Skip to content

Latest commit

 

History

History
81 lines (66 loc) · 3.84 KB

README.md

File metadata and controls

81 lines (66 loc) · 3.84 KB

Autonomous Systems Simulation

Table of Contents

Introduction

This project simulates autonomous car systems using a game environment called City Car Gaming, known for its realistic mechanics and graphics, sufficient to track objects and simulate real-world scenarios.

Features

Object Tracking

object_2object_2 The system tracks various objects in the game environment using yolov8n model.

Bird's-eye View Mapping

object_2object_2 Transforms the front camera view into a bird's-eye view to map surroundings and predict the next path of objects. Currently, it is able to:

  • Show a bird's-eye view for close distance objects
  • Warn about possible collisions based on short-range predictions but only in straight road for now

Line Detection

object_2object_2 Detects road lines using a sliding window technique, which assists in maintaining lane integrity.

Installation

git clone https://github.com/elymsyr/autonomous-systems-simulation.git
pip install -r requirements.txt
python run_simulation.py

Future Plans

Improved Bird's-eye View

Enhancing the bird's-eye view system to generate a more detailed environmental map and implement car pose detection. The goal is to provide accurate 3D terrain detection and car orientation.

Improved Next Path Prediction

Improving object path prediction to handle long-range scenarios and diverse object movement. The goal is to develop more accurate trajectory predictions for moving objects.

Navigator

This system will provide self-driving capabilities by creating a path for the car to follow autonomously. The plan includes path planning algorithms and integration with the object detection module.

Collision Avoidance System

Introducing a system to prevent collisions using real-time object detection and prediction. Future implementations might include autobraking or steering interventions.

Contributing

Contributions are welcome. Please check the issues tab and submit pull requests. See .github

License

See the LICENSE file for details.

Acknowledgments

I would like to thank the following resources and projects that inspired or contributed to features:

    @article{gosala21bev,
    author={Gosala, Nikhil and Valada, Abhinav},
    journal={IEEE Robotics and Automation Letters}, 
    title={Bird’s-Eye-View Panoptic Segmentation Using Monocular Frontal View Images}, 
    year={2022},
    volume={7},
    number={2},
    pages={1968-1975},
    doi={10.1109/LRA.2022.3142418}}