Automated Speed and Lane Change Decision Making using Deep Reinforcement Learning
Paper i proceeding, 2018

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.
https://arxiv.org/abs/1803.10056

Artificial Intelligence (cs.AI)

Robotics (cs.RO)

Machine Learning (cs.LG)

Författare

Carl-Johan E Hoel

Chalmers, Mekanik och maritima vetenskaper, Fordonsteknik och autonoma system

Volvo Group

Krister Wolff

Chalmers, Mekanik och maritima vetenskaper, Fordonsteknik och autonoma system

Leo Laine

Chalmers, Mekanik och maritima vetenskaper

Volvo Group

IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC

Vol. 2018-November 2148-2155 8569568
978-172810323-5 (ISBN)

21st IEEE International Conference on Intelligent Transportation Systems (ITSC)
Maui, Hawaii, USA,

Styrkeområden

Informations- och kommunikationsteknik

Transport

Drivkrafter

Hållbar utveckling

Innovation och entreprenörskap

Ämneskategorier

Data- och informationsvetenskap

Datavetenskap (datalogi)

Fundament

Grundläggande vetenskaper

DOI

10.1109/ITSC.2018.8569568

Mer information

Senast uppdaterat

2024-07-17