Project Description

Heavy morning traffic at a large intersection, next to it a school, bike paths, regular buses and a crossing tram – a challenging situation for traffic planning as well as for all road users.

Increasing traffic in inner-city areas aggravates the conflict between safety, traffic efficiency, and environmental pollution. Intersections are critical nodes in traffic networks, which, however, are mostly only controlled via fixed, pre-defined traffic light phase schedules. Inflexible, sub-optimal traffic light control can lead to unnecessary traffic jams and emissions. Individual needs, requirements and possibilities of different road users cannot be considered up to now.

Project Goals

In principle, modern information & communication technologies and control concepts offer the potential to comprehensively manage and optimize intersection traffic in real time. The requirements of all road users can be considered, the situations can be interpreted in the best possible way, and coordinated, cooperative control strategies can be implemented to realize a holistic optimum. Therefore, in this project, novel, integrated and flexible communication, control and simulation methods are to be developed to implement an “intelligent intersection” system that utilizes available real-time information on the positions, speeds and expected behavior of the road users to simultaneously

• ensure traffic safety for all road users in the context of an intersection,
• efficiently regulate the flow of traffic, and thus
• optimize the overall energy consumption and minimize traffic emissions resulting from passing the intersection.

Multi-Agent Model Architecture for Intersection Traffic Simulation

Urban traffic can naturally be modeled as a distributed and heterogeneous multi-agent dynamic system. To allow a scalable solution for simulation, control, and information management, we developed a simple but powerful generic model architecture that supports various traffic participant types. While the agent actions are restricted to a small set of basic operations, a remarkably rich global system functionality and complex information flow can be modeled. The resulting traffic dynamics and information model enables efficient and parallelizable simulation, control, and prediction computations.

Model-Predictive Obstacle Avoidance

Autonomous obstacle avoidance (OA) in urban driving is achieved by coupling two model-predictive controllers (MPCs) that realize OA constraints with different levels of complexity. The resulting Two-Layer OA-MPC architecture enables efficient and collision-free OA in real time by exploiting linear time-invariant MPC design, mixed-integer programming (MIP) and quadratic programming (QP).

Figure 3 illustrates an exemplary co-simulation of an unprotected left turn scenario with the CARLA Simulator: The reference path and road boundary constraints of the ego vehicle 1 are highlighted in the first snapshot, while the predictions of detected traffic participants inside the dashed blue detection range of the ego vehicle are visualized with red circles. The inflated shapes of traffic participants that need to be actively avoided are highlighted in red. The position prediction and history of the ego vehicle are visualized with green circles and green dots, respectively. Traffic lights are also detected and, depending on their phase plan predictions, considered as (static) obstacles. For further insights please read our journal article, opens an external URL in a new window.

© IEEE (2024) with permission under CC-BY-4.0

Figure 3: Co-simulation of an unprotected left turn scenario with the CARLA Simulator, from [3].

Figure 3: Co-simulation of an unprotected left turn scenario with the CARLA Simulator, from [3].

Soft Inputs for Human Drivers

We extend the interaction capabilities between human-driven vehicles (HDVs) and an automated intersection with warnings and maneuver recommendations, such as lane-change or speed recommendations (similar to green light optimal speed advisory (GLOSA) functionalities). These soft inputs enable HDVs to travel more efficiently and safely by exploiting 5G communication, collective perception, and state-of-the-art intersection control concepts. A suitable maneuver recommendation formulation for a conflicting unprotected left-turn scenario is tested in typical urban intersection simulation studies that show high achievable performance with respect to the HDV penetration rate and compliance behavior.

Platoon Control Concepts

Download: Platoon Control Concepts (Springer), opens an external URL in a new window

Gratzer, A.L., Schirrer, A., Thormann, S., Jakubek, S. (2022). Platoon Control Concepts. In: Schirrer, A., Gratzer, A.L., Thormann, S., Jakubek, S., Neubauer, M., Schildorfer, W. (eds) Energy-Efficient and Semi-automated Truck Platooning. Lecture Notes in Intelligent Transportation and Infrastructure. Springer, Cham. doi.org/10.1007/978-3-030-88682-0_8

Publications

[0] Walch, Manuel, Matthias Neubauer, Wolfgang Schildorfer, and Alexander Schirrer. "Modelling interrelations between C-ITS impact categories: a system-dynamics approach using causal loop diagrams, opens an external URL in a new window." European Transport Research Review 16, no. 1 (2024): 60.

[1] Gratzer, Alexander L., Florian Schlägl, Alexander Schirrer, Faruk Pasic, Maryna Kolisnyk, Christoph F. Mecklenbräuker, and Stefan Jakubek. "Modeling and Analysis of Human Drivers' Compliance Behavior to Maneuver Recommendations in Form of Soft Inputs., opens an external URL in a new window" In 2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), pp. 3632-3638. IEEE, 2024.

[2] Gratzer, Alexander L., Alexander Schmiedhofer, Alexander Schirrer, and Stefan Jakubek. "Agile Mixed-Integer-based Lane-Change MPC for Collision-Free and Efficient Autonomous Driving, opens an external URL in a new window." IEEE Transactions on Intelligent Vehicles (2024).

[3] Gratzer, Alexander L., Maximilian M. Broger, Alexander Schirrer, and Stefan Jakubek. "Two-Layer MPC Architecture for Efficient Mixed-Integer-Informed Obstacle Avoidance in Real-Time, opens an external URL in a new window." IEEE Transactions on Intelligent Transportation Systems 25, no. 10 (2024): 13767-13784.

[4] Gratzer, Alexander L., Maximilian M. Broger, Alexander Schirrer, and Stefan Jakubek. "Flatness-Based Mixed-Integer Obstacle Avoidance MPC for Collision-Safe Automated Urban Driving, opens an external URL in a new window." In 2023 9th International Conference on Control, Decision and Information Technologies (CoDIT), pp. 1844-1849. IEEE, 2023.

[5] Gratzer, Alexander L., Alexander Schirrer, and Stefan Jakubek. "Agile Multi-Agent Model Architecture for Intelligent Intersection Traffic Simulation, opens an external URL in a new window." IFAC-PapersOnLine 55, no. 27 (2022): 89-95.

[6] Gratzer, Alexander L., Alexander Schirrer, Elvira Thonhofer, Faruk Pasic, Stefan Jakubek, and C. F. Mecklenbräuker. "Short-Term Collision Estimation by Stochastic Predictions in Multi-Agent Intersection Traffic, opens an external URL in a new window" In 2022 International Conference on Electrical, Computer and Energy Technologies (ICECET), pp. 1-6. IEEE, 2022.

[7] Gratzer, Alexander L., Sebastian Thormann, Alexander Schirrer, and Stefan Jakubek. "String stable and collision-safe model predictive platoon control, opens an external URL in a new window." IEEE Transactions on Intelligent Transportation Systems 23, no. 10 (2022): 19358-19373.

Cooperation Partners

• TU Wien, Institute of Telecommunications
• Andata Artificial Intelligence
• swarco futurit
• komobile
• Kuratorium für Verkehrssicherheit

Duration

• October 2020 - February 2024

More Projects

• ConnAT

Contact