2021

Ramin Niroumand, North Carolina State University
Leila Hajibabai, North Carolina State University
Ali Hajbabaie (ahajbab@ncsu.edu), North Carolina State University

This study presents a vehicle-level distributed coordination strategy to control a mixed traffic stream of connected automated vehicles (CAVs) and connected human-driven vehicles (CHVs) through signalized intersections. A new phase (a.k.a. white phase) is incorporated into the signals, during which CAVs will negotiate the right-of-way and CHVs must follow their immediate front vehicle. Each CAV optimizes its own trajectory while leading a group of CHVs and navigating them through the intersection during the white phase. The white phase will not be activated under low CAV penetration rates, where vehicles must wait for green signals to travel through the intersection. We have formulated this problem as a distributed mixed-integer non-linear program and developed a methodology to form an agreement among all vehicles in the intersection vicinity on their trajectories and signal timing parameters. The agreement on trajectories is reached through an iterative process, where CAVs receive the planned trajectory of other vehicles, make necessary adjustments to their trajectory to avoid collisions, and share their updated trajectories with others. Additionally, the agreement on signal timing parameters is formed through a voting process with the participation of all vehicles, where the most voted feasible signal timing parameters are selected. The numerical experiments indicate that the proposed methodology can efficiently control vehicle movements at signalized intersections under various CAV market penetration rates. The proposed technique yield reductions in total delay, ranging from 2.7% - 98.7%, compared to that of a fully-actuated signal control obtained from a state-of-practice traffic signal optimization software.

Qiangqiang Guo, University of Washington
Zhijun Liu, University of Washington
Ohay Angah, University of Washington
Xuegang Ban (banx@uw.edu), University of Washington

Eco-Driving has great potential to reduce the fuel consumption of road vehicles, especially under the connected and autonomous vehicles (CAV) environment. Traditional model-based Eco-Driving methods usually require sophisticated models thus can not deal with complex driving scenarios. This paper proposes a reinforcement learning (RL) based Eco-Driving algorithm considering both longitudinal acceleration/deceleration and lateral lane-changing operations. Given the information of the front vehicle and upcoming signals, a deep deterministic policy gradient (DDPG) algorithm is designed to learn the longitudinal operations to reduce the fuel consumption as well as maintain acceptable travel times. Collecting the state-action value of each single lane and integrating the information of adjacent lanes, a deep Q-learning algorithm is developed to make the lane-changing decisions. Together, a hybrid deep Q-learning and policy gradient (HDQPG) algorithm is developed for vehicles driving along multi-lane signalized roads. The numerical experiments on a three-lane and five-intersection corridor show that, under the studied scenarios, the HDQPG can reduce fuel consumption by up to 46% and meanwhile limit the increment of travel time under 12%. The controlled vehicle learns to use longitudinal fuel-saving strategies such as braking earlier before stop at red signals and “pulsing and gliding” while cruising to reduce fuel consumption, and perform appropriate lane-changing operations to avoid congested lanes to improve or at least maintain the performance of travel time.

Xiao Liang, Pennsylvania State University
S. Ilgin Guler, Pennsylvania State University
Vikash Gayah (gayah@engr.psu.edu), Pennsylvania State University, University Park

This paper proposes a decentralized signal control algorithm that leverages connected vehicle information to improve traffic operations along arterials. The proposed algorithm obtains real-time vehicle locations and speeds, as well as information on pedestrians waiting to cross individual intersections, to optimize signal phasing and timing plans. Signal timing is optimized at individual intersections; however, information about vehicle platoons passing a given intersection is shared with neighboring intersections to facilitate natural coordination between adjacent intersections. The decentralized algorithm is compared to a centralized algorithm that optimizes signal timing at all intersection simultaneously, as well as a traditional coordination strategy. The proposed algorithm is shown both to be more computationally efficient than the centralized approach and provide better operational performance (in terms of person, vehicle and pedestrian delay) than both the centralized algorithm and the traditional strategy. The algorithm is robust to a range of demand patterns and can be applied under scenarios in which not all vehicles are connected or full information about pedestrian arrivals is not available.

Samira Ziyadidegan (samiraziyadg@tamu.edu), Texas A&M University, College Station
Xiaoyu Skye Guo, District Department of Transportation
Luca Quadrifoglio, Texas A&M University

Signalized intersections are one of the most critical bottlenecks in a network, since their throughputs are remarkably lower than the available flow in their upstream section. Platooned vehicles passing through the intersection can brings many benefits and alleviate such issues. However, when a platoon is crossing an intersection, it can be split due to the coordination of the signal phases, gap length and individual vehicle’s motions (i.e., velocity and acceleration) which affects mobility and sustainability. To address this issue, this paper develops general motion equations and determines an optimal size of a subset (n vehicles) in an N-vehicle platoon (n ≤ N) under different scenarios to minimize the average delay among all N vehicles. Further, a sensitivity analysis is conducted to study the impacts of different parameters on the optimized size. Results show that the intersection, vehicle’s motions, and vehicle platoon attributes have effects on determining the optimal size. The determined optimal platoon size would be beneficial for research areas such as eco-driving simulations and signal optimizations by minimizing the average delay and number of stops per vehicle in the network. Moreover, this optimal size would play a potential role in developing a platooning policy for vehicle-to-vehicle flow.

Tang Li (tang.li16@imperial.ac.uk), Imperial College London
Fangce Guo, Imperial College London
Rajesh Krishnan, Imperial College London
Aruna Sivakumar, Imperial College London

With the emergence of connected and automated technologies, Connected Autonomous Vehicles (CAVs) are able to communicate and interact with other vehicles and signal controllers. Vehicle to Vehicle (V2V) and Vehicle to Infrastructure (V2I) communications open up an opportunity to improve routing and signal timing efficiency with additional information from CAVs, such as prior travel time and signal green time. Most of the existing research on routing and signal timing for Human Driven Vehicles (HDVs) has to face the fact that human drivers only have partial knowledge about travel cost and traffic status on the road network, which typically reduces the system efficiency. In this paper, the impacts of additional information from CAVs on routing and signal timing efficiency in terms of total travel time has been investigated. An Optimal Routing and Signal Timing (ORST) control strategy for CAVs has been proposed and compared with four existing routing and signal timing strategies where drivers have different levels of information. The results of the simulation demonstrate that with additional information from CAVs, ORST can reduce about 49% of the total travel time compared with Stochastic User Equilibrium (SUE) and about 10% of the total travel time compared with User Equilibrium (UE).

Huiyu Chen, University of Alberta
Tony Qiu (zqiu1@ualberta.ca), University of Alberta

In the existing centralized joint dynamic route guidance and signal control (DRG-SC) studies, the exponentially growing data-processing needs and problem-solving complexity have naturally led to an un-timely solution which is unsuitable to be applied to the dynamically changing connected vehicle (CV) environment. Mobile edge computing (MEC) pushes the data storage and computation from the remote cloud to local vehicles or infrastructures, which reduces response time and improves network bandwidth when further combined with the 5G technology. Facilitated by MEC and 5G benefits, this study first developed a novel simulation-based distributed DRG-SC framework with detailed data communication, processing, and computation procedure. Vehicles cooperate and are guided to reach either user optimal (UO) or system optimal (SO) traffic state. In tandem, the proposed adaptive signal control (ASC) changes the signal timing plan based on the adjacent intersections’ traffic volume. Finally, a comprehensive case study is implemented in SUMO software to test the effectiveness of the proposed method. The results demonstrate significant reductions in vehicles’ average departure delay, waiting time and travel time. What’s more, the effectiveness of adopting the distributed framework in saving computation time is illustrated by multiprocessing-based parallelism. Overall, the proposed method is scalable, computation effective and gives considerable network performance. The whole study provides valuable and practical insights for intelligent traffic operation and control.

Jiangchen Li, Nanjing University of Aeronautics and Astronautics
Liqun Peng, University of Alberta
Tony Qiu (zqiu1@ualberta.ca), University of Alberta

Although connected vehicle (CV)-based signal coordination systems have been investigated from both offline and online strategic perspective s, the existing work has yet to address certain coordination performance issues, including the dynamic platoon dispersion effect and real-time high-frequency communication. As such, our study proposes a real-time predictive coordination system consisting of a probabilistic single-vehicle-based dynamic platoon dispersion model, an extended link performance function, and a real-time model predictive control (MPC)-based coordination framework to improve coordination performance. The probabilistic single-vehicle-based dynamic platoon dispersion model is used to model platoon dispersion effects in high-frequency communication by utilizing the potential of a single vehicle’s position and trajectory information. The proposed coordination system was comprehensively investigated by a software-in-loop simulation platform that addresses different practical corridor scenarios in the ACTIVE CV testbed in Canada. As we demonstrate, simulation results show that the MPC-based signal coordination continuously outperformed existing signal control with lower delays for major streets with different demand profiles and different CV penetration rates, even in the low penetration conditions. We argue here that the proposed single-vehicle-based platoon dispersion model, along with real-time MPC-based coordination, offer significant potential to further improve the system performance of signal coordination in a low penetration environment.

Roozbeh Mohammadi (roozbeh.mohammadi@aalto.fi), Aalto University
Claudio Roncoli, Aalto University
Miloš Mladenović, Aalto-yliopisto

Connected vehicle (CV) technology is expected to bring unprecedented opportunities to share, collect, and exploit various information on vehicles and their occupants. This capability can be exploited to develop novel traffic management strategies, replacing or complementing the current dominant strategies, which have been mainly designed to facilitate vehicle flow. Assuming that CVs are able to transmit on-board users and vehicle data, we propose a user-based signal timing optimisation (UBSTO) strategy, designed to optimize user throughput for signalized intersections. In the CV environment, the inputs of the proposed algorithm consist of position and speed of CVs, as well as the number of passengers travelling in each vehicle, while the output is the optimum green time duration for each signal phase. In addition, our proposed strategy is able to adapt the cycle length to the traffic volume condition. In case of missing users data, the same strategy can also operate in vehicle-based mode, where the objective is vehicle-throughput maximization. The performance of the proposed strategy is compared with a fully actuated controller (FAC) in microscopic simulation for several scenarios, including  different CV penetration rates. Our findings show that UBSTO can effectively increase user throughput and decrease average user delay in comparison with FAC, while also prioritizing vehicles with higher number of users on-board. These findings have implications for further development of prioritization strategies for public transport and ride-sharing vehicles.

Zhihong Yao (zhyao@swjtu.edu.cn), Southwest Jiaotong University
Yangsheng Jiang, Southwest Jiaotong University
Bin Ran, University of Wisconsin, Madison

With the development of connected automated vehicles (CAVs) technologies, CAVs trajectories not only provide more real-time information by vehicles to infrastructure (V2I), but also can be controlled and optimized by the infrastructure to vehicles (I2V) to further reduce travel time and gasoline consumption. This paper proposes a two-level model for traffic signal timing and trajectories planning of multiple connected automated vehicles to reduce vehcle delay and gasonline. The proposed method consists of two levels, i.e., traffic signals and vehicle’s arrival time optimization, and multiple CAVs trajectories planning. The former optimizes signal timing plan and vehicles’ arrival time for random arrival CAVs to minimize the average vehicle’s delay. The latter optimizes multiple CAVs trajectories considering average gasoline consumption. The dynamic programming (DP) and the General Pseudospectral Optimal Control Software (GPOPS) are applied to solve the two-level optimization problem. Numerical studies are conducted to compare the proposed method with a fixed-time plan when vehicle trajectories are optimized. The results show significantly reduce both average vehicle delay and gasoline consumption under different demand levels by the proposed method. The reduced average vehicle’s delay and gasoline consumption can be as much as 26.91% and 10.38%, respectively, for a simple two-phase intersection. Sensitivity analysis suggests that the minimum green time and free flow speed have a noticeable impact on the average vehicle’s delay and gasoline consumption reduction.

Qinzheng Wang, Leidos, Inc.
Xianfeng Yang (xtyang@umd.edu), University of Maryland, College Park
Yun Yuan, University of Utah
Zhitong Huang, Leidos, Inc.

The rapid development of wireless communication enables connected vehicles (CVs) technology to reach a level of maturity. Enriched information collected from CVs describe traffic states near an intersection, and supplements a data source for an effective signal control. This paper proposes a traffic signal control system integrating adaptive traffic signal control and dynamic signal coordination control in a connected environment. This system consists of optimization problems at intersection and corridor levels. At the intersection level, a real-time adaptive control model is developed to assign optimal green times by minimizing total vehicle delay. When the market penetration rate is low, a method depending on limited CVs data is presented to estimate the vehicle arrival information. At the corridor level, a real-time optimization model is formulated to design the optimal coordination plan for critical paths (i.e. paths with high flows), and the objective is to maximize green bandwidth for vehicles travelling along this corridor. These two optimization models are solved using dynamic programming. A simulated arterial in VISSIM is modeled to evaluate the effectiveness and efficiency of the proposed traffic signal control system. Various market penetration rates of CVs is tested. The results indicate that the proposed control system can improve the performance of all critical paths under all market penetration rates and the arterial performance under high market penetration rates compared with other control systems.

Wanjing Ma, Tongji University
Lijuan Wan, Tongji University
Chunhui Yu (hughyu90@tongji.edu.cn), Tongji University
Li Zou, Shenzhen Urban Transport Planning Center
Jianfeng Zheng, Didi Chuxing Inc.
Dongbo Liu, Southeast University

Existing fixed-time traffic signal optimization methods mainly use traffic volumes collected by infrastructure-based detectors (e.g., loop detectors). These infrastructure-based detectors generally have high maintenance costs and low coverage. With the deployment of probe vehicles, vehicle trajectory data provide more information about traffic states and can be utilized for signal timing. However, most related studies assume high penetration rates of probe vehicles or short sampling intervals. This paper develops a hierarchical multi-objective optimization framework to optimize fixed-time traffic signals based on sampled vehicle trajectories at isolated signalized intersections, which is applicable to low-resolution trajectory data. Cycle length and green splits are optimized under both under- and slightly over-saturated traffic conditions. The number of over-saturated phases and average vehicle delays are adopted as the primary and the secondary objectives, respectively. The aggregation of sampled trajectory data during the same period across multiple cycles and Same-ratio Principles (SRPs) are proposed to compensate for the limitations of low penetration rates of probe vehicles. The evolution of sampled trajectories with varying signal timings are formulated explicitly. A sampled-trajectory-density method is proposed to identify over-saturated phases. Then a mixed integer non-linear programming (MINLP) model is formulated and solved by solving several MILP models. Simulation studies validate the advantages of the proposed model over the one in Synchro Studio. Sensitivity analysis shows that the proposed model can handle sampling intervals as long as 15 s when sufficient sampled trajectories are collected. The proposed model is implemented with field data to demonstrate its applicability in real world.

Md Abu Sufian Talukder (mtalukder@crimson.ua.edu), University of Alabama
Abhay Lidbe, University of Alabama
Elsa Tedla, Alabama Transportation Institute
Alexander Hainen, University of Alabama
Travis Atkison, University of Alabama

The recent advancement in connected vehicle (CV) technology has presented significant prospect of reforming conventional traffic signal operation. This CV technology has facilitated the access to detailed vehicle trajectory information such as location, speed, and acceleration by using high speed information transaction between vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I). Such trajectory information can provide more insightful information about the traffic conditions nearby and can potentially be utilized for enhanced signal control operation. However, low penetration rate of CV enabled vehicles on the road and limited deployments of V2I communications has made the implementation of CV technology impractical. Thus, this paper proposes a novel signal control approach, which utilizes vehicle trajectory information with conventional traffic signal controllers in a limited use or even absence of V2I communications. A weighted delay-based algorithm (WDBA) was developed to demonstrate delay optimization at an isolated signalized intersection to improve its operational performance. The intersection was modelled in VISSIM microsimulation and the proposed algorithm was coded in python. The simulation results were compared with existing free control operation for a variety of traffic demand scenarios. Analysis results confirmed that the proposed WDBA significantly improved intersection operational performance in terms of vehicle delay, stop delay, and queue length thereby demonstrating a more efficient control method than existing actuated control.