Skip to content

POSTECH-IMLAB/PIMNET_Autonomous_Driving

 
 

Repository files navigation

carla_yaw

End-to-end Autonomous Driving System using yaw information.

We implement a vision-based autonomous driving mode using yaw and localization based End-to-end autonomous driving model with pytorch.

Implemented Version

  • Python 3.6
  • Pytorch 1.3
  • Torchvision 0.2
  • CARLA 0.8.2

Kinematic Vehicle Equation

We utilize the bicycle model form kinematic vehicle dynamics. The yaw angle is calculated by Kinematic Vehicle Equation.

How to Run

  • Train: python main_auxi_v0.py --lr=1e-4 --train-dir=your dir --eval-dir=your dir
  • Eval: python run_auxi_ver0.py --model-path=trained_model.pth

CARLA benchmark

The CARLA simulator has a large variety of driving environments, such as traff ic lights and dynamic obstacles, including dynamic vehicles and pedestrians.

You can download the CARLA simulator from here (benchmark).

licence : LGPL 2.1

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%