Automated Speed and Lane Change Decision Making using Deep Reinforcement Learning

TL;DR

This paper presents a deep reinforcement learning approach using a Deep Q-Network to automate speed and lane change decisions in highway driving, demonstrating its effectiveness and generality across different scenarios.

Contribution

The paper introduces a novel deep reinforcement learning method with a convolutional neural network for high-level decision making in driving tasks, applicable to multiple scenarios.

Findings

The agent matched or outperformed a reference model in highway driving.

The same algorithm successfully generalized to overtaking scenarios with oncoming traffic.

A new convolutional neural network approach was proposed for high-level input representation.

Abstract

This paper introduces a method, based on deep reinforcement learning, for automatically generating a general purpose decision making function. A Deep Q-Network agent was trained in a simulated environment to handle speed and lane change decisions for a truck-trailer combination. In a highway driving case, it is shown that the method produced an agent that matched or surpassed the performance of a commonly used reference model. To demonstrate the generality of the method, the exact same algorithm was also tested by training it for an overtaking case on a road with oncoming traffic. Furthermore, a novel way of applying a convolutional neural network to high level input that represents interchangeable objects is also introduced.