Embedded Files
APP & Web applications related to smart mobility
APP & Web applications related to smart mobility
Our team has developed a variety of software toolbox and apps for enhancing the efficiency and sustainability of various kinds of smart mobility systems, including ride-hailing services, electric vehicle systems, and e-hailing taxi markets. For more details, please visit https://www.civil.hku.hk/kejintao/softwares.html
A large-scale taxi simulation platform
A large-scale taxi simulation platform
We propose a novel multi-functional and open-sourced simulation platform for ride-sourcing systems, which can simulate the behaviors and movements of various agents on a real transportation network. It provides a few accessible portals for users to train and test various optimization algorithms, especially reinforcement learning algorithms, for a variety of tasks, including on-demand matching, idle vehicle repositioning, and dynamic pricing.
Related Paper: Feng S., Chen T., Zhang Y., Ke J.* & H. Yang, 2023. A multi-functional simulation platform for on-demand ride service operations. Preprint at ArXiv:2303.12336. https://doi.org/10.48550/arXiv.2303.12336.
Open-sourced Code Link: https://github.com/HKU-Smart-Mobility-Lab/Transpotation_Simulator
An Taxi Idle Cruising APP
An Taxi Idle Cruising APP
It is found that 40-60% taxi time is spent on idle cruising for searching passengers, which leads to a waste of taxi resource. Our APP can recommend the optimal idle searching/cruising route to taxi drivers in Hong Kong, to mitigate supply and demand imbalance. The cruising recommendation algorithm is enpowered by tailored reinforcement learning algorithm.
Charge Express APP
Charge Express APP
By 2025, HK Goverment plans to expand the number of charging stations to 5,000, nearly doubling the current amount. This app allows taxi drivers and other road users to easily find the locations and details of both charging stations and taxi stands. It provides navigation and guides users on the best routes to these sites, helping them avoid wasting time.
Driver-Order Matching Tool
Driver-Order Matching Tool
The taxi market operates under two matching methods: dispatch and broadcast. In this toolbox, users can select the matching mode and upload travel request information from drivers and passengers to obtain matching results. These results can be downloaded or displayed on a map.
A simulator for model evaluations for ride-sourcing markets
A simulator for model evaluations for ride-sourcing markets
Ride-sourcing services have become increasingly important in meeting people's mobility needs since their emergence. Compared to traditional street-hailing taxi services, ride-sourcing services significantly reduce the matching frictions in the markets by matching drivers and passengers with relatively distant distances through an online platform. Motivated by this new feature as well as the need for designing operating and regulating stra-tegies, researchers have attempted to describe these innovative ride-sourcing markets through mathematical models, the core of which is the matching functions for characterizing matching frictions. Previous studies have developed a variety of matching functions for ride-sourcing markets, including perfect matching function, Cobb-Douglas type matching function, queuing models, and some physical models. However, less is known about the applicability and performance of these matching functions, that is, under what situations each of these matching functions well characterizes the real market. To address this issue, this project makes one of the first attempts to calibrate, validate, and compare the prevailing matching functions in the literature, and ascertain the conditions of their applicability. In particular, we establish a simulator to simulate a total of 420 scenarios of the ride-sourcing market under different combinations of supply and demand. The key performance metrics, including the matching rate in the market, passengers' average matching time, passengers' average pick-up time, and passengers' average total waiting time, are utilized to test and compare seven widely used matching functions under various market scenarios. An open-sourced package of simulation and model evaluations is provided.
Related Paper: Wei, S., Feng, S., Ke, J., & Yang, H. (2022). Calibration and validation of matching functions for ride-sourcing markets. Communications in Transportation Research, 2, 100058.
Open-sourced Code Link: https://github.com/hku-kejintao/simulator-matching-function-validation
Google Sites
Report abuse