/FNI-RL

This repo is the implementation of "Fear-Neuro-Inspired Reinforcement Learning for Safe Autonomous Driving".

FNI-RL

This repo is the implementation of "Fear-Neuro-Inspired Reinforcement Learning for Safe Autonomous Driving".

The code will be released soon.

Demonstration

The video demonstration of our work can be found: Please click to watch the video.

In all demonstrations, the red-colored car represents the FNI-RL-driven autonomous vehicle.

1. Unprotected left turn at an unsignalized intersection with oncoming traffic

Scenario (a)

2. Right turn at an unsignalized intersection with crossing traffic

Scenario (b)

3. Unprotected left turn at an unsignalized intersection with mixed traffic flows

Scenario (c)

4. Crossing negotiation at an unsignalized intersection with mixed traffic flows

Scenario (d)

5. Long-term goal-driven navigation with mixed traffic flows

Scenario (e1)

Scenario (e2)

Installation

This repo is developed using Python 3.7 and PyTorch 1.3.1+CPU in Ubuntu 16.04.

We utilize the proposed FNI-RL approach to train the autonomous driving agent in the popular Simulation of Urban Mobility (SUMO, Version 1.2.0) platform.

We believe that our code can also run on other operating systems with different versions of Python, PyTorch and SUMO, but we have not verified it.

The required packages can be installed using

pip install -r requirements.txt