Skip to content

Deep learning-based emergency braking system built with CARLA simulator

License

Notifications You must be signed in to change notification settings

ethanmclark1/carla_aebs

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

52 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Autonomous Emergency Braking System

Overview

The Autonomous Emergency Braking System described here integrates two neural networks for enhanced vehicle safety and lane adherence:

1. Convolutional Neural Network (CNN) for Lane Following:

  • This network is responsible for determining the steering angle.
  • The output is based on visual inputs, ensuring the vehicle remains centered within its lane.

2. Dueling Deep Q-Network (Dueling DQN):

  • This network receives the steering angle from the CNN, combined with data on velocity and distance.
  • It then computes Q-values for potential actions.
  • The action with the highest Q-value is selected, guiding the vehicle's next move.

Together, these networks enable the vehicle to autonomously apply brakes as needed, preventing collisions with vehicles ahead while ensuring consistent lane positioning.

Setup Environment

Run AEBS

  • Open a terminal and enter command: ./CarlaUE.sh -opengl to run CARLA simulator
  • Activate conda environment in another terminal then enter command: python3 driver.py
  • To visualize the agent in action, open a third terminal and enter the command: sudo docker run -it --network="host" -e CARLAVIZ_HOST_IP=localhost -e CARLA_SERVER_IP=localhost -e CARLA_SERVER_PORT=2000 mjxu96/carlaviz:0.9.11 then open a browser and go to localhost: 127.0.0.1:8080/

References

  1. Chae, Hyunmin, et al. "Autonomous braking system via deep reinforcement learning." 2017 IEEE 20th International conference on intelligent transportation systems (ITSC). IEEE, 2017.
  2. Bojarski, Mariusz, et al. "End to end learning for self-driving cars." arXiv preprint arXiv:1604.07316 (2016).
  3. Wang, Ziyu, et al. "Dueling network architectures for deep reinforcement learning." International conference on machine learning. PMLR, 2016.