Simulating self-driving 🚗 in javascript

Abstract:

I made this project to demonstrate learning to self-drive a car using Neural network. To stimulate the vehicle I used canvas_API and connected the controls with a neural network. The neural network had 5 inputs and 4 outputs.The changing of weights is visualized in the application to visualize how network works.The Network has to be manually trained and the optimum version can be saved and reused. The use of this project lies in demonstrating how Neural networks learn to control a vehicle.

---

Introduction:

The project is a implementation of car which is driven by a neural network, the network controls the car based on signal inputs and weights. To implement the car i used javascript. In javascript i used cavasAPI and basic mathematical functions.

---

Process:

- 1.Car driving mechanism
- 2.Road Definition
- 3.Artificial Sensors
- 4.Collision Detection
- 5.Traffic Simulation
- 6.Neural Networks
- 7.Visualizing NN
- 8.Optimizing NN
- 9.Fine Tuning

Video:

car-dr.mp4

Basic working of car

final-demo.mp4

Final demonstration of car

---

Images:

Final demo of vehicle

---

Diagrams:

Diagramtic representation of Car and NN ⬆️

---

Explanation:

Autonomous.Car.Learns.to.Navigate.1.mp4

Formulaes:

---

References:

1. Canvas_api
2. canvas rendering

---

@Contributors:

Extras:

• use excaldraw extension to read the excalidraw.png

Flow:

How to Run:

• fork this repo
  • git clone https://github.com/MimoHasPurpose/autonomous-car.git
• install live-server extension
• open index.html file
• run this file using live-server extension

---

References

• Autonomous Vehicle
• Challanges to autonomous vehicle
• autonomous planes and other types of vehicles which can be automated ,
• Future of autonomous vehicle