Autonomous Car Simulation (TensorFlow)

A virtual traffic environment for training autonomous cars using reinforcement learning.The agent learns to navigate dynamic environments with moving obstacles, achieving successful obstacle avoidance.

Demo:

with static obstacles:

with dynamic (moving) obstacles:

Features:

• Designed with TensorFlow for efficient training.
• Utilizes reinforcement learning algorithms (e.g., Q-Learning) for autonomous car behavior optimization.
• Simulates dynamic traffic environments with customizable numbers of moving obstacles for obstacle avoidance training.

Dependencies:

Library	Description
P5.js	A JavaScript library for creative coding. Used in this project for easy visualization of the virtual traffic environment.
TensorFlow.js	A JavaScript library enabling machine learning on the web browser. Used for implementing reinforcement learning algorithms for the autonomous car agent.
Collide2d.js	A library for efficient collision detection between objects in 2D space. Useful for accurately simulating collisions between the car and obstacles.