Autonomous Car Programming for AWS DeepRacer

This project focuses on training an autonomous vehicle using AWS DeepRacer. AWS DeepRacer allows you to get hands-on with reinforcement learning, an advanced machine learning technique. The project involves setting up the simulation environment, training the model, and evaluating its performance. I have presented and effectively communicated complex technical concepts and results to both technical and non-technical audiences in multiple symposiums. Two videos which show the training phase and testing phase can be viewed at the bottom.

Project Details

Technologies Used: AWS DeepRacer, Python, Reinforcement Learning
Key Features:

• Simulation setup in AWS DeepRacer console
• Custom reward function development
• Training and evaluating the reinforcement learning model
• Deployment for real-world testing

Some of the Code

# Custom Reward Function for AWS DeepRacer
def reward_function(params):
    # Read input parameters
    track_width = params['track_width']
    distance_from_center = params['distance_from_center']
    speed = params['speed']
    steering_angle = abs(params['steering_angle'])

    # Define reward and penalties
    reward = 1.0
    if distance_from_center <= track_width * 0.1:
        reward += 1.0
    elif distance_from_center <= track_width * 0.25:
        reward += 0.5
    else:
        reward -= 1.0

    # Penalize if the car is steering too much
    if steering_angle > 15:
        reward *= 0.8

    # Reward higher speeds
    if speed > 2.0:
        reward += 1.0

    return float(reward)

            

Poster

Videos