E-scooter sharing lets people rent an e-scooter while the system owner manages the fleet. Relocation is fundamental to increase system utilization and revenues, but it is also an expensive task. In this paper we aim at assessing the benefits of relocation while quantifying its economic costs. For this, we rely on trace driven simulations where we build upon millions of actual rentals from two cities, Austin and Louisville. Firstly, we build prediction models to estimate which areas will present a surplus or a lack of e-scooters. We compare a simple stationary model with a state-of-art deep-learning model. Secondly, we replay the exact same traces to quantify the benefits of a relocation heuristic, comparing different system options. Our results show that relocation is fundamental to maximize the number of trips the system can satisfy. Interestingly, even a light and simple relocation policy with few relocations per hour can improve the percentage of satisfied trips by up to 42%. This can also translate in a fleet size reduction without impacting the performances. However, when projected into the economic benefits, the additional costs of relocation must be carefully considered to avoid wasting its benefits.

Benefits of Relocation on E-scooter Sharing - a Data-Informed Approach / Tolomei, Leonardo; Fiorini, Stefano; Ciociola, Alessandro; Vassio, Luca; Giordano, Danilo; Mellia, Marco. - ELETTRONICO. - (2021), pp. 3170-3175. ((Intervento presentato al convegno 2021 IEEE International Intelligent Transportation Systems Conference (ITSC) [10.1109/ITSC48978.2021.9564809].

Benefits of Relocation on E-scooter Sharing - a Data-Informed Approach

Ciociola, Alessandro;Vassio, Luca;Giordano, Danilo;Mellia, Marco
2021

Abstract

E-scooter sharing lets people rent an e-scooter while the system owner manages the fleet. Relocation is fundamental to increase system utilization and revenues, but it is also an expensive task. In this paper we aim at assessing the benefits of relocation while quantifying its economic costs. For this, we rely on trace driven simulations where we build upon millions of actual rentals from two cities, Austin and Louisville. Firstly, we build prediction models to estimate which areas will present a surplus or a lack of e-scooters. We compare a simple stationary model with a state-of-art deep-learning model. Secondly, we replay the exact same traces to quantify the benefits of a relocation heuristic, comparing different system options. Our results show that relocation is fundamental to maximize the number of trips the system can satisfy. Interestingly, even a light and simple relocation policy with few relocations per hour can improve the percentage of satisfied trips by up to 42%. This can also translate in a fleet size reduction without impacting the performances. However, when projected into the economic benefits, the additional costs of relocation must be carefully considered to avoid wasting its benefits.
978-1-7281-9142-3
File in questo prodotto:
File Dimensione Formato  
ITSC____Scooter_Relocation.pdf

accesso aperto

Descrizione: Versione finale
Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 1.25 MB
Formato Adobe PDF
1.25 MB Adobe PDF Visualizza/Apri
Benefits_of_Relocation_on_E-scooter_Sharing_-_a_Data-Informed_Approach.pdf

non disponibili

Descrizione: Versione Pubblicata
Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 220.56 kB
Formato Adobe PDF
220.56 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2934928