Abstract:On-demand ride services reshape urban transportation systems, human mobility, and travelers' mode choice behavior. Compared to the traditional street-hailing taxi, an on-demand ride services platform analyzes ride requests of passengers and coordinates real-time supply and demand with dynamic operational strategies in the ride-sourcing market. To test the impact of dynamic optimization strategies on the ride-sourcing market, this paper proposes a dynamic vacant car-passenger meeting model. In this model, the accumulative arrival rate and departure rate of passengers and vacant cars determine the waiting number of passengers and vacant cars, while the waiting number of passengers and vacant cars in turn influence the meeting rate (which equals to the departure rate of both passengers and vacant cars). The departure rate means the rate at which passengers and vacant cars match up and start a paid trip. Compared with classic equilibrium models, this model can be utilized to characterize the influence of short-term variances and disturbances of current demand and supply (i.e., arrival rates of passengers and vacant cars) on the waiting numbers of passengers and vacant cars. Using the proposed meeting model, we optimize dynamic strategies under two objective functions, i.e., platform revenue maximization, and social welfare maximization, while the driver's profit is guaranteed above a certain level. We also propose an algorithm based on approximate dynamic programming (ADP) to solve the sequential dynamic optimization problem. The results show that our algorithm can effectively improve the objective function of the multi-period problem, compared with the myopic algorithm. A broader range of surge pricing and commission rate and the introduction of incentives are helpful to achieve better optimization results. The dynamic optimization strategies help the on-demand ride services platform efficiently adjust supply and demand resources and achieve specific optimization goals. (C) 2020 Elsevier Ltd. All rights reserved.

Deep Reinforcement Learning-based Trajectory Pricing on Ride-hailing Platforms

Spatial-temporal Pricing for Ride-Sourcing Platform with Reinforcement Learning

Dynamic Optimization Strategies for On-Demand Ride Services Platform: Surge Pricing, Commission Rate, and Incentives

Joint Optimization of Pricing, Dispatching and Repositioning in Ride-Hailing with Multiple Models Interplayed Reinforcement Learning

Seeking in Ride-on-Demand Service: A Reinforcement Learning Model with Dynamic Price Prediction

Supply-Demand-aware Deep Reinforcement Learning for Dynamic Fleet Management

Optimizing Drivers’ Revenue Efficiency for Ride-On-Demand Services: A Reinforcement Learning Approach with Dynamic Price Prediction

Spatio-temporal Incentives Optimization for Ride-hailing Services with Offline Deep Reinforcement Learning

A Distributed Model-Free Ride-Sharing Approach for Joint Matching, Pricing, and Dispatching using Deep Reinforcement Learning

Scalable Deep Reinforcement Learning for Ride-Hailing

Seeking Based on Dynamic Prices: Higher Earnings and Better Strategies in Ride-on-Demand Services

Reinforcement Learning For Optimizing Driving Policies On Cruising Taxis Services

Competitive pricing for ride-sourcing platforms with MARL

CoRLNF: Joint Spatio-Temporal Pricing and Fleet Management for Ride-Hailing Platforms

Dynamic Balancing-Charging Management for Shared Autonomous Electric Vehicle Systems: A Two-Stage Learning-Based Approach

Dynamic Pricing for Vehicle Dispatching in Mobility-as-a-Service Market Via Multi-Agent Deep Reinforcement Learning

Dynamic Pricing on E-commerce Platform with Deep Reinforcement Learning: A Field Experiment

Integrated ridesharing services with chance-constrained dynamic pricing and demand learning

Learning Dynamic Pricing Rules for Flight Tickets.

Dynamic Pricing in Shared Mobility on Demand Service

Deep Reinforcement Learning Algorithm for Dynamic Pricing of Express Lanes with Multiple Access Locations