arDiosnostic

AR-assisted PCB and motherboard diagnostics.

arDiosnostic is a bench-first diagnostic platform for electronics repair. It combines computer vision, board profiles, guided measurements, and later augmented reality to help technicians inspect, compare, measure, and confirm faults on PCBs and motherboards.

Philosophy: Measure → Compare → Locate → Suggest → Confirm

The first goal is not magic AI diagnosis. The first goal is a reliable workshop tool that produces evidence.

---

Product Positioning

Do not sell this as:

AI that repairs motherboards automatically

Sell it as:

Visual board diff + measurement assistant for PCB repair.

---

Bench MVP

The MVP intentionally avoids live AR, full AI diagnosis, and oscilloscope-level analysis at the beginning.

MVP Scope

1. Load healthy board image
2. Load damaged board image
3. Align both images
4. Detect visual differences
5. Mark suspicious zones
6. Attach manual component/test-point metadata
7. Read basic voltage/current measurements
8. Generate a JSON/PNG report

Evidence Sources

arDiosnostic should never diagnose without evidence. A suggestion must be backed by at least one of:

• visual difference
• electrical measurement
• board profile data
• repair history
• thermal evidence

---

Core Features

1. Visual Board Diff

Compare a known-good board image against a suspect board image.

Detects:

• missing components
• moved components
• burnt zones
• lifted pads
• solder irregularities
• revision differences

2. Board Profile JSON

A board profile stores component positions, rails, test points, expected voltages, and metadata.

3. Static Overlay

Before live AR, the system draws labels and warnings on a still image:

[TP1 3.3V]    [C245 suspicious]
[U3 PMIC]     [PP3V3 rail]

4. Measurement Assistant

Initial measurement support:

• DC voltage
• DC current
• continuity / low impedance checks
• basic digital HIGH/LOW activity

Oscilloscope-level ripple, PWM, clocks, and buses are future features.

5. Rule-Based Diagnostics

Diagnosis starts as structured rules, not vague AI responses.

Example:

{
  "rule_id": "short_on_rail_01",
  "condition": {
    "rail": "PP3V3_G3H",
    "measured_voltage": {"lt": 0.5},
    "current": {"gt": 0.3}
  },
  "action": {
    "diagnosis": "possible_short",
    "next_step": "inject_limited_current",
    "priority": 1
  }
}

---

Repository Structure

arDiosnostic/
├── apps/
│   └── web/
├── core/
│   ├── vision/
│   ├── diagnostics/
│   ├── measurements/
│   └── board_db/
├── data/
│   └── boards/
│       └── demo_board_001/
├── docs/
├── tools/
├── requirements.txt
├── ROADMAP.md
└── README.md

---

Quick Start

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Run a visual diff demo:

python -m core.vision.diff \
  --reference data/boards/demo_board_001/reference_top.png \
  --candidate data/boards/demo_board_001/candidate_top.png \
  --out reports/demo_diff.png

Read a hardware probe:

python -m core.measurements.serial_reader --port auto

Windows example:

python -m core.measurements.serial_reader --port COM3

macOS/Linux example:

python -m core.measurements.serial_reader --port /dev/ttyUSB0

---

Safety

This project can interact with powered electronics. Use strict safety practices:

• do not inject voltage without current limit
• discharge large capacitors before probing
• avoid measuring high voltage without isolation
• do not use liquid nitrogen in MVP workflows
• avoid condensation on energized boards
• use ESD protection
• treat in-circuit resistance readings as unreliable unless confirmed

---

Long-Term Vision

The full platform may later include:

• mobile AR overlay
• thermal hotspot detection
• repair history intelligence
• continuity assistant
• OpenBoardView export
• oscilloscope bridge
• collaborative board profile database
• training mode for technicians

---

Slogan

No adivina.
Mide.
Compara.
Localiza.
Confirma.