Paper Image

Reinforcement learning for autonomous vehicle lane changing

Published on:

25 September 2023

Primary Category:

Machine Learning

Paper Authors:

Emanuel Figetakis,

Yahuza Bello,

Ahmed Refaey,

Lei Lei,

Medhat Moussa

Bullets

Key Details

Proposes integrated car-following and lane-changing model using deep reinforcement learning

Models scenario of highway construction as Markov decision process

Uses Deep Q-Network algorithm trained on edge server to optimize decisions

Evaluates model in simulation using epsilon-greedy and Boltzmann policies

Epsilon-greedy policy learns optimal actions for dynamic environment

AI generated summary

Reinforcement learning for autonomous vehicle lane changing

This paper proposes a reinforcement learning framework for autonomous vehicles to make optimal lane changing decisions in response to sudden highway construction. The scenario is modeled as a Markov decision process and trained using a Deep Q-Network algorithm on a multi-access edge computing server. The model is evaluated in simulation using two policies, epsilon-greedy and Boltzmann, with epsilon-greedy significantly outperforming for this dynamic environment.

Answers from this paper

Comments

No comments yet, be the first to start the conversation...

Sign up to comment on this paper

Sign Up