Project files

🚗 OpenKidCar

🏎️ Ganbei No.1 — AI-powered Smart Children's Kart

Dad wants to build you a little car, Dad wants to spend more time with you.

Not just buying you a toy car, but setting an example for you.

🌱 Project Origin

My son is three years old this year.

At this age, children are interested in all kinds of vehicles—buses, trucks, motorcycles, race cars, excavators, cranes, ride-hailing cars, and more.

Most parents would choose to buy a ready-made children’s car.

But I wanted to do something different:

Build a car for him with my own hands.

This car doesn’t have to be the fastest or the most luxurious in the world.

But it will have:

• A design created by Dad;
• Code written by Dad;
• An electronic system assembled by Dad;
• The entire creative process completed by Dad.

I hope that when he grows up, he will remember not just a car.

But rather:

That once, Dad tried to create something with his own hands.

🚀 Project Vision

OpenKidCar is an open-source project for intelligent children's electric cars based on AI + open-source hardware + embedded systems.

The goal is not simply to manufacture a children's toy car, but to explore:

In the AI era, can an ordinary person leverage AI to accomplish hardware creation that previously required a professional team?

OpenKidCar aims to build:

• An open-source software platform;
• A modular hardware architecture;
• An expandable intelligent car system;
• A Maker project that ordinary families can participate in.

✨ Project Features

🤖 AI-Driven

Assisted by AI in:

• Product design
• Software development
• Hardware learning
• Circuit design
• 3D modeling
• Documentation writing
• Debugging

OpenKidCar itself is also an experiment about:

How AI can help ordinary people create real-world products

🔧 Modular Architecture

OpenKidCar uses a layered design similar to a real car:

                    AI Intelligence Layer

              Brain

                    │

              Communication Interface

                    │

              Vehicle Control Layer


                    │

        ┌───────────┼───────────┐

        │           │           │

      Motors      Sensors     Lights

                    │

              Electric Vehicle Hardware Platform

🏗️ System Architecture

AI Control Unit (Raspberry Pi)

Responsibilities:

• AI interaction
• Voice recognition
• Camera vision
• Data processing
• Mobile device connection
• Advanced control logic

Vehicle Control Unit (Arduino)

Responsibilities:

• Throttle control
• Motor control
• Lighting control
• Sensor reading
• Safety protection
• Real-time control

Hardware Execution Layer

Includes:

• Battery system
• Motor system
• Motor controller
• Chassis
• Sensors
• Custom shell

📌 Current Stage

✅ Software layer is ready → Moving towards the real car

The current stage is: "Digital twin has fully verified the decision-making logic and is transitioning to real motors/frames"

• ✅ The real signal chain of the brain (decision-making) + cerebellum (real-time execution) has been fully tested in simulation
• 🚧 Powertrain sizing calculation document: docs/powertrain_sizing.md — Basis for motor/battery/ESC selection
• 🚧 Wiring guide: docs/wiring_guide.md — Follow the instructions even without an electronics background
• 🚧 Firmware to-do list: docs/firmware_todo.md — Software gaps from simulation to real car

Car-building philosophy: Use mature, off-the-shelf components for safety-critical parts such as frame, load-bearing, and brakes; 3D printing is only used for appearance decoration (can be bypassed). First, let the existing car "come to life," then focus on aesthetics.

V0.1 Desktop Smart Car Control Demo (Digital Twin Dual-End)

Objective:

To complete the core system verification of the smart car using the "brain + cerebellum" real signal chain without a physical car body.

Achieved:

✅ Raspberry Pi and Arduino communication (protocol V0.3, CRC check / heartbeat / safety commands)

✅ Real collaboration between brain (decision-making) and cerebellum (execution), with signals passing through a real protocol chain

✅ Digital twin dual-end: car end (dashboard + control area) + parent end (GPS / battery level / walkie-talkie / video)

✅ Cheers Assistant (AI voice assistant): hold to speak → Xiaozhi official server → streaming voice reply (replacing the old offline Vosk voice control)

✅ Cheers Assistant voice control car: say "turn on headlights / LED strip / mute / gear / emergency brake" → MCP tool → brain → car execution

✅ Real sound effects (engine / horn CC0 recordings, speed changes with throttle)

✅ Walkie-talkie (two-way push-to-talk), video call (car pushes stream to parent)

🚗 Car-side (http://localhost:8000 (opens in a new tab))

One-screen display: circular speed gauge, throttle/brake vertical rod (steering-linked turn signals), gear, 8 function keys (headlights/mute/LED strip/horn + cheers assistant/emergency brake/video/intercom), battery/power/voltage/current/temperature/GPS, light status bar, event log.

📱 Parent's End (http://localhost:8001 (opens in a new tab), mobile-friendly)

One-screen monitoring on mobile: GPS real-time trajectory map (with trajectory playback), battery monitoring, video calls (display windows as needed), walkie-talkie, remote emergency stop + remote gear setting + remote power on/off.

Demo effect: When the child says "turn on the headlights," Cheers Assistant (AI voice) recognizes → brain makes decision → protocol → cerebellum executes, turning on the LED. Parents can see the GPS location, perform remote emergency stops, engage in voice chat, and view the video feed in real time on their phones.

🛣️ Roadmap

V0.1: Smart Car Electronic Control Prototype

Objective:

Verify the software and electronic architecture.

V0.2: Power System Integration

Objective:

Add:

• Motors
• Motor controllers
• Battery system
• Speed control

V0.3: Real Vehicle Platform

Goals:

Completion:

• Modification of children's electric car chassis
• Custom body design
• Instrumentation system
• Lighting system

V1.0: OpenKidCar Smart Kids' Car

Goals:

Implementation:

• AI voice assistant
• Parental control
• Mobile app
• Automatic safety protection
• 3D-printed appearance
• Complete open-source documentation

📂 Project Directory

OpenKidCar/
├── README.md
├── LICENSE
├── .github/               # Community documentation + Issue/PR templates
│   ├── CONTRIBUTING.md
│   ├── CODE_OF_CONDUCT.md
│   └── SECURITY.md
├── docs/                  # Product requirements, architecture, protocols, roadmap
│   ├── product_requirements.md
│   ├── roadmap.md
│   ├── architecture.md
│   ├── hardware_io_map.md       # Pin assignment diagram (interface contract)
│   ├── hardware_deployment_guide.md  # Hardware deployment guide (wiring/download/flashing)
│   ├── communication_protocol.md
│   ├── hardware_bom.md
│   ├── powertrain_sizing.md     # Powertrain sizing calculation (motor/battery/ESC)
│   ├── wiring_guide.md          # Wiring guide (ready-to-use version)
│   ├── firmware_todo.md         # Firmware to-do list (from simulation to real car)
│   ├── xiaozhi_integration.md   # Xiaozhi voice assistant integration plan
│   ├── development_notes.md     # Development notes and status archive
│   └── development_log.md
├── hardware/
│   ├── bom/               # BOM list
│   ├── circuit/           # Circuit design
│   └── pcb/               # PCB design
├── firmware/
│   └── arduino/           # Arduino firmware
├── software/
│   └── raspberry_pi/      # Raspberry Pi software
│       └── tools/         # Development aids (OCR image reading / CDP real screenshots)
├── cad/
│   └── body_design/       # 3D car body model
├── media/
│   ├── images/            # Images and video materials
│   ├── 宣传物料/           # Renderings/orthographic views/exploded views/promotional videos
│   ├── prototype/         # Prototype photos and videos
│   └── videos/            # Project videos
└── logs/
    └── 开发日记/           # Development process records

🖥️ Dual-End Digital Twin (v1, in development)

Uses real brain (Raspberry Pi) + cerebellum (Arduino) code to process real signals, with both ends sharing the same car. No 3D car model — dashboard + control area + parent area.

A process called TwinCore (real Brain + CerebellumSim)
 ├─ http://localhost:8000  Car end: dashboard + control area
 └─ http://localhost:8001  Parent end: GPS / battery / intercom / video (mobile-friendly)

How to run:

# One-click startup: dual-end digital twin + Xiaozhi assistant bridge
cd software/raspberry_pi
scripts/start_xiaozhi.bat            # Windows: simultaneously starts twin_server + xiaozhi_bridge

# Or manually start separately:
.venv/Scripts/python.exe digital_twin/backend/twin_server.py 8000 8001
py-xiaozhi/.venv/Scripts/python.exe digital_twin/backend/xiaozhi_bridge.py 8010
# Open browser to http://localhost:8000 (car end) and http://localhost:8001 (parent end)
# Access parent end from phone: use the computer's local network IP on the same Wi-Fi, e.g., http://192.168.x.x:8001

The dual-end digital twin web pages are fully open-sourced on GitHub (software/raspberry_pi/digital_twin/), with code available in Development Log Day 4.

🚀 Quick Start

The project is currently in the V0.1 prototype development stage. You can first read the Product Requirements Document to understand the complete functional planning of Ganbei No. 1, or start learning about technical details from the System Architecture, Communication Protocol, and Development Log.

Want to participate in joint development? Please first read the Contribution Guide and Code of Conduct.

🧩 Hardware Plan

Current experimental platform:

📖 Development Log

This project will document the entire process, including:

• Learning electronic control from scratch;
• First time making Raspberry Pi and Arduino communicate;
• First time controlling a motor;
• First time designing a car body;
• First time letting a child drive a car they built themselves.

The development process itself is also part of the project.

Latest update: Development Log Day 7 — From simulation to real car: car route planning + power selection + wiring scheme

🤝 Welcome to Get Involved

OpenKidCar welcomes everyone to participate.

No matter if you excel in:

• AI development
• Python
• Arduino
• Embedded systems
• Circuit design
• Mechanical design
• 3D modeling
• Product design

You are welcome to join.

Before contributing, please read the Contribution Guide and Code of Conduct. If you discover a security issue, please privately report it as described in SECURITY.md.

You don’t need to be an expert.

Because this project itself is an exploration of:

How ordinary people can learn and create with the help of AI.

❤️ To My Child

Perhaps many years from now,

you won’t remember this car.

But I hope you remember:

your dad once tried hard to create a work that belongs to you.

This world can not only be bought.

It can also be created.

📜 License

This project is licensed under the MIT License, a permissive open-source license that allows free use, modification, and distribution, including for commercial purposes, provided that the copyright notice is retained.