This is a self-driving car simulation implemented purely in JavaScript, HTML, and CSS — no external libraries are used. The project demonstrates AI-controlled cars navigating a road while avoiding obstacles and following lanes using simple neural networks.
- AI-controlled cars using neural networks
- Sensor-based collision detection
- Multiple lanes and realistic road simulation
- Traffic (dummy cars) for AI cars to avoid
- Real-time visualization of the neural network
- Gradient and styling improvements for a better UI
- Cars:
- AI cars (blue) are controlled by neural networks.
- Dummy cars (red) move along lanes as obstacles.
- Each car has a sensor system to detect road borders and other cars.
- Road:
- Divided into 3 lanes (default), with dashed lane markers.
- Styled to resemble a real road with gradient effect.
- Network Visualization:
- Displays neural network of the best car in real-time.
- Gradient lines for a more appealing look.
| File | Description |
|---|---|
index.html |
Main HTML structure with canvas elements and buttons. |
styles.css |
CSS styling for canvases, buttons, and layout. |
main.js |
Simulation setup, animation loop, and AI logic. |
car.js |
Car class (AI & dummy), movement, and drawing. |
road.js |
Road class and lane calculation. |
controls.js |
Keyboard control and AI output mapping. |
sensor.js |
Sensor system for cars to detect obstacles. |
network.js |
Neural network logic and feed-forward calculations. |
visualizer.js |
Visualization of the neural network. |
utils.js |
Utility functions like intersection detection and RGBA conversion. |
- Contains stable version 1 of the simulation.
- Basic AI-controlled cars and traffic.
- Basic neural network visualization.
- Contains Version 2 improvements:
- Better road styling with gradients and lanes.
- Improved network visualization styling.
- Added multiple demo cars as obstacles.
- Buttons for saving/discarding the best AI brain on the left side.
- More appealing UI with shadows and rounded corners.
- Reserved for experimentation and training:
- You can test new AI algorithms here.
- Does not affect the main or test branch unless merged.
- Clone the repository:
git clone https://github.com/Samprit74/self-driven-car-simulation-without-library.git
Open index.html in a web browser (Chrome recommended).
Use the save 💾 and discard 🗑️ buttons to manage the best AI car brain.
Watch AI-controlled cars learn to navigate the road in real-time!
⚙️ Libraries
No external libraries are used.
Pure JavaScript, HTML, and CSS.
Neural networks and visualization are implemented manually in network.js and visualizer.js.
🌟 Future Improvements
Train AI with different neural network architectures.
Add more dynamic obstacles and road layouts.
Add stats dashboard for AI performance.