, regulatory filings, and manufacturing process records.

  • 28
    My Full OpenClaw Setup

    Local AI assistant configuration and deployment. Reference for setting up privacy-first, on-device LLM infrastructure suitable for engineering environments handling ITAR-controlled or proprietary device design data.

  • 33
    How to Quickly Summarize Requirements — IBM DOORS Next + AI Hub

    AI-assisted requirements management in IBM DOORS Next. Direct reference for applying language models to ISO 13485 design input/output traceability and FDA 21 CFR Part 820 design control documentation.

  • 17
    Claude Code Best Practices

    Systematic approaches to agentic coding with Claude. Reference for building AI-assisted engineering automation tools — from design script generation to automated test protocol drafting in regulated device development.

  • 298
    Python RAG Tutorial (with Local LLMs): AI For Your PDFs

    Practical implementation of document-grounded RAG using local models. Reference for building offline-capable AI systems that operate on design history files, SOPs, and regulatory submissions without sending data to external APIs.

Repository References

  • LightRAG ↗ HKUDS / LightRAG

    Fast graph-and-vector RAG pipeline from EMNLP 2025. Reference architecture for building knowledge retrieval systems over engineering corpora — regulatory standards libraries, manufacturing SOPs, and device design documentation.

    Python
  • RAG-Anything ↗ HKUDS / RAG-Anything

    All-in-one RAG framework supporting text, images, tables, equations, and mixed-modality documents. Reference for processing the heterogeneous document types that appear in medical device technical files — test reports, engineering drawings, and clinical evaluation reports.

    Python
  • anything-llm ↗ Mintplex-Labs / anything-llm

    Privacy-first, on-device AI productivity stack with document ingestion, RAG, and multi-LLM support. Evaluated as an on-premises alternative for regulated engineering environments where cloud API usage is restricted by ITAR or confidentiality requirements.

    JavaScript
  • NemoClaw ↗ NVIDIA / NemoClaw

    Run OpenClaw inference securely inside NVIDIA OpenShell with managed inference. Reference for deploying AI assistants in air-gapped or restricted-network manufacturing and research environments.

    TypeScript
  • ollama ↗ ollama / ollama

    Local LLM runtime supporting Llama, DeepSeek, Gemma, and other open models. The foundation layer for offline AI tooling in device development workflows — enabling model inference without external API dependencies or data egress.

    Go