JetRacer Autonomous Driving

Description

Autonomous driving for Final Degree Project in Robotics.

Software

The employed architecture of convolutional neural network (CNN) is YOLOv8n-seg. This model has been trained with a dataset of 500 images of the following circuit (Figure 1). The achieved metrics of this model are:

Development

List TODO:

• Test JetRacer motors

• Keyboard Control JetRacer

• Gamepad Control JetRacer

• Create own dataset.

• Train YOLO.

• Lane segmentation.

• Object segmentation.

• Car follower.

• Security stop.

• Implement EKF.

License

JetRacer Autonomous Driving is available under next license:

• AGPL-3.0 License: See LICENSE file for details.

Author:

(c) 2023 (Marcos Fernández)