4.7 Article

Neuro-dynamic programming for optimal control of macroscopic fundamental diagram systems

Journal

Publisher

PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.trc.2020.102628

Keywords

Macroscopic fundamental diagram; Hamilton-Jacobi-Bellman equation; Neuro-dynamic programming; Policy iteration; Saturated state and input

Funding

  1. National Key R&D Program of China [2018YFB1600500]
  2. Research Grants Council of Hong Kong [11216819]

Ask authors/readers for more resources

The macroscopic fundamental diagram (MFD) can effectively reduce the spatial dimension involved in dynamic optimization of traffic performance for large-scale networks. Solving the Hamilton-Jacobi-Bellman (HJB) equation takes center stage in yielding solutions to the optimal control problem. At the core of solving the HJB equation is the value function that represents choosing a sequence of actions to optimize the system performance. However, this problem generally becomes intractable for possible discontinuities in the solution and the curse of dimensionality for systems with all but modest dimension. To address these challenges, a neural network is used to approximate the value function to obtain the optimal controls through policy iteration. Furthermore, a saturated operator is embedded in the neural network approximator to handle the difficulty caused by the control and state constraints. This policy iteration can be implemented as an iterative data-driven technique that integrates with the model-based optimal design based on real-time observations. Numerical experiments are conducted to show that the neuro-dynamic programming approach can achieve optimization goals while stabilizing the system by regulating the traffic state to the desired uncongested equilibrium.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

Editorial Material Transportation

Dynamic modelling and optimisation of transportation systems in the connected era

Andy H. F. Chow, Yong-Hong Kuo, Panagiotis Angeloudis, Michael G. H. Bell

TRANSPORTMETRICA B-TRANSPORT DYNAMICS (2022)

Editorial Material Transportation

Special issue on methodological advancements in understanding and managing urban traffic congestion

Renxin Zhong, Zhengbing He, Andy H. F. Chow, Victor Knoop

TRANSPORTMETRICA A-TRANSPORT SCIENCE (2022)

Article Transportation Science & Technology

Adaptive signal control for bus service reliability with connected vehicle technology via reinforcement learning

Andy H. F. Chow, Z. C. Su, E. M. Liang, R. X. Zhong

Summary: This paper introduces an adaptive signal controller that effectively manages traffic delays and urban bus service reliability using fully adaptable acyclic timing plans. The controller is built upon a reinforcement learning framework combining model-based and data-driven components to reduce traffic delays and bus service variabilities.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2021)

Article Transportation Science & Technology

Adaptive network traffic control with an integrated model-based and data-driven approach and a decentralised solution method*

Z. C. Su, Andy H. F. Chow, R. X. Zhong

Summary: This paper presents an adaptive traffic controller for stochastic road networks, using an integrated model-based and data-driven solution framework. The model-based component facilitates the training of the data-driven ADP-based state approximator to improve the overall performance of the control system. A decentralised solution approach is further developed to enhance and stabilise the performance of the overall control system, even under congested conditions.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2021)

Article Computer Science, Artificial Intelligence

Bus arrival time prediction and reliability analysis: An experimental comparison of functional data analysis and Bayesian support vector regression

Y. P. Huang, C. Chen, Z. C. Su, T. S. Chen, A. Sumalee, T. L. Pan, R. X. Zhong

Summary: Accurate bus arrival time prediction is crucial for maintaining stability and attracting more passengers to improve transit services. This paper proposes data-driven approaches based on FDA and BSVR, with a probabilistic nested delay operator, to increase prediction accuracy and conduct journey time reliability analysis. Empirical studies in Guangzhou show that the proposed methods are competitive in various traffic conditions, with FDA providing more accurate results and anticipating uncertainties effectively.

APPLIED SOFT COMPUTING (2021)

Article Transportation

Lane-based estimation of travel time distributions by vehicle type via vehicle re-identification using low-resolution video images

Cheng Zhang, H. W. Ho, William H. K. Lam, Wei Ma, S. C. Wong, Andy H. F. Chow

Summary: This paper proposes a new method for estimating lane-based travel time distributions by vehicle type using low-resolution vehicle video images captured by conventional traffic surveillance cameras. The method utilizes deep learning and graph matching techniques, and performs well in vehicle type-specific traffic management schemes.

JOURNAL OF INTELLIGENT TRANSPORTATION SYSTEMS (2023)

Article Economics

Reinforcement learning for logistics and supply chain management: Methodologies, state of the art, and future opportunities

Yimo Yan, Andy H. F. Chow, Chin Pang Ho, Yong-Hong Kuo, Qihao Wu, Chengshuo Ying

Summary: This paper provides a comprehensive review of the development and applications of reinforcement learning techniques in logistics and supply chain management. The most popular approach, Q-learning, is adopted by many studies, and recent research in urban logistics has been growing rapidly. Potential directions for future research are also presented.

TRANSPORTATION RESEARCH PART E-LOGISTICS AND TRANSPORTATION REVIEW (2022)

Article Engineering, Civil

Vehicle Re-identification for Lane-level Travel Time Estimations on Congested Urban Road Networks Using Video Images

Cheng Zhang, Bi Yu Chen, William H. K. Lam, H. W. Ho, Xiaomeng Shi, Xiaoguang Yang, Wei Ma, S. C. Wong, Andy H. F. Chow

Summary: The study proposes a new vehicle re-identification method to estimate lane-level travel time distributions by considering lane-level traffic conditions, vehicles' lane changing behaviors, and visual features. A comprehensive case study in Hong Kong demonstrates that the proposed method outperforms existing methods and provides accurate lane-level travel time distribution information on congested urban roads.

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS (2022)

Article Engineering, Civil

Two-Stage Stochastic Program for Dynamic Coordinated Traffic Control Under Demand Uncertainty

Lubing Li, Wei Huang, Andy H. F. Chow, Hong K. Lo

Summary: This study develops a cell-based two-stage stochastic program to address the dynamic, spatial, and stochastic characteristics of traffic flow for arterial adaptive signal control. By incorporating the concept of Phase Clearance Reliability (PCR) and using a gradient-based solution algorithm, the study enhances solution efficiency and validates findings through VISSIM. Results show the importance of capturing dynamic, spatial, and stochastic features for traffic control in order to avoid delay performance degradation.

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS (2022)

Article Engineering, Civil

Adaptive Metro Service Schedule and Train Composition With a Proximal Policy Optimization Approach Based on Deep Reinforcement Learning

Cheng-Shuo Ying, Andy H. F. Chow, Yi-Hui Wang, Kwai-Sang Chin

Summary: This study presents an integrated metro service scheduling and train unit deployment approach based on deep reinforcement learning framework with a proximal policy optimization method. By parameterizing the value function and control policy through artificial neural networks and incorporating operational constraints through a devised mask scheme, the optimization problem was successfully solved in real-world scenarios with superior performance. Results show the advantages of flexible train compositions in saving operational costs and reducing service irregularities.

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS (2022)

Article Economics

Multi-agent deep reinforcement learning for adaptive coordinated metro service operations with flexible train composition

Cheng-Shuo Ying, Andy H. F. Chow, Hoa T. M. Nguyen, Kwai-Sang Chin

Summary: This paper presents an adaptive control system for coordinated metro operations using a multi-agent deep reinforcement learning approach. The system outperforms previous methods in terms of solution quality and performance achieved.

TRANSPORTATION RESEARCH PART B-METHODOLOGICAL (2022)

Article Economics

Hierarchical control for stochastic network traffic with reinforcement learning

Z. C. Su, Andy H. F. Chow, C. L. Fang, E. M. Liang, R. X. Zhong

Summary: This study proposes a hierarchical control framework to maximize the throughput of a road network driven by travel demand with uncertainties. The upper level uses a reinforcement learning algorithm to regulate the traffic influx into the core road network without the need for an underlying system model and macroscopic fundamental diagram. The lower level is a local signal control system that regulates the spatial distribution of traffic flow within the core network. The study contributes to the management of urban road networks with advanced computing technologies.

TRANSPORTATION RESEARCH PART B-METHODOLOGICAL (2023)

Article Transportation Science & Technology

Adaptive rail transit network operations with a rollout surrogate-approximate dynamic programming approach

Hoa T. M. Nguyen, Andy H. F. Chow

Summary: This paper presents an adaptive optimization framework for dynamic rail transit network operations using a rollout surrogate-approximate dynamic programming method. The proposed framework reduces passengers' waiting times significantly with reasonable computational time. The results suggest the potential of the proposed optimizer for real-time applications in large-scale rail transit networks.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2023)

Proceedings Paper Computer Science, Artificial Intelligence

OAM: An Option-Action Reinforcement Learning Framework for Universal Multi-Intersection Control

Enming Liang, Zicheng Su, Chilin Fang, Renxin Zhong

Summary: Efficient traffic signal control is crucial for alleviating urban traffic congestion. Reinforcement learning has the potential to devise optimal signal plans that can adapt to dynamic congestion, but faces challenges. To address these challenges, a universal multi-intersection control framework is proposed, which incorporates a well-known cell transmission model, regularized delay as reward, and a universal neural network structure. Results demonstrate that the proposed framework outperforms state-of-the-art controllers in reducing average travel time.

THIRTY-SIXTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTY-FOURTH CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE / THE TWELVETH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE (2022)

Article Transportation Science & Technology

3-Strategy evolutionary game model for operation extensions of subway networks

Yue Zhao, Liujiang Kang, Huijun Sun, Jianjun Wu, Nsabimana Buhigiro

Summary: This study proposes a 2-population 3-strategy evolutionary game model to address the issue of subway network operation extension. The analysis reveals that the rule of maximum total fitness ensures the priority of evolutionary equilibrium strategies, and proper adjustment minutes can enhance the effectiveness of operation extension.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2024)

Article Transportation Science & Technology

Integrated optimization of container allocation and yard cranes dispatched under delayed transshipment

Hongtao Hu, Jiao Mob, Lu Zhen

Summary: This study investigates the challenges of daily storage yard management in marine container terminals considering delayed transshipment of containers. A mixed-integer linear programming model is proposed to minimize various costs associated with transportation and yard management. The improved Benders decomposition algorithm is applied to solve the problem effectively and efficiently.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2024)

Article Transportation Science & Technology

Range-constrained traffic assignment for electric vehicles under heterogeneous range anxiety

Zhandong Xu, Yiyang Peng, Guoyuan Li, Anthony Chen, Xiaobo Liu

Summary: This paper studied the impact of range anxiety among electric vehicle drivers on traffic assignment. Two types of range-constrained traffic assignment problems were defined based on discrete or continuous distributed range anxiety. Models and algorithms were proposed to solve the two types of problems. Experimental results showed the superiority of the proposed algorithm and revealed that drivers with heightened range anxiety may cause severe congestion.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2024)

Article Transportation Science & Technology

Demand forecasting and predictability identification of ride-sourcing via bidirectional spatial-temporal transformer neural processes

Chuanjia Li, Maosi Geng, Yong Chen, Zeen Cai, Zheng Zhu, Xiqun (Michael) Chen

Summary: Understanding spatial-temporal stochasticity in shared mobility is crucial, and this study introduces the Bi-STTNP prediction model that provides probabilistic predictions and uncertainty estimations for ride-sourcing demand, outperforming conventional deep learning methods. The model captures the multivariate spatial-temporal Gaussian distribution of demand and offers comprehensive uncertainty representations.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2024)

Article Transportation Science & Technology

Partial trajectory method to align and validate successive video cameras for vehicle tracking

Benjamin Coifman, Lizhe Li

Summary: This paper develops a partial trajectory method for aligning views from successive fixed cameras in order to ensure high fidelity with the actual vehicle movements. The method operates on the output of vehicle tracking to provide direct feedback and improve alignment quality. Experimental results show that this method can enhance accuracy and increase the number of vehicles in the dataset.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2024)

Article Transportation Science & Technology

Dynamic routing for the Electric Vehicle Shortest Path Problem with charging station occupancy information

Mohsen Dastpak, Fausto Errico, Ola Jabali, Federico Malucelli

Summary: This article discusses the problem of an Electric Vehicle (EV) finding the shortest route from an origin to a destination and proposes a problem model that considers the occupancy indicator information of charging stations. A Markov Decision Process formulation is presented to optimize the EV routing and charging policy. A reoptimization algorithm is developed to establish the sequence of charging station visits and charging amounts based on system updates. Results from a comprehensive computational study show that the proposed method significantly reduces waiting times and total trip duration.

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES (2024)