Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function

Jialin Hao; Rola Naja; Djamal Zeghlache

doi:10.1109/WiMob55322.2022.9941534

Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function

Jialin Hao
, Rola Naja
, Djamal Zeghlache

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper provides a Lane Change Assistance (LCA) platform that communicates with Unmanned Aerial Vehi-cles (UAV). The proposed platform is based on a reinforcement learning technique, where a Deep Q-Network (DQN) is trained to make lane change decisions. The reward function of the DQN agent considers safety, comfort and efficiency perspectives. Specifically, the safety reward, based on the road vehicular density, is adapted dynamically by the drone during the training phase. Performance analysis proves that the proposed platform improves the total travel time while reducing the collision rate and responding to urgent lane changes in a timely manner.

Original language	English
Title of host publication	2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022
Publisher	IEEE Computer Society
Pages	314-320
Number of pages	7
ISBN (Electronic)	9781665469753
DOIs	https://doi.org/10.1109/WiMob55322.2022.9941534
Publication status	Published - 1 Jan 2022
Event	18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022 - Thessaloniki, Greece Duration: 10 Oct 2022 → 12 Oct 2022

Publication series

Name	International Conference on Wireless and Mobile Computing, Networking and Communications
Volume	2022-October
ISSN (Print)	2161-9646
ISSN (Electronic)	2161-9654

Conference

Conference	18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022
Country/Territory	Greece
City	Thessaloniki
Period	10/10/22 → 12/10/22

Keywords

dynamic reward function
lane change
reinforcement learning

Access to Document

10.1109/WiMob55322.2022.9941534

Cite this

Hao, J., Naja, R., & Zeghlache, D. (2022). Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function. In 2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022 (pp. 314-320). (International Conference on Wireless and Mobile Computing, Networking and Communications; Vol. 2022-October). IEEE Computer Society. https://doi.org/10.1109/WiMob55322.2022.9941534

Hao, Jialin ; Naja, Rola ; Zeghlache, Djamal. / Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function. 2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022. IEEE Computer Society, 2022. pp. 314-320 (International Conference on Wireless and Mobile Computing, Networking and Communications).

@inproceedings{6fc052f8fb1e4c6f86a2c236950a6932,

title = "Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function",

abstract = "This paper provides a Lane Change Assistance (LCA) platform that communicates with Unmanned Aerial Vehi-cles (UAV). The proposed platform is based on a reinforcement learning technique, where a Deep Q-Network (DQN) is trained to make lane change decisions. The reward function of the DQN agent considers safety, comfort and efficiency perspectives. Specifically, the safety reward, based on the road vehicular density, is adapted dynamically by the drone during the training phase. Performance analysis proves that the proposed platform improves the total travel time while reducing the collision rate and responding to urgent lane changes in a timely manner.",

keywords = "dynamic reward function, lane change, reinforcement learning",

author = "Jialin Hao and Rola Naja and Djamal Zeghlache",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022 ; Conference date: 10-10-2022 Through 12-10-2022",

year = "2022",

month = jan,

day = "1",

doi = "10.1109/WiMob55322.2022.9941534",

language = "English",

series = "International Conference on Wireless and Mobile Computing, Networking and Communications",

publisher = "IEEE Computer Society",

pages = "314--320",

booktitle = "2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022",

}

Hao, J, Naja, R & Zeghlache, D 2022, Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function. in 2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022. International Conference on Wireless and Mobile Computing, Networking and Communications, vol. 2022-October, IEEE Computer Society, pp. 314-320, 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022, Thessaloniki, Greece, 10/10/22. https://doi.org/10.1109/WiMob55322.2022.9941534

Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function. / Hao, Jialin; Naja, Rola; Zeghlache, Djamal.
2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022. IEEE Computer Society, 2022. p. 314-320 (International Conference on Wireless and Mobile Computing, Networking and Communications; Vol. 2022-October).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function

AU - Hao, Jialin

AU - Naja, Rola

AU - Zeghlache, Djamal

PY - 2022/1/1

Y1 - 2022/1/1

N2 - This paper provides a Lane Change Assistance (LCA) platform that communicates with Unmanned Aerial Vehi-cles (UAV). The proposed platform is based on a reinforcement learning technique, where a Deep Q-Network (DQN) is trained to make lane change decisions. The reward function of the DQN agent considers safety, comfort and efficiency perspectives. Specifically, the safety reward, based on the road vehicular density, is adapted dynamically by the drone during the training phase. Performance analysis proves that the proposed platform improves the total travel time while reducing the collision rate and responding to urgent lane changes in a timely manner.

AB - This paper provides a Lane Change Assistance (LCA) platform that communicates with Unmanned Aerial Vehi-cles (UAV). The proposed platform is based on a reinforcement learning technique, where a Deep Q-Network (DQN) is trained to make lane change decisions. The reward function of the DQN agent considers safety, comfort and efficiency perspectives. Specifically, the safety reward, based on the road vehicular density, is adapted dynamically by the drone during the training phase. Performance analysis proves that the proposed platform improves the total travel time while reducing the collision rate and responding to urgent lane changes in a timely manner.

KW - dynamic reward function

KW - lane change

KW - reinforcement learning

U2 - 10.1109/WiMob55322.2022.9941534

DO - 10.1109/WiMob55322.2022.9941534

M3 - Conference contribution

AN - SCOPUS:85142713035

T3 - International Conference on Wireless and Mobile Computing, Networking and Communications

SP - 314

EP - 320

BT - 2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022

PB - IEEE Computer Society

T2 - 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022

Y2 - 10 October 2022 through 12 October 2022

ER -

Hao J, Naja R, Zeghlache D. Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function. In 2022 18th International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob 2022. IEEE Computer Society. 2022. p. 314-320. (International Conference on Wireless and Mobile Computing, Networking and Communications). doi: 10.1109/WiMob55322.2022.9941534

Drone-Assisted Lane Change Maneuver using Reinforcement Learning with Dynamic Reward Function

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this