This project is a voice-controlled Iron Man–style helmet powered by an ESP32 and a local AI system running on a computer.
The helmet reacts to spoken commands by moving the mask and controlling the eye lights, and responds conversationally using a locally hosted language model — inspired by the idea of JARVIS.
The system is designed to be modular, privacy-friendly, and fully local, without reliance on cloud services.
- 🎙️ Voice input captured directly from the helmet
- 🔊 Spoken responses through an onboard speaker
- 🤖 Conversational AI using a local LLM
- 🦾 Physical actuation (mask open/close) using servo motors
- 👀 LED eye control via voice commands
- 🔌 Simple and robust power system (power bank based)
The project is split into two main parts:
- ESP32 microcontroller
- Digital I2S microphone
- Servo motors for mask movement
- LEDs for eye illumination
- Speaker with I2S audio amplifier
- Powered by a 5 V power bank
The helmet acts as a smart input/output device: it captures voice, executes physical actions, and plays audio responses.
- Voice recognition (ASR)
- Command detection
- Local language model (LLM)
- Optional local TTS
- Bluetooth communication with the helmet
All AI processing is done locally, ensuring low latency and full offline operation.
- User speaks into the helmet
- ESP32 sends audio to the computer via Bluetooth
- Computer converts speech to text
- If the text matches a command:
- A control instruction is sent back to the helmet
- If it does not match a command:
- The text is processed by a local LLM
- A spoken response is generated and sent back
- Helmet reacts with movement, lights, or audio
- ESP32 development board
- INMP441 I2S microphone
- 2× MG90 servo motors
- I2S audio amplifier
- 8 Ω / 2 W speaker
- LEDs (eyes) + resistors or MOSFETs
- 5 V / 2 A power bank
- Bulk and decoupling capacitors
- Helmet shell and mechanical mounts
The project intentionally avoids complex battery management in early and final builds by using a 5 V power bank, providing:
- Stable voltage
- High peak current capability
- Built-in protection
- Simplified wiring
Mechanical and software constraints are used to ensure reliable operation within the current limits.
🚧 In active development
Current focus:
- Validating full voice → response loop
- Refining servo motion and mechanics
- Optimizing latency and interaction feel
The project is intentionally developed in iterative stages, prioritizing reliability and clarity over premature optimization.
- Bluetooth audio streaming optimization
- Local ASR model integration
- Local LLM selection and tuning
- Mechanical refinement of mask system
- Final helmet integration and testing
- Fully local AI (no cloud dependency)
- Clear separation between embedded control and AI logic
- Wearable-friendly power and thermal behavior
- Clean, maintainable architecture
This project is released under the MIT License.
Feel free to use, modify, and build upon it.