Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection

<p dir="ltr">In the evolving urban transportation, the emergence of Micro-Mobility (MM), symbolized by Electric Scooters (ESs), has become a pivotal response to private automobiles’ environmental and logistical challenges. However, the limited battery capacity of ESs presents a chall...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Karim Aboeleneen (21841766) (author)
مؤلفون آخرون: Nizar Zorba (16888728) (author), Ahmed M. Massoud (16896417) (author)
منشور في: 2024
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513542684672000
author Karim Aboeleneen (21841766)
author2 Nizar Zorba (16888728)
Ahmed M. Massoud (16896417)
author2_role author
author
author_facet Karim Aboeleneen (21841766)
Nizar Zorba (16888728)
Ahmed M. Massoud (16896417)
author_role author
dc.creator.none.fl_str_mv Karim Aboeleneen (21841766)
Nizar Zorba (16888728)
Ahmed M. Massoud (16896417)
dc.date.none.fl_str_mv 2024-06-16T12:00:00Z
dc.identifier.none.fl_str_mv 10.1109/access.2024.3395286
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Reinforcement_Learning-Based_E-Scooter_Energy_Minimization_Using_Optimized_Speed-Route_Selection/29715254
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Engineering
Manufacturing engineering
Information and computing sciences
Artificial intelligence
Machine learning
Electric scooters
reinforcement learning
energy minimization
user dissatisfaction
route and speed selection
Roads
Energy consumption
Minimization
Vehicle dynamics
Temperature
Resistance
Mathematical models
Electric vehicles
Reinforcement learning
dc.title.none.fl_str_mv Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p dir="ltr">In the evolving urban transportation, the emergence of Micro-Mobility (MM), symbolized by Electric Scooters (ESs), has become a pivotal response to private automobiles’ environmental and logistical challenges. However, the limited battery capacity of ESs presents a challenge in realizing their full potential. This paper addresses the problem of optimizing energy consumption in ESs by jointly considering path and speed selection all while considering user dissatisfaction levels. Our approach considers two types of ESs, one with regenerative braking (i.e., able to recharge the battery from kinetic energy of movement) and the other without regenerative braking. In order to build a realistic environment, we considered dynamic factors such as time-varying road congestion, road conditions, and ambient temperature. We considered a comprehensive energy consumption model for the ES that includes parameters such as rolling resistance, air friction, road gradient, auxiliary power and ambient temperature influence. Moreover, we introduced a user dissatisfaction model that accounts for traffic conditions, congestion, and ambient temperature to enhance the user experience. The optimization problem was then formulated and solved with Deep Reinforcement Learning (DRL-DQN) approach considering the time-varying environment, road-specific parameters (i.e., road angle, road shading, road speed limit, and road condition), and user dissatisfaction levels. The DRL approach was designed to make timely and context-aware decisions the minimize the energy consumption of the ES. Rigorous validation and comprehensive testing demonstrate the effectiveness of our approach. We evaluated the proposed solution’s performance against alternative methodologies used by fleet operators in different tests, including energy consumption, average user dissatisfaction, and average trip duration. The results showed that the proposed approach saved nearly 53-67% of energy for regenerative braking cases and 25-55% for non-regenerative braking cases when compared with other approaches and offers high adaptability to the environment and less complexity when compared with the exhaustive solution.</p><h2>Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2024.3395286" target="_blank">https://dx.doi.org/10.1109/access.2024.3395286</a></p>
eu_rights_str_mv openAccess
id Manara2_cf965ddfcce1999e5c76d19bde7b4cfc
identifier_str_mv 10.1109/access.2024.3395286
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/29715254
publishDate 2024
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route SelectionKarim Aboeleneen (21841766)Nizar Zorba (16888728)Ahmed M. Massoud (16896417)EngineeringManufacturing engineeringInformation and computing sciencesArtificial intelligenceMachine learningElectric scootersreinforcement learningenergy minimizationuser dissatisfactionroute and speed selectionRoadsEnergy consumptionMinimizationVehicle dynamicsTemperatureResistanceMathematical modelsElectric vehiclesReinforcement learning<p dir="ltr">In the evolving urban transportation, the emergence of Micro-Mobility (MM), symbolized by Electric Scooters (ESs), has become a pivotal response to private automobiles’ environmental and logistical challenges. However, the limited battery capacity of ESs presents a challenge in realizing their full potential. This paper addresses the problem of optimizing energy consumption in ESs by jointly considering path and speed selection all while considering user dissatisfaction levels. Our approach considers two types of ESs, one with regenerative braking (i.e., able to recharge the battery from kinetic energy of movement) and the other without regenerative braking. In order to build a realistic environment, we considered dynamic factors such as time-varying road congestion, road conditions, and ambient temperature. We considered a comprehensive energy consumption model for the ES that includes parameters such as rolling resistance, air friction, road gradient, auxiliary power and ambient temperature influence. Moreover, we introduced a user dissatisfaction model that accounts for traffic conditions, congestion, and ambient temperature to enhance the user experience. The optimization problem was then formulated and solved with Deep Reinforcement Learning (DRL-DQN) approach considering the time-varying environment, road-specific parameters (i.e., road angle, road shading, road speed limit, and road condition), and user dissatisfaction levels. The DRL approach was designed to make timely and context-aware decisions the minimize the energy consumption of the ES. Rigorous validation and comprehensive testing demonstrate the effectiveness of our approach. We evaluated the proposed solution’s performance against alternative methodologies used by fleet operators in different tests, including energy consumption, average user dissatisfaction, and average trip duration. The results showed that the proposed approach saved nearly 53-67% of energy for regenerative braking cases and 25-55% for non-regenerative braking cases when compared with other approaches and offers high adaptability to the environment and less complexity when compared with the exhaustive solution.</p><h2>Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2024.3395286" target="_blank">https://dx.doi.org/10.1109/access.2024.3395286</a></p>2024-06-16T12:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1109/access.2024.3395286https://figshare.com/articles/journal_contribution/Reinforcement_Learning-Based_E-Scooter_Energy_Minimization_Using_Optimized_Speed-Route_Selection/29715254CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/297152542024-06-16T12:00:00Z
spellingShingle Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
Karim Aboeleneen (21841766)
Engineering
Manufacturing engineering
Information and computing sciences
Artificial intelligence
Machine learning
Electric scooters
reinforcement learning
energy minimization
user dissatisfaction
route and speed selection
Roads
Energy consumption
Minimization
Vehicle dynamics
Temperature
Resistance
Mathematical models
Electric vehicles
Reinforcement learning
status_str publishedVersion
title Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
title_full Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
title_fullStr Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
title_full_unstemmed Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
title_short Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
title_sort Reinforcement Learning-Based E-Scooter Energy Minimization Using Optimized Speed-Route Selection
topic Engineering
Manufacturing engineering
Information and computing sciences
Artificial intelligence
Machine learning
Electric scooters
reinforcement learning
energy minimization
user dissatisfaction
route and speed selection
Roads
Energy consumption
Minimization
Vehicle dynamics
Temperature
Resistance
Mathematical models
Electric vehicles
Reinforcement learning