CircuitSense AI is a hybrid diagnostic framework designed for embedded systems such as Arduino and ESP32. The system integrates Large Language Model (LLM) reasoning with deterministic electrical validation to provide reliable, structured, and hardware-aware debugging assistance.
Unlike generic AI chat systems, CircuitSense AI separates probabilistic reasoning from physics-based validation to reduce hallucinated outputs and improve technical reliability.
Embedded debugging involves:
- Hardware-software coupling
- Electrical constraints (voltage, current, logic levels)
- Peripheral communication protocols (I2C, UART, SPI)
- Firmware flashing and bootloader dependencies
Traditional debugging relies on:
- Manual datasheet lookup
- Forum search
- Trial-and-error
This increases debugging latency and reduces productivity.
CircuitSense AI introduces a hybrid architecture:
LLM-Based Semantic Reasoning
+
Deterministic Electrical Rule Engine
+
Module-Specific Knowledge Graph
This layered approach ensures structured and reliable embedded diagnostics.
User Interface
↓
Flask REST API
↓
Parallel Diagnostic Layer
→ LLM Diagnostic Engine
→ Electrical Validation Engine
→ Module Knowledge Base
↓
Response Synthesizer
↓
Structured Debug Report
- Parses embedded error logs
- Identifies probable root causes
- Generates structured troubleshooting steps
- Ranks failure hypotheses
Implements deterministic validation:
Voltage Divider: Vout = Vin × (R2 / (R1 + R2))
ADC Scaling: ADC_value = (Vin / Vref) × 1023
Tolerance: Flags anomaly if deviation > ±5%
This prevents physically impossible AI suggestions.
Includes structured troubleshooting for:
- HC-05 Bluetooth
- L298N Motor Driver
- I2C LCD
- Voltage Sensors
Each module defines:
- Failure states
- Dependency relationships
- Corrective sequences
To mitigate AI hallucination:
- Physics-based validation layer
- Root cause ranking
- Structured output formatting
- Module constraint enforcement
This increases diagnostic trustworthiness.
Phase 1 – Web Prototype
Phase 2 – IDE Plugin Integration
Phase 3 – Real-Time Serial Monitor Analysis
Phase 4 – Circuit Image Recognition
CircuitSense-AI/ │ ├── README.md ├── requirements.md ├── design.md └── (future implementation files)
CircuitSense AI follows a hybrid intelligence model:
Probabilistic reasoning
+
Deterministic physical validation
+
Domain-specific constraint graphs
This ensures structured, reliable, and technically sound embedded debugging.
✔ Documentation Completed
✔ System Architecture Defined
✔ Hybrid AI Framework Designed
⬜ Prototype Under Development
⬜ AI Integration Planned
This repository represents the architectural and conceptual foundation of CircuitSense AI for the idea submission stage.