Anastasios Kouvelas


Last Name

Kouvelas

First Name

Anastasios

ORCID

0000-0003-4571-2530

Organisational unit

02610 - Inst. f. Verkehrspl. u. Transportsyst. / Inst. Transport Planning and Systems

Search Results

Publications 1 - 10 of 251
  • Consistent vehicle trajectory extraction from aerial recordings using oriented object detection
    Item type: Journal Article
    Riehl, Kevin; El-Baklish, Shaimaa K.; Kouvelas, Anastasios; et al. (2025)
    Scientific Reports
    Vehicle trajectories offer valuable insights for a wide range of road transportation applications. Due to the rise of drone technology, a growing branch of literature explores optical vehicle trajectory extraction from aerial videos, where object detection using neural networks is an important component. Horizontal bounding box object detection struggles to differentiate well between rotated vehicles, especially when dealing with complex backgrounds or densely packed vehicles. This work proposes a generalizable computation pipeline that leverages angular information to extract high-quality trajectories starting from video recordings and ending in trajectories in Cartesian and lane coordinates. A trajectory reconstruction algorithm is designed to be vehicle- and driver-informed and to maximize the physical consistency of the reconstructed trajectories both on the individual vehicles’ and platoon levels. A comprehensive benchmark of 18 object detection models on a real-world video dataset demonstrates how oriented object detection and the use of angular information can be used to significantly improve the consistency of extracted trajectories (15% better internal, and 20% better platoon consistency), and that orientation-informed trajectories can be reconstructed to lane coordinates of higher quality. The reconstructed vehicle trajectories better capture car-following and traffic dynamics, thereby improving their usability for traffic flow studies.
  • Time-to-Green predictions for fully-actuated signal control systems with supervised learning
    Item type: Working Paper
    Genser, Alexander; Makridis, Michail; Yang, Kaidi; et al. (2022)
    arXiv
    Recently, efforts have been made to standardize signal phase and timing (SPaT) messages. These messages contain signal phase timings of all signalized intersection approaches. This information can thus be used for efficient motion planning, resulting in more homogeneous traffic flows and uniform speed profiles. Despite efforts to provide robust predictions for semi-actuated signal control systems, predicting signal phase timings for fully-actuated controls remains challenging. This paper proposes a time series prediction framework using aggregated traffic signal and loop detector data. We utilize state-of-the-art machine learning models to predict future signal phases' duration. The performance of a Linear Regression (LR), a Random Forest (RF), and a Long-Short-Term-Memory (LSTM) neural network are assessed against a naive baseline model. Results based on an empirical data set from a fully-actuated signal control system in Zurich, Switzerland, show that machine learning models outperform conventional prediction methods. Furthermore, tree-based decision models such as the RF perform best with an accuracy that meets requirements for practical applications.
  • Antifragile perimeter control
    Item type: Working Paper
    Sun, Linghang; Makridis, Michail; Genser, Alexander; et al. (2024)
    arXiv
    The optimal operation of transportation networks is often susceptible to unexpected disruptions, such as traffic incidents and social events. Many established control strategies rely on mathematical models that struggle to cope with real-world uncertainties, leading to a significant decline in effectiveness when faced with substantial disruptions. While previous research works have dedicated efforts to improving the robustness or resilience of transportation systems against disruptions, this paper applies the cutting-edge concept of antifragility to better design a traffic control strategy for urban road networks. Antifragility sets itself apart from robustness and resilience as it represents a system's ability to not only withstand stressors, shocks, and volatility but also thrive and enhance performance in the presence of such adversarial events. Hence, modern transportation systems call for solutions that are antifragile. In this work, we propose a model-free deep Reinforcement Learning (RL) scheme to control a two-region urban traffic perimeter network. The system exploits the learning capability of RL under disruptions to achieve antifragility. By monitoring the change rate and curvature of the traffic state with the RL framework, the proposed algorithm anticipates imminent disruptions. An additional term is also integrated into the RL algorithm as redundancy to improve the performance under disruption scenarios. When compared to a state-of-the-art model predictive control approach and a state-of-the-art RL algorithm, our proposed method demonstrates two antifragility-related properties: (a) gradual performance improvement under disruptions of constant magnitude; and (b) increasingly superior performance under growing disruptions.
  • MESO-U
    Item type: Conference Poster
    Ni, Ying-Chuan; Kouvelas, Anastasios; Makridis, Michail (2025)
  • Resilience enhancement of urban roadway network during disruption via perimeter control
    Item type: Journal Article
    Zhu, Chunli; Wen, Guanghui; Li, Nan; et al. (2024)
    IEEE Transactions on Network Science and Engineering
    Frequent happened extreme weather events (EWEs) cause severe disruptions to the operation of large-scale urban road network. Perimeter control is of high application potential in the target scenarios. However, few studies are concerned about the lacking knowledge of the system's response features under EWE. In this work, we proposed a network resilience curve (NRC)-based perimeter control strategy to facilitate the network equilibrium, therefore enhancing the network resilience. The proposed NRC is an extension of the classical macroscopic fundamental diagram (MFD) under disruptions. A real-world trajectory dataset under normal and rainstorm day has been analyzed comparatively by using the present NRC, in which average velocity immediately reduces while average flow reveals the hysteresis effect. We compared the strategies of fixed plan, NRC-based proportional-integral (PI) control without or with connected vehicles, and min-max model predictive control. Case studies show that the proposed NRC-based PI controller improves the average weighted speed by 11.03% over fixed time strategy and recovered 7.14% ahead of the other strategies. Results demonstrate the feasibility and stability of the proposed strategy, which contributes to exploit the reasonable regulatory mechanism of EWE type disruptions.
  • Real-time merging traffic control for throughput maximization at motorway work zones
    Item type: Other Conference Item
    Tympakianaki, Athina; Spiliopoulou, Anastasia D.; Kouvelas, Anastasios; et al. (2012)
  • Link travel time estimation with spatial correlations based on OD data
    Item type: Conference Paper
    Ghandeharioun, Zahra; Rau, Moritz; Kouvelas, Anastasios (2021)
    Understanding complicated city traffic patterns has been recognized as a key goal by twenty- first century urban planners and traffic management systems, resulting in a significant rise in the quantity and variety of traffic data gathered. For example, taxi firms in a growing number of large cities have begun to collect metadata for each individual vehicle trip, such as origin, destination, and travel duration. Taxi data offer information on traffic patterns, allowing the study of urban flow – what will traffic look like between two sites at a certain day and time in the future? In this paper, we propose a method based on sparse GPS probe data that focuses on how to allocate travel time data to the different links traveled between GPS observations. This model incorporates the spatial correlations between the links in a network. The main goal of this work is to show, how with a simple adjustment in previously known parametric methods we can consider spatial correlations and improve our results in a more realistic way. For estimation of arterial travel time, the methodology is applied to a case study for the partial network of New York City based on the data, collected from the taxicabs in New York City, providing the locations of origins, destinations and travel times. The model estimates quarter hourly averages of urban link travel times using OD trip data. This study proposes a more accurate approach for estimating link travel times that fully utilizes the partial information received from taxis data in cities.
  • A nonlinear model predictive control approach for coordinated control of highways
    Item type: Conference Paper
    Chavoshi, Kimia; Kouvelas, Anastasios (2020)
    The growing level of freeway traffic congestion comprises an everyday life issue with social, economic, and environmental implications for modern metropolitan areas. Although there is evidence that Variable Speed Limits (VSL) and Ramp Metering (RM) are two effective practical approaches to ameliorate traffic congestion. The positive effects that these approaches can have on traffic flow and congestion can be demonstrated with the augmented METANET model, which is one of the most widely used macroscopic models for freeway traffic. Since the augmented METANET is a nonlinear model, nonlinear model predictive control (NLMPC) is an appropriate control method for this system. It performs as a recursive on-line finite-horizon optimization of nonlinear problems, subject to the system dynamics and additional constraints, and has the privilege of prediction of future system states. In this paper, we utilized the NLMPC method for the coordination of VSL and RM in highway networks. Afterwards, we provide a case study where we simulate the implementation of the proposed control method on a freeway that contains a typical setting of on-ramps, off-ramps, as well as a lane drop that creates a bottleneck. The simulation results demonstrate significant improvement in the traffic flow conditions and provide useful insights about the way that VSL and RM manage to achieve this improvement. Understanding the special characteristics of capacity drop in highways, and how to ameliorate it, is crucial for future large-scale implementations.
  • Green-Pressure
    Item type: Conference Paper
    Riehl, Kevin; Kouvelas, Anastasios; Makridis, Michail (2025)
    2025 IEEE 64th Conference on Decision and Control (CDC)
    Urban transportation networks increasingly suffer from congestion. Negative externalities resulting from noise and pollution, affect public health, quality of life, and the economy. The major traffic bottlenecks in cities are conflicts at intersections, leading to this pressing issue. Intelligent transportation systems leverage sensors to optimize traffic flows, mainly by control of traffic lights. Green-Pressure is an extension of the Max-Pressure algorithm, that leverages vehicle category information from loop-detectors for a weighted queue-length approach, to reduce emissions at signalized intersections. A multi-modal, case study of a real-world artery network with seven intersections, and 96 traffic signals, demonstrates the feasibility of the proposed method using a calibrated microsimulation model. Interestingly, the differentiation of vehicle categories at traffic lights not only enables reductions in emissions up to 9% but also improves traffic efficiency significantly (5% reduction of total travel time) when compared with the (unweighted) Max-Pressure controller. This is achieved by systematic prioritization of transporters, trucks, and buses, at the cost of slightly larger delays for passenger cars and motorcycles. Ultimately, the proposed method has the potential to achieve more sustainable road traffic leveraging existing sensor infrastructure. Code & resources: https://github.com/DerKevinRiehl/green_pressure_control
  • Multi-region control for urban road networks using adaptive optimization
    Item type: Conference Paper
    Kouvelas, Anastasios; Saeedmanesh, Mohammadreza; Geroliminis, Nikolas (2015)
    In this paper we present a control scheme for heterogeneous transportation networks which is based on the concept of the Macroscopic Fundamental Diagram (MFD) integrated with an adaptive optimization technique. The heterogeneous transportation network is first partitioned into a number of regions with homogeneous traffic conditions and well-defined MFDs. An appropriate clustering algorithm that consumes traffic information is applied for that purpose. A macroscopic MFD-based model is used to describe the traffic dynamics of the resulting multi-region transportation system. A multivariable proportional integral (PI) feedback regulator is implemented to control the nonlinear system in real-time. The control variables consist of the inter-transferring flows between neighbourhood regions and the actuators correspond to the traffic lights of these areas (e.g. boundaries between regions). The recently proposed Adaptive Fine-Tuning (AFT) algorithm is used to optimize the gain matrices as well as the vector with the set-points of the PI controller. AFT is an iterative adaptive algorithm that optimizes the values of the tuneable parameters of the controller (e.g. gains and set-points) based on measurements of a performance index (e.g. total delay) for different perturbations of the parameters. In each iteration AFT receives (a) the total delay of the system and (b) some aggregate information about the demand of each region and calculates the control parameters to be used at the next iteration. The overall control scheme is tested in simulation and different performance criteria are studied. The performance of the initial PI controller which corresponds to a fixed-time policy is compared to the final controller that is obtained after the convergence of AFT. A key advantage of the proposed approach is that it does not require high computational effort and future demand data if the current state of each region can be observed with loop detector data.
Publications 1 - 10 of 251