Look-Ahead Insertion Policy for a Shared-Taxi System Based on Reinforcement Learning

Wei, Chong; Wang, Yinhu; Yan, Xuedong; Shao, Chunfu

doi:10.1109/access.2017.2769666

Cited by 26 publications

(25 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Rivera et al introduced an approximate dynamic programming method with the probabilistic knowledge on future requests for freight selection [21]. Wei et al proposed a reinforcement learning method taking the uncertainty of future requests into account [22]. The method made a look-ahead decision to improve the service quality of a shared-taxi system.…”

Section: A Review Of Demand Prediction Methodsmentioning

confidence: 99%

A Spatiotemporal Thermo Guidance Based Real-Time Online Ride-Hailing Dispatch Framework

Guo

Zhang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Online ride-hailing platforms can gather travel requests and allocate service vehicles to balance transportation demands and supplies, which may result in an increase in the utilization rate of service resources and improve the transportation efficiency and social welfare. An effective and flexible dispatching strategy can significantly reduce the waiting time of passengers and increase the profit of service vehicles. In this paper, we propose a real-time service vehicle dispatching framework in the context of large-scale online ride-hailing, which considers all the main issues involved in the problem in a unified and integrated way. Firstly, we formally define the problem and introduce a novel mathematical model based on the spatiotemporal thermo to guide the dispatching process so as to achieve a better balance of demands and supplies. Secondly, we propose an effective machine learning method to predict the spatiotemporal thermo based on the historical data of multi-dimensional factors. Afterward, decision rules to dynamically segment timeframes and a Kuhn-Munkres based algorithm are developed to solve the dispatching problem. Finally, extensive experiments on large-scale real-world instances are conducted to verify the effectiveness of the proposed framework. INDEX TERMS transportation, vehicle dispatching, online ride-hailing, spatiotemporal thermo, dynamic timeframe = , and is on duty during time range [ = , = ]. We use = to denote the vehicle's current service status. = equals 1 when the vehicle is assigned to serve a request, and otherwise equals 0. Definition 4 (Request): A request of a passenger, represented by = < B , B , B >, submitted at time B , indicates a transportation need from the origin node B to the destination node B. A request will be canceled by the passenger if it is not served after a fixed time D. Definition 5 (Service-pair): When a service vehicle is assigned to answer a request , a service-pair < , > is created. Note that we consider the circumstance that a vehicle only serves one request at a time.

show abstract

Section: A Review Of Demand Prediction Methodsmentioning

confidence: 99%

A Spatiotemporal Thermo Guidance Based Real-Time Online Ride-Hailing Dispatch Framework

Guo

Zhang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…One alternative is to model each available vehicle as an agent [21, Forthcoming Order Figure 1: Ride-hailing task in thermodynamics view. 37,39]. However, such setting needs to maintain thousands of agents interacting with the environment, which brings a huge computational cost.…”

Section: Introductionmentioning

confidence: 99%

“…Greedily matching vehicles with long-distance orders can receive high immediate gain at a single order dispatching stage, but it would harm order response rate (ORR) and future revenue especially during rush hour because of its long drive time and unpopular destination. Recent attempts [21,37,39] deployed RL to combine instant order reward from online planning with future state-value as the final matching value. However, the coordination between different regions is still far from optimal.…”

Section: Introductionmentioning

confidence: 99%

CoRide

Jin

Zhou

Zhang

et al. 2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

How to optimally dispatch orders to vehicles and how to trade off between immediate and future returns are fundamental questions for a typical ride-hailing platform. We model ride-hailing as a large-scale parallel ranking problem and study the joint decisionmaking task of order dispatching and fleet management in online ride-hailing platforms. This task brings unique challenges in the following four aspects. First, to facilitate a huge number of vehicles to act and learn efficiently and robustly, we treat each region cell as an agent and build a multi-agent reinforcement learning framework. Second, to coordinate the agents from different regions to achieve long-term benefits, we leverage the geographical hierarchy of the region grids to perform hierarchical reinforcement learning. Third, to deal with the heterogeneous and variant action space for joint order dispatching and fleet management, we design the action as the ranking weight vector to rank and select the specific order or the fleet management destination in a unified formulation. Fourth, to achieve the multi-scale ride-hailing platform, we conduct the decision-making process in a hierarchical way where a multihead attention mechanism is utilized to incorporate the impacts of neighbor agents and capture the key agent in each scale. The whole novel framework is named as CoRide. Extensive experiments based on multiple cities real-world data as well as analytic synthetic data demonstrate that CoRide provides superior performance in terms of platform revenue and user experience in the task of citywide hybrid order dispatching and fleet management over strong baselines.

show abstract

“…The system allows several people to share the capacity 242 ISBN: 978-81-936820-0-5 Proceedings DOI: 10.21467/proceedings.1 Series: AIJR Proceedings of the same vehicle if their destination routes are on the same route. Therefore, the system can use less vehicles to satisfy the requests and dynamically respond to the travel demand [1]. The proposed design in this paper shows the daily operation of each individual taxi driver and makes the profit, distribution of profit among drivers is fair and equitable without imposing any additional cost on passengers.…”

Section: Introductionmentioning

confidence: 99%

“…With the development of smart phone commercial applications of shared-taxi systems such as UberPool, Ola, LiftLine, Split and DiDiXiaoba came into existence around the world. These systems benefit both urban traffic and travelers around the world [1]. Though today's taxi system is far from being efficient: a most common problem is that people often have difficulties in getting a taxi during rush hours and weekends where there will be more demand for taxis, though the occupied taxis might still have available seats which are taken by one or two passengers per taxi.…”

Section: Introductionmentioning

confidence: 99%

A Survey on the Implementation of Reinforcement Learning on Shared Taxi System

Madhurya¹,

Karthik²

2018

Nciccnda

View full text Add to dashboard Cite

Reinforcement learning (RL) is an area of machine learning concerned with how software agents used to take actions in a nature so as to maximize some faith of increasing benefit. In machine learning, the nature is typically formulated as a Markov decision process (MDP), as many of the reinforcement learning algorithms for this context utilize active programming techniques. Taxis plays a important role in modern public transportation networks, especially in countries where public transportation services are still underdeveloped. Taxi drivers currently rely on a simple first come-first serve approach, with a high coefficient of luck controlling their profit Problems arising from the current system are low taxi utilization, long passenger waiting times, road safety issues, and traffic congestion. In this paper, we present a system that improves the shared taxi service by benefiting both the taxi drivers and the passengers. This study proposed a various reinforcement learning method for a shared-taxi system. The present system proposes a solution in the form of a good, efficient and economical transportation service.

show abstract

Look-Ahead Insertion Policy for a Shared-Taxi System Based on Reinforcement Learning

Cited by 26 publications

References 28 publications

A Spatiotemporal Thermo Guidance Based Real-Time Online Ride-Hailing Dispatch Framework

A Spatiotemporal Thermo Guidance Based Real-Time Online Ride-Hailing Dispatch Framework

CoRide

A Survey on the Implementation of Reinforcement Learning on Shared Taxi System

Contact Info

Product

Resources

About