rban signalized intersections remain safety-critical despite phase-based control, particularly in scenarios where right-turning vehicles and straight-going Vulnerable Road Users (VRUs) approach from the same direction and may enter a shared conflict area with overlapping arrival times. This risk is amplified by the increasing prevalence of electric bicycles, whose higher speeds and accelerations reduce the available reaction time. Green Light Optimized Speed Advisory (GLOSA) can improve efficiency by coordinating vehicle approach speeds with Signal Phase and Timing~(SPaT), but conventional implementations are largely vehicle-centric and do not explicitly manage VRU conflicts. This paper presents a VRU-aware GLOSA that augments classical SPaT-based speed advice with a conflict-aware safety overlay. We evaluate three strategies in SUMO: baseline (no advisory), classical queue-compensated GLOSA, and VRU-aware GLOSA. The simulation is based on a digital twin of a real-world signalized intersection in Ingolstadt, Germany. Results demonstrate that VRU-aware GLOSA successfully improves safety by reducing conflicts and increasing temporal gaps, while also enhancing traffic flow through fewer vehicle stops across diverse bicycle types and riding styles.
Motorcycles exhibit significant roll motion during cornering, which affects their physical characteristics in traffic simulations. However, SUMO does not natively provide roll angle estimation or lean aware vehicle geometry for motorcycles since usually a car model is used for motorcycles. This paper presents a lightweight method to estimate motorcycle roll dynamics in SUMO using lane-level network information and vehicle kinematics. The approach computes local road curvature from lane geometry and then computes the roll angle at each simulation step. The method is implemented and evaluated across multiple scenarios, including curved roads, junction turns, and straight segments. Results demonstrate stable, physically consistent roll evolution, enhancing the realism of motorcycle modelling and providing a foundation for future applications such as safety analysis or V2X message generation.
In the DACH region (Germany, Austria, and Switzerland), traffic signal control is commonly implemented using LISA+® signal logic files that process inputs from traffic detectors and public transport priority systems. However, reproducing such control systems in SUMO typically requires high manual effort, as the signal logic must be reimplemented at a low-level via the standard TraCI interface, or relies on GUI-based black-box tools. This reimplementation often introduces simplifications or approximations, reducing simulation fidelity and limiting the transferability of results across different scenarios and control setups. We present a SUMO- and Python-based platform that enables the direct import of LISA+® signal control logic and the integration of DACH-standard detector and telegram inputs into SUMO simulations. The platform directly uses LISA+® logic files and maps their signal groups, phase structures, and timing logic to SUMO traffic signal controllers. To drive the imported logic, we provide a traffic sensor input layer that maps simulated traffic information in SUMO to the detector and message inputs required by the control logic. By reproducing both the original signal logic and its associated sensor inputs, the proposed platform enables high-fidelity and fully reproducible simulation of DACH-standard signalised intersections without reformulating the SUMO signal controller, supporting both research and engineering workflows.
Activity-based models have emerged over the past four decades and have demonstrated a stronger capacity than traditional four-step models to capture the sequence of an individual’s daily activities, associated trips, and travel behavior. However, the state of the art shows that demand generation in activity-based models still relies heavily on detailed travel diary data, which require extensive data-collection efforts. This paper introduces an activity-based modeling framework developed for the hybridPy/SUMO simulation environment. The aim is to generate individual plans based on generally available data such as aggregated population data, vehicle ownership and other mobility-related data, as well as building-scale land-use data from open sources. The model enables to generate a detailed daily travel plan for each agent, in which the land-use type of the visited locations must match the respective activity type. In addition, the plans must meet multiple constraints, such as building capacity, individual daily travel time budgets, and opening hours, where available. The overall framework comprises three key sub-modules: (i) household synthesis, which reconstructs a synthetic household population; (ii) individual disaggregation, which disaggregates households into individual-level attributes; and (iii) daily diary and schedule generation, which produces complete activity-travel diaries with primary activities and optional secondary activities. For each sub-module, the key input data, parameters, and outputs are presented to enable users to construct daily travel-demand scenarios. This framework offers several advantages in terms of reliable, behavior-based travel demand modeling and practical applicability across different urban contexts. By presenting a step-by-step model development workflow from the user’s perspective, it enables scientists and transport planners to quickly and easily build large-scale microscopic travel-demand models without relying on disaggregated travel diary surveys or other location-specific statistics.
A path-accurate simulation of individual bus, tram and train trips through a certain area requires accurate route data. If this is not given, e.g. as shape file from the public transport company, a reconstruction in SUMO is possible by means of the existing “gtfs2pt” tool, which imports and processes public GTFS and OSM data. However, the reconstruction of true routes often fails due to missing or outdated shapes in OSM, or due to inaccurate polylines in the GTFS shapes file, which often includes only stop locations and no further waypoints. Then gtfs2pt searches the fastest path between stops, which mostly results in the true route, particularly for a tram or train line (if alternative pathes simply do not exist), but not in cases of bus lines within a dense road network. Therefor a map-matching approach based on GPS traces was recently proposed at the last SUMO Conference (by Schweppenhäuser et al., 2025). While these traces were given for Berlin in the corresponding GTFS shape file, this approach is not applicable to areas, e.g. Magdeburg, where they are not available in the local GTFS data. Hence, a more generic approach for bus routes is required and proposed in this article, which is in line with gtfs2pt in terms of using open data sources (GTFS, OSM) and intends to improve the optimal path searching. This is achieved by a) limiting the set of potential edges for bus routes according to reasonable edge criteria (e.g. used by other bus lines given from OSM, type and priority of edge, etc.), and b) searching the shortest path between all consecutive stops of a bus trip (i.e. not isolated between only two consecutive stops). As the latter was already applied and tested successfully for tram lines in a previous article (by Kaiser and Schmaus, 2025), it is adapted and tested more verbose for bus lines in this article. Finally, the results and computation time of this approach are compared with the ground-truth as well as the output of gtfs2pt by an example of public transport trips in Magdeburg, to evaluate accuracy and performance.
