R2E: A Real-Time Routing and Recharging Recommendation System for Electric Taxi Drivers
-
2018-11-12
-
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report 09.01.2017-12.17.2018
-
Corporate Publisher:
-
Abstract:Electric taxi (eTaxi) has been introduced into the public transportation systems to accelerate the greening of industrial firms all over the world. Different from traditional taxis that can refuel in minutes, eTaxis’ recharging cycles can be as long as one hour. Currently, most of existing taxi recommendation systems focus on maximizing immediate reward from picking up next passengers. However, in the real world, the reward of a taxi driver is strongly correlated with the effective driving hours, especially for eTaxi drivers who are suffering from long recharging cycles. Therefore, how to maximize total reward of the eTaxi driver and when, where and how long to recharge an eTaxi have already emerged as urgent and crucial problems to be solved for the widely deployment of the eTaxi. To combat these problems, we propose a real-time routing and recharging recommendation system, called R2E, to recommend an entire route with recharging plan to maximize total reward, which considers both immediate reward from picking up next passengers and future reward from the rest of working hours. In this paper, we first formulate the serving and recharging problem of an eTaxi as a Markov Decision Process and leverage MDP to compute future reward in the rest of working hours. Then, based on immediate reward computed from historical data and future reward obtained from computational results of MDP, we propose a maximized reward routing algorithm, called MARA, to provide a fine-grained entire route including both finding next passengers and recharging plan in order to achieve maximum reward. Lastly, we evaluate our R2E on a real-world data set collected from the Shenzhen City in China. The experimental results clearly validate the effectiveness of our proposed R2Esystem.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type: