VEHITS 2024 Abstracts

Area 1 - Connected Vehicles
Full Papers Short Papers
Area 2 - Intelligent Transport Systems and Infrastructure
Full Papers Short Papers
Area 3 - Intelligent Vehicle Technologies
Full Papers Short Papers
Area 4 - Data Analytics
Full Papers Short Papers
Area 5 - Smart Mobility and Sustainable Transport Services
Full Papers

Area 1 - Connected Vehicles

Full Papers
Paper Nr: 54
Title:

What Is the Right Bounding Box of a VRU Cluster in V2X Communication? How to Form a Good Shape?
Authors:

Leonardo Barbosa da Silva, Silas C. Lobo, Evelio G. Fernández and Christian Facchi
Abstract: Among the possible traffic members on a Vehicle-to-Everything network, the term Vulnerable Road User (VRU) is assigned e.g. to pedestrians and cyclists. The VRU Awareness Message (VAM) is used by VRUs to inform other users of their presence and ensure they are perceived in a traffic system. Since the number of VRUs in crowded areas might be very high, the over-the-air traffic might be overloaded. To reduce channel overload, VAMs offer a clustering feature in which VRUs with similar kinematics and positions can group themselves so that only one device transmits messages. The VRU Basic Service specification describes the cluster as a bounding box that must cover all its members using a geometric shape so that other vehicles in the vicinity can avoid colliding with the contained VRUs. This paper contributes to the standardization effort by introducing a data structure, the Cluster Map, for the clustering in the VRU Basic Service. Furthermore, this work is the first to suggest strategies for forming bounding box shapes. Simulation results show that each of the geometry types is useful in different situations, thus further research on the topic is advised.
Download
Short Papers
Paper Nr: 43
Title:

Optimal Velocity Model Based CACC Controller for Urban Scenarios
Authors:

Anas Abulehia, Reza Dariani and Julian Schindler
Abstract: To address the current high level of congestion, a connected vehicle system in the form of a platoon or Cooperative Adaptive Cruise Control (CACC) presents a promising solution. This system significantly reduces stop-and-go traffic, as well as fuel consumption. A Cooperative Adaptive Cruise Control (CACC) system comprises two or more closely-following vehicles traveling at a desired cruising velocity and distance headway. Compared with human drivers, such a system has the advantage of reducing inter-vehicle distance, making it a promising solution for mitigating traffic congestion as well as reducing aerodynamic drag, and fuel consumption. This work aims to introduce a new Cooperative Adaptive Cruise Control (CACC) based on the optimal velocity model in traffic dynamics. Several controllers for the introduced CACC system will be presented, particularly various versions of the linear quadratic controller. Simulation scenarios for these controllers will also be discussed.
Download
Paper Nr: 17
Title:

Protocol Interoperability for Intelligent Transportation Systems
Authors:

Jonas Vogt and Hans D. Schotten
Abstract: Transportation is an essential part of daily life. To ensure optimal transportation of people and goods, all stakeholders, including traffic participants, infrastructure providers, and service providers, must be interconnected. This interconnectivity enables efficient traffic management, energy and noise reduction, and safe driving. However, integrating different systems via various communication technologies and protocols can pose challenges to ensuring information quality, reliability, security, and privacy. As communication becomes a more critical safety factor for connected and automated driving, and more systems are involved in communication, ensuring data quality becomes increasingly crucial. This includes data availability, range, precision, privacy, and security. This paper presents an evaluation mechanism for assessing the quality of standardized protocols in terms of interoperability. It can be concluded that, although many protocols exist in the field of intelligent transportation systems, they transport similar information but differ in details that may impact the applications working with that information.
Download
Paper Nr: 30
Title:

Assessing Trustworthiness of V2X Messages: A Cooperative Trust Model Against CAM- and CPM-Based Ghost Vehicles in IoV
Authors:

Runbo Su, Yujun Jin and Ye-Qiong Song
Abstract: A number of V2X (Vehicle-to-Everything) messages are standardized by the European Telecommunication Standardization Institute (ETSI), such as CAM (Cooperative Awareness Message) and CPM (Collective Perception Message). Since road safety and traffic efficiency are on the basis of the assumption that correct and accurate V2V messages are shared, ensuring the trustworthiness of these V2X messages becomes an essential task in IoV (Internet of Vehicles) security. However, containing safety-related information makes V2X messages susceptible to malicious insider attacks from compromised vehicles after the PKI (Public Key Infrastructure) authentication step (Farran and Khoury, 2023), such as Ghost Vehicles (GV) (Gyawali and Qian, 2019), passively or actively reaching a ’ghost’ state in terms of communication, position, etc. By integrating CPS (Collective Perception Service) in the Veins simulator, our work aims to propose a trust assessment model in IoV against several types of CAM- and CPM-based GV to increase security. The simulation results provide a preliminary analysis of the feasibility of the proposed model and show the effectiveness in terms of assessing V2X messages’ trustworthiness.
Download
Paper Nr: 44
Title:

CARISMA: CAR-Integrated Service Mesh Architecture
Authors:

Kevin Klein, Pascal Hirmer and Steffen Becker
Abstract: The amount of software in modern cars is increasing continuously with traditional electric/electronic (E/E) architectures reaching their limit when deploying complex applications, e.g., regarding bandwidth or computational power. To mitigate this situation, more powerful computing platforms are being employed and applications are developed as distributed applications, e.g., involving microservices. Microservices received widespread adoption and changed the way modern applications are developed. However, they also introduce additional complexity regarding inter-service communication. This has led to the emergence of service meshes, a promising approach to cope with this complexity. In this paper, we present an architecture applying the service mesh approach to automotive E/E platforms comprising multiple interlinked High-Performance Computers (HPCs). We validate the feasibility of our approach through a prototypical implementation.
Download
Paper Nr: 61
Title:

A Prototype Preceding Vehicle Identification System Development and Field Evaluation
Authors:

Zeyu Mu, Guancheng Tu, Austin Shi, Kun Yang, Yixin Sun, Cong Shen and B. B. Park
Abstract: Preceding vehicle identification is crucial for establishing cooperative platooning. This paper presents the development of a prototype preceding vehicle identification system (PVIS) and its field evaluation for the assessment of commercial viability. We designed and assembled a prototype consisting of a processing unit (Jetson Nano board), a communication device (Wi-Fi dongle), a GPS unit, and a distance measurement sensor (Terabee sensor). The Jetson Nano integrates the SparkFun GPS-RTK-SMA unit, the Terabee time-of-flight sensor, and the Wi-Fi dongle. The PVIS prototype in the ego vehicle measures the distance to its preceding vehicle and receives the GPS data from potential preceding vehicles with the PVIS prototypes. With these, the PVIS in the ego vehicle determines the connectivity of the preceding vehicle. The field evaluation results showed that the prototype PVIS works as designed, and each successful identification takes about 5.3 seconds. However, it was found that the Terabee (time of flight) sensor, at times, did not properly measure distances, likely due to an angle issue caused by the roadway surface and vibration of the vehicle. We discussed how to overcome the challenges identified and enhance the prototype for successful commercialization.
Download

Area 2 - Intelligent Transport Systems and Infrastructure

Full Papers
Paper Nr: 31
Title:

Traffic State Estimation on Urban Roads Using Perception-Enriched Floating Car Data
Authors:

Moritz Schweppenhäuser, Karl Schrab, Robert Protzmann and Ilja Radusch
Abstract: Modern-day navigation systems by developers like Google© and TomTom© require user participation primarily in the form of Floating Car Data (FCD) for accurate Traffic State Estimation (TSE). However, to provide reliable information, systems rely on large road user participation of at least 5 %, which is only truly available to the big players. We propose a method to soften the participation requirement by utilizing modern perception sensors (e.g., radar, lidar, camera) of connected vehicles (CVs) to enrich the FCD set, compensating reduced data quantity with increased data quality. By using position and speed estimates of surrounding vehicles we increase the sample size and can additionally collect estimates of segments that are not traversed by CVs. To validate and assess the proposed system, we utilize Eclipse MOSAIC and conduct a simulation-based test series on the calibrated large-scale BeST scenario. Initial findings indicate improved estimation performance on selected road segments, especially at lower rates of market penetrations. In a network-wide investigation, we show that travel time estimates of the proposed method are often more accurate than conventional approaches, while also requiring smaller penetration rates.
Download
Paper Nr: 55
Title:

Spatial Performance Indicators to Evaluate Spatiotemporal Traffic Prediction
Authors:

Muhammad Farhan Fathurrahman and Sidharta Gautama
Abstract: Traffic prediction is vital for traffic management systems and helps enhance traffic management efficiency over a traffic network. Recently, spatiotemporal prediction models have been proposed that extend single traffic node temporal prediction. They employ the spatial context of the combined nodes in the urban network to improve prediction. However, the key performance indicators (KPI) of these methods are still limited to accuracy averaged over the full traffic network. They do not yet describe local spatiotemporal behaviour that can affect the traffic prediction accuracy in the traffic network. In this paper, we explore three spatial KPIs: Global Moran’s I, Geary’s C, and Getis-Ord General G to evaluate traffic flow prediction for freeway traffic networks. The study is conducted by evaluating traffic flow prediction results in the PeMSD8 dataset using spatiotemporal prediction and calculating different KPIs. Several synthetic scenarios based on the prediction results are created to showcase what the standard KPI cannot distinguish. The Global Moran’s I and Geary’s C can identify different levels of spatial autocorrelation and the Getis-Ord General G can distinguish spatial clustering in prediction results. The findings aim to improve the evaluation of different traffic prediction methods towards a better traffic management system.
Download
Short Papers
Paper Nr: 6
Title:

Examining the Impact of Weather, Temporal Factors, and User Traits on Multimodal Shared Micromobility Systems in Non-Urban Campus Environments: The MORE Sharing Case Study
Authors:

Maryna Pobudzei and Silja Hoffmann
Abstract: Although shared micromobility systems in cities have been extensively studied, their potential for non-urban settings such as university campuses and rural communities has not been explored much yet. This study aims to fill this gap by examining a multimodal shared micromobility service that offers various options through a single app, such as city bikes, e-bikes, e-cargo bikes, e-mopeds, and e-scooters. The study analyzed this campus-based system’s first four months, considering factors like weather, time, user demographics, pre-reservation duration, and vehicle types. Machine learning models like Negative Binomial Regression, Random Forests, Gradient Boosted Regression Trees, and Neural Networks were used to analyze the data. The study found that e-scooters were the most popular, followed by e-bikes. E-mopeds were used less but were reserved for more extended periods. Most trips were taken on weekdays, especially between 8 AM and 6 PM. Reservation numbers peaked in the first month, and subsequent months showed longer reservation durations and distances. Rain decreased trip numbers and distances but increased reservation durations. Reservations on Fridays, weekends, and holidays were shorter but covered more distance. Female users tended to travel longer distances. These findings can benefit similar non-urban environments, broadening the application of shared micromobility systems.
Download
Paper Nr: 16
Title:

Muli-Quay Combined Berth and Quay Crane Allocation Using the Cuckoo Search Algorithm
Authors:

Sheraz Aslam, Michalis P. Michaelides and Herodotos Herodotou
Abstract: This study investigates the combined berth allocation problem (BAP) and quay crane allocation problem (QCAP) while considering a multi-quay setting. First, a mixed integer linear programming mathematical model is developed based on various constraints and real port settings. Then, the multi-quay combined BAP and QCAP is solved using both the exact method and a metaheuristic optimization method, namely, the cuckoo search algorithm (CSA). This analysis pertains to a one-week planning scenario, utilizing data from a real port. The results of the comparative analysis show that the proposed CSA can provide a near-optimal solution (<1.02% from the optimal) at a fraction of the computational time (10 times faster), as compared to the exact solution. This makes it suitable for solving larger instances of the combined BAP and QCAP for bigger terminals and extended planning horizons.
Download
Paper Nr: 23
Title:

History-Based Road Traffic Anomaly Detection Using Deep Learning and Real-World Data
Authors:

Alexander Michielsen, Mohammadmahdi Rahimiasl, Ynte Vanderhoydonc and Siegfried Mercelis
Abstract: Detecting anomalies in road traffic, such as accidents and traffic jams, can provide various benefits to road users and road infrastructure managers, including optimal route planning, redirecting traffic flows, and reducing congestion caused by traffic accidents. Recently, many history-based traffic prediction deep learning methods have been developed to perform this task. These methods detect anomalous traffic by comparing the current traffic situation with a predicted one based on historical data. This paper investigates the possibility of detecting traffic anomalies using a novel combination of traffic prediction and graph anomaly detection algorithms, both using deep learning, in a real-world dataset of highways near Antwerp, Belgium. It first benchmarks configurations with different time resolutions of prediction algorithms in terms of accuracy. Then, a combined configuration including anomaly detection is benchmarked in terms of traffic anomaly detection accuracy. Furthermore, it examines which traffic features can contribute to anomaly detection e.g. speed, vehicle length. Finally, the entire system is tested on real-world traffic data containing anomalies. The results show a decreased anomaly detection performance when using both vehicle speed and length as features instead of only speed, and an increased performance when using larger time resolutions.
Download
Paper Nr: 42
Title:

Enhancing Individual Mobility: A Multistage Personalization Approach for Itinerary Planning in Multimodal Networks
Authors:

Alexandra Wins, Christoph Becker, Sascha Alpers, Lukas Kneis and Andreas Oberweis
Abstract: Individual mobility is an essential element of a prosperous society. Multimodal transportation can offer greater time, cost, and environmental efficiency than relying on a single mode of transport. Personalized itinerary planning is crucial to enhance the appeal of multimodal transport. Our proposed approach for recommending personalized itineraries tailors them by integrating diverse mobility preferences, routing services, and calibrating parameters of these services to provide individualized options. We optimize itineraries within the existing routing services and available data. The aim of this approach is to enhance travel experiences, making them more efficient, cost-effective, and aligned with each traveler’s unique needs and preferences. The approach was evaluated in a mid-sized German city by analyzing real-world mobility preferences, available routing services, and mobility providers. Personalization criteria relevant to the evaluation area were selected. A simulation was conducted, which demonstrated a 10.48% increase in travel utility when compared to the shortest path itinerary recommendation.
Download
Paper Nr: 46
Title:

Optimizing Traffic Adaptive Signal Control: A Multi-Objective Simulation-Based Approach for Enhanced Transportation Efficiency
Authors:

Sarah Salem and Axel Leonhardt
Abstract: This research aims to improve traffic flow efficiency, reduce congestion, and enhance the overall performance of the transportation system for different road users, while keeping in mind the ease of implementation of the provided approach. That is achieved by optimizing the stage length parameter in the VAP files for VISSIM using ParMOO, a powerful optimization tool. The VAP files contain crucial information about traffic signal control logic, including signal timings, stage durations, and cycle lengths. The maximum stage length parameter within VAP files represents the maximum allowable time for a particular traffic signal stage before transitioning to the next stage. Optimizing this parameter can significantly impact traffic performance by reducing delays and improving overall traffic flow efficiency. Average delays for passenger cars and pedestrians are chosen as objective functions to be minimized. Sensitivity analysis is employed to validate the optimized solutions. Comparing the traffic performance measures using the optimized VAP files with the base case, we found that the optimized solutions consistently outperformed the observed performance. The research contributes by utilizing the ParMOO algorithm and integrating it within VISSIM software, enabling researchers to readily apply the methodology and advance the field of traffic signal control with practical and industry-relevant solutions.
Download
Paper Nr: 60
Title:

Dynamic Prices in Ride-Sharing Scenarios
Authors:

Lech Duraj and Grzegorz Herman
Abstract: We describe a dynamic pricing strategy applicable to ride-sharing scenarios in public transportation services. The strategy incorporates data about relation popularity and price acceptance rates. Crucially, it captures interdependencies between tickets for different relations served by a single vehicle, and thus is able to balance out locally optimal pricing with the expected future evolution of vehicle occupancy state. Based on historical data about a real-world ride-sharing operator, we demonstrate that the proposed method is robust to imperfections in the input data, and estimate it to be more profitable than both fixed-price strategies (even theoretically optimal), and an actual dynamic pricing strategy prescribed by business experts.
Download
Paper Nr: 47
Title:

Personalisation in Mobility-as-a-Service: Where We Are and How to Move Forward
Authors:

Kamaldeep Singh Oberoi
Abstract: Within urban mobility ecosystem, Mobility-as-a-Service (MaaS) has come up as a promising approach to promote sustainable modes of transport and increase the attractiveness of public and shared multimodal mobility. It aims to become a viable alternative to personal cars for door-to-door trips. The long-term objective of MaaS is to change the people’s travel behaviour by nudging them to make sustainable choices. However, changing people’s travel behaviour is not easy and MaaS has to provide personalised mobility services, catering to the needs of each individual user, in order to be considered as convenient as a personal car. In this paper, we look at the existing literature on personalisation in MaaS proposed by the research community as well as different private MaaS service providers. This brief literature review helps in better understanding the current trends on personalisation and highlights certain limitations in the way it is incorporated within existing MaaS solutions. Based on these limitations, we discuss certain challenges which need to be resolved in order to improve MaaS in the future. These challenges present interesting research directions towards the development of personalised sustainable urban mobility ecosystem.
Download
Paper Nr: 73
Title:

Infrastructure-Based Communication Trust Model for Intelligent Transportation Systems
Authors:

Malek Lachheb, Rihab Abidi, Nadia Ben Azzouna and Nabil Sahli
Abstract: Intelligent Transportation Systems (ITS) aim to enhance traffic management through Vehicle-to-Vehicle (V2V), Vehicle-to-Infrastructure (V2I) and Infrastructure-to-Infrastructure (I2I) communications. However, the wireless medium and dynamic nature of these networks expose them to security threats from faulty nodes or malicious attacks. While cryptography-based mechanisms provide security against outsider attacks, the network remains vulnerable to attacks from legitimate but malicious nodes. Trust models have hence been proposed to evaluate node and data credibility to make informed security decisions. Existing models are either vehicle-centric with limited stability due to mobility or infrastructure-based with risks of single points of failure. This paper proposes a self-organizing, infrastructure-based trust model for securing ITS communication leveraging Smart Roadside Signs (SRSs). The model introduces a trust-based clustering algorithm using a fuzzy-based Dempster Shafer Theory (DST). This eliminates dependence on external trusted authorities while enhancing stability through infrastructure oversight. The decentralized trust formation and adaptive clustering balance security assurance with scalability. The results of the simulations show that our model is resilient against on-off attack, packet drop attack, jamming attack, bad-mouthing attack and collusion attack.
Download

Area 3 - Intelligent Vehicle Technologies

Full Papers
Paper Nr: 13
Title:

Bayesian Network for Analysis and Prediction of Traffic Congestion Using the Accident Data
Authors:

Kranthi K. Talluri and Galia Weidl
Abstract: Traffic congestion has become a significant concern regarding social safety and economic impact. Understanding the relationship between congestion and accidents is vital in providing the patterns to the Traffic Management System to mitigate the congestion as early as possible. Furthermore, traffic accidents lead to property damage, casualties, and increased congestion levels. So, a lot of research is going on to tackle this problem of accidents and congestion. This paper proposes a Bayesian Network (BN) to predict and analyze the factors of the probability of traffic congestion using accident data. A novel technique of labeling the congestion is being introduced, namely the formula-based and hotspot-based approaches, utilizing the accident dataset. Different scenarios are developed to understand the patterns causing congestion, and two classification models are used to evaluate the performance of the BN model. Model results are compared with different machine learning models. Results show that the proposed model outperforms in terms of accuracy and precision. It shows comparative performance concerning other machine learning algorithms.
Download
Paper Nr: 14
Title:

Detecting Edge Cases from Trajectory Datasets Using Deep Learning Based Outlier Detection
Authors:

Marcel Sonntag, Lennart Vater, Roman Vuskov and Lutz Eckstein
Abstract: The biggest challenge to overcome for automated vehicles is to prove their safety, as these vehicles are solely responsible for the passengers’ safety. The scenario-based testing approach promises an efficient safety validation procedure by only testing the safety in relevant scenarios. An open question is how to select the relevant scenarios for testing. So-called edge cases are frequently named in the automated driving domain to be important scenarios for testing automated vehicles. However, it is not an easy task to define what an edge case is and to find and validate them. In this work, we present a novel data-driven approach to finding edge cases in trajectory datasets using deep learning-based outlier detection. We develop a method that calculates embeddings for driving scenarios in a two-stage process. In the dimensionally reduced embedding space, outliers represent potential edge cases. We apply the approach to the exiD dataset and find potential edge cases. For validation, we present the found potential edge cases to a group of experts. The experts validate that the approach is capable of detecting edge cases in trajectory datasets.
Download
Paper Nr: 20
Title:

Performance Evaluation of a ROS2 Based Automated Driving System
Authors:

Jorin Kouril, Bernd Schäufele, Ilja Radusch and Bettina Schnor
Abstract: Automated driving is currently a prominent area of scientific work. In the future, highly automated driving and new Advanced Driver Assistance Systems will become reality. While Advanced Driver Assistance Systems and automated driving functions for certain domains are already commercially available, ubiquitous automated driving in complex scenarios remains a subject of ongoing research. Contrarily to single-purpose Electronic Control Units, the software for automated driving is often executed on high performance PCs. The Robot Operating System 2 (ROS2) is commonly used to connect components in an automated driving system. Due to the time critical nature of automated driving systems, the performance of the framework is especially important. In this paper, a thorough performance evaluation of ROS2 is conducted, both in terms of timeliness and error rate. The results show that ROS2 is a suitable framework for automated driving systems.
Download
Paper Nr: 22
Title:

Efficient Deployment of Neural Networks for Thermal Monitoring on AURIX TC3xx Microcontrollers
Authors:

Christian Heidorn, Frank Hannig, Dominik Riedelbauch, Christoph Strohmeyer and Jürgen Teich
Abstract: This paper proposes an approach for efficiently deploying neural network (NN) models on highly resourceconstrained microcontroller architectures, particularly AURIX TC3xx microcontrollers. Here, compression and optimization techniques of the NN model are required to reduce execution time while maintaining accuracy on the target microcontroller. Furthermore, especially on AURIX TriCores that are frequently used in the automotive domain, there is a lack of support for automatic conversion and deployment of pretrained NN models. In this work, we present an approach that fills this gap, enabling the conversion and deployment of so-called thermal neural networks on AURIX TC3xx microcontrollers for the first time. Experimental results on three different NN types show that, when pruning of convolutional neural networks is applied, we can achieve a speedup of 2.7× compared to state-of-the-art thermal neural networks.
Download
Paper Nr: 24
Title:

Reinforcement Learning and Optimal Control: A Hybrid Collision Avoidance Approach
Authors:

Simon Gottschalk, Matthias Gerdts and Mattia Piccinini
Abstract: In this manuscript, we consider obstacle avoidance tasks in trajectory planning and control. The challenges of these tasks lie in the nonconvex pure state constraints that make optimal control problems (OCPs) difficult to solve. Reinforcement Learning (RL) provides a simpler approach to dealing with obstacle constraints, because a feedback function only needs to be established. Nevertheless, it turns out that often we get a long lasting training phase and we need a large amount of data to obtain appropriate solutions. One reason is that RL, in general, does not take into account a model of the underlying dynamics. Instead, this technique relies solely on information from the data. To address these drawbacks, we establish a hybrid and hierarchical method in this manuscript. While classical optimal control techniques handle system dynamics, RL focuses on collision avoidance. The final trained controller is able to control the dynamical system in real time. Even if the complexity of a dynamical system is too high for fast computations or if the training phase needs to be accelerated, we show a remedy by introducing a surrogate model. Finally, the overall approach is applied to steer a car on a racing track performing dynamic overtaking maneuvers with other moving cars.
Download
Paper Nr: 27
Title:

Using Attention Mechanisms in Compact CNN Models for Improved Micromobility Safety Through Lane Recognition
Authors:

Chinmaya Kaundanya, Paulo Cesar, Barry Cronin, Andrew Fleury, Mingming Liu and Suzanne Little
Abstract: The use of personal transportation devices such as e-bikes and e-scooters (micromobility) necessitates the development of improved safety support systems using highly-accurate, real-time lane recognition. However, the constrained operating environment, both computationally and physically, on such devices restricts the applicability of existing sensor-based solutions. One option is to leverage vision-based systems and AI models. However, these are typically built using high-spec processors and high-memory platforms and the models need to be adapted to low-spec platforms such as microcontrollers. A significant barrier to the development and evaluation of these potential solutions is the lack of lane recognition datasets that focus on the first-person (rider) perspective. We contribute a lane recognition dataset of micromobility first-person perspective images from e-mobility rides. This dataset is utilized to assess the impact of channel and spatial attention on compact CNN models, driven by the aim to maximize utilization through the addition of cost-effective operations like these attention mechanisms, which introduce only a modest increase in the number of parameters. We find that adding channel and spatial attention can improve the performance of the standard compact CNN classification models and specifically that adding the spatial branch improves the performance of the model with channel attention. The MobileNetV3 model with the fewest parameters among those with channel plus spatial attention maintained high overall performance. Our code and dataset are publicly accessible at: https://github.com/Luna-Scooters/Compact-Attention-based-CNNs-on-MLRD.
Download
Paper Nr: 41
Title:

Improving Lane Level Dynamics for EV Traversal: A Reinforcement Learning Approach
Authors:

Akanksha Tyagi, Meghna Lowalekar and Praveen Paruchuri
Abstract: Emergency vehicles (EVs) perform a critical task of attending medical emergencies and delay in their operations can result in loss of lives to long term or permanent health implications. Therefore, it is very important to design strategies that can reduce the delay of EVs caused by slow moving traffic. Most of the existing work on this topic focuses on assignment and dispatch of EVs from different base stations to hospitals or finding the appropriate routes from dispatch location to hospital. However, these works ignore the effect of lane changes when EV is travelling on a stretch of a road. In this work, we focus on lane level dynamics for EV traversal and showcase that a pro-active picking of lanes can result in significant reductions in traversal time. In particular, we design a Reinforcement Learning (RL) model to compute the most optimal lane for an EV to travel at each timestep. We propose RLLS (Reinforcement Learning based Lane Search) algorithm for a general purposes EV traversal problem and perform a series of experiments using the well-known traffic simulator SUMO. Our experimentation demonstrates that our model outperforms the default SUMO algorithm and is also significantly better than the existing state-of-the-art heuristic approach BLS (Best Lane Search) strategy in normal traffic conditions. We also simulate worst case scenarios by introducing slowed down vehicles at regular time intervals into the traffic and observe that our model generalizes well to different traffic scenarios.
Download
Paper Nr: 56
Title:

Archetypes of Carsharing Relocation Algorithms: A Perspective on Problem Space, Solution Space and Evaluation
Authors:

Christoph Prinz, Mathias Willnat, Tim-Benjamin Lembcke and Lutz M. Kolbe
Abstract: Shared vehicle services like carsharing enable society to achieve a more favorable tradeoff between the societal cost and the individual benefits of physical mobility. To realize this value proposition, numerous carsharing types with unique constraints have emerged. A key challenge of making such offerings available, is the real time coordination of fleet supply tailored to short term customer demands. Researchers developed frameworks, algorithms, and decision support systems to address the corresponding vehicle relocation challenge on strategic, tactical, and operational level. However, subsequent vehicle relocation knowledge must be systematized to ensure that subsequent insights can be reused and further developed. Consequently, we develop a holistic taxonomy for vehicle relocation algorithms in carsharing, which contributes to current research by (1) providing consistent descriptions and analyses of vehicle relocation problems, solutions, and evaluation approaches, (2) identifying archetypes of algorithm instances, and (3) guiding research to work on subsequent research gaps. As a result, we substantiate a resilient and validated relationship between vehicle relocation’s problem and solution space.
Download
Paper Nr: 70
Title:

Efficient Real-Time Obstacle Avoidance Using Multi-Objective Nonlinear Model Predictive Control and Semi-Smooth Newton Method
Authors:

Mostafa Emam, Thomas Rottmann and Matthias Gerdts
Abstract: This work discusses the theory and methodology of applying Nonlinear Model Predictive Control (NMPC) in an efficient manner to achieve real-time path planning and obstacle avoidance for autonomous vehicles. First, we explain the optimization problem formulation and the numerical solution approach using a semi-smooth Newton method adapted for nonlinear problems. Then, an MPC path planning problem is described in terms of the vehicle model, the controller design, and the mathematical representation of obstacles as proper system constraints. Afterwards, the developed controller is numerically evaluated for different vehicle models in a simulated environment to dynamically assess its flexibility and real-time performance, which serves as a prerequisite to deferred real-life testing.
Download
Short Papers
Paper Nr: 12
Title:

Autonomous Driving Validation and Verification Using Digital Twins
Authors:

Heiko Pikner, Mohsen Malayjerdi, Mauro Bellone, Barış C. Baykara and Raivo Sell
Abstract: With the introduction of autonomous vehicles, there is an increasing requirement for reliable methods to validate and verify artificial intelligence components that are part of safety-critical systems. Validation and verification (V&V) in real-world physical environments is costly and unsafe. Thus, the focus is moving towards using simulation environments to perform the bulk of the V&V task through virtualization. However, the viability and usefulness of simulation is very dependent on its predictive value. This predictive value is a function of the modeling capabilities of the simulator and the ability to replicate real-world environments. This process is commonly known as building the digital twin. Digital twin construction is non-trivial because it inherently involves abstracting particular aspects from the multi-dimensional real world to build a virtual model that can have useful predictive properties in the context of the model-of-computation of the simulator. With a focus on the V&V task, this paper will review methodologies available today for the digital twinning process and its connection to the validation and verification process with an assessment of strengths/weaknesses and opportunities for future research. Furthermore, a case study involving our automated driving platforms will be discussed to show the current capabilities of digital twins connected to their physical counterparts and their operating environment.
Download
Paper Nr: 21
Title:

Real-Time Lane Departure Detection Using Google Routes
Authors:

Nafisa Z. Tasnim, Attiq U. Zaman and M. I. Hayee
Abstract: Our previously developed Lane Departure Detection and Warning System (LDWS) used a standard GPS receiver and two algorithms to detect an unintentional lane departure. The first algorithm generated the Road Reference Heading (RRH) from a vehicle’s past trajectories, while the second algorithm predicted lane departure in real time using RRH. A significant limitation of this system is the dependency on past trajectories. A vehicle must travel on the road at least once in the past to use that trajectory for RRH generation needed for future lane departure detection. To avoid dependency on past trajectories, this work uses Google routes instead of past trajectories to extract the RRH of any given road. We also compared the RRH generated from a Google route with that of a past trajectory and found both RRHs to be comparable indicating that our LDWS does not need to rely on RRH from past trajectories. To evaluate the accuracy of lane departure detection using Google RRH, we performed many field tests on a freeway. Our field test results show that our LDWS can accurately detect all lane departures on long straight sections of the freeway irrespective of whether the RRH was generated from a Google route or past trajectory.
Download
Paper Nr: 25
Title:

VRU-Net: Convolutional Neural Networks-Based Detection of Vulnerable Road Users
Authors:

Abdelhamid Mammeri, Abdul J. Siddiqui and Yiheng Zhao
Abstract: Research work on object detection for transportation systems have made considerable progress owing to the effectiveness of deep convolutional neural networks. While much attention has been given to object detection for automated vehicles (AVs), the problem of detecting them at road intersections has been underexplored. Specifically, most research work in this area have, to some extent, ignored vulnerable road users (VRUs) such as persons using wheelchairs, mobility scooters, or strollers. In this work, we seek to fill the gap by proposing VRU-Net, a CNN-based model designed to detect VRUs at road intersections. VRU-Net first learns to predict a VRUMask representing grid-cells in an input image that are highly probable of containing VRUs of interest. Based on the predicted VRUMask, regions/cells of interest are extracted from the image/feature maps and fed into the further layers for classification. In this way, we greatly reduce the number of regions to process when compared to popular object detection works such as Faster RCNN and the likes, which consider anchor points and boxes all over the image. The proposed model achieves a speedup of 4.55× and 13.2% higher mAP when compared to the Faster RCNN. Our method also achieves 9% higher mAP, comparing to SSD (Single Shot Multibox Detection).
Download
Paper Nr: 32
Title:

Human Centric Intersection Crossing Control Using C-ITS Information
Authors:

Abhishek Kalose, Dehlia Willemsen and Jochem Brouwer
Abstract: One of the challenges in automated transport of passengers is comfort and trust of the passengers during their travel. This especially comes into play when the automated driving vehicle has to react to external influences from e.g. traffic lights. Much research has been put into recognizing the traffic lights and their state with on-board sensors and into optimal traffic regulation at signalized intersections, however, optimal vehicle control for passenger trust and comfort seems lacking. To advance in this area, in the EU-project SHOW, an in-car traffic light control algorithm was designed and implemented in TNO’s carlab to be evaluated with passengers. The outcome of experimental tests with a limited number of participants as passenger, seems promising and will be a basis for future research on this topic. The implemented approach was found to be an adequate methodology to tune the intersection crossing functionality of an automated vehicle in order to optimize comfort and increase passenger trust.
Download
Paper Nr: 37
Title:

Making Radar Detections Safe for Autonomous Driving: A Review
Authors:

Tim Brühl, Lukas Ewecker, Robin Schwager, Tin S. Sohn and Sören Hohmann
Abstract: Radar sensors rank among the most common sensors used for highly automated driving functions due to their solid distance and velocity measurement capabilities and their robustness against adversarial environmental conditions. However, radar point clouds are noisy and must therefore be filtered. This work reviews current research with the aim to make radar detections usable for safe perception functions which require a guarantee for correctness of the measured environmental representation. The impact on radar errors on the distinct downstream tasks is explained. Besides, the term of safety for automated driving functions is illuminated under consideration of the current standards and state-of-the-art research interpreting these standards is presented. Furthermore, this work discusses safe radar signal processing and filtering, approaches to enrich radar data points by information fusion, e.g. from cameras and other radars, and development tools for safe radar-based perception functions. Finally, next steps on the way towards safety guarantees for radar sensors are identified.
Download
Paper Nr: 39
Title:

A Qualitative Review of Full Sized Autonomous Racing Vehicle Sensors: A Case Study
Authors:

Manuel Mar and Eric Dietz
Abstract: This paper explores into the challenges and advancements encountered in the development and operation of full-sized autonomous cars built for motorsports competitions. Concentrating on a qualitative examination of the sensor configuration, structure, and real-time assessment of vehicle platforms in the Indy Autonomous Challenge and Roborace. The scrutiny is centered on recent years’ research and the vehicles’ performance in demanding conditions, systematically highlighted and summarized in this paper. The analysis furnishes a more concise and condensed comprehension of the prevailing trends in such competitions, offering insights into the future of autonomy in the coming years.
Download
Paper Nr: 50
Title:

Introducing Scaled Model Development to on-Sight Automatic Train Operation
Authors:

Tobias Hofmeier and Martin Cichon
Abstract: Rail systems are often not an economic option in terms of flexibility and cost in intermodal competition. To address this issue, there is a push towards implementing automation and digitalization components. In recent years, there has been a strong focus on on-sight automated train operation systems. As the applications move beyond protected areas such as metro systems, to complex on-sight driving scenarios, the demands on system development, verification, and validation increase. Methods from the automotive industry are well known for overcoming these challenges with virtual development and final field testing. Fundamentally different operating conditions prevent sufficient field testing, as rail infrastructure and vehicles are difficult to procure for development and testing purposes. In science and research, scaled models are being promoted for similar problems. These models allow for simulations to be verified and favorable estimates to be made. This paper demonstrates the possibility of using scaled model methods for the development of on-sight automatic train operation (ATO) functions. A demonstrator of a highly automated shunting locomotive is being built as a scaled model and equipped with sensors for environment detection and localization as well as communication interfaces. The feasibility of ATO functions in the scaled model is demonstrated using defined use cases.
Download
Paper Nr: 51
Title:

Multi-Pedestrian Tracking and Map-Based Intention Estimation for Autonomous Driving Scenario
Authors:

Ali Dehghani and Lucila P. Studencki
Abstract: Pedestrian intentions estimation and tracking have become essential for the development of autonomous vehicles (AVs). The vehicles need to be aware of pedestrians to avoid fatalities even in complex urban traffic. This requires understanding the most probable trajectory of pedestrians to accordingly plan the vehicle’s maneuvers. This complex task requires modeling how multiple pedestrians interact with each other and move depending on their environment. This paper employs a Gaussian Mixture Probability Hypothesis Density Filter, enhanced by the Generalized Potential Field Approach (GMPHD-GPFA), to simultaneously track multiple pedestrians and determine and predict their behavior seconds ahead. The model used considers the static environment of the pedestrians to estimate their intentions and improve prediction accuracy. The paper evaluates both the tracking efficiency of the algorithm and its capability to predict the intentions of multiple pedestrians.
Download
Paper Nr: 52
Title:

Knowledge-Based Approach to Generate Scenarios for Testing Highly Automated On-Sight Train Operations
Authors:

Lucas Greiner-Fuchs and Martin Cichon
Abstract: Scenario-based test methods are cumulatively used for developing and testing highly automated railway vehicles, similar to the automotive industry. However, due to significant differences between the two technologies, existing approaches in the automotive sector cannot be directly applied to railways. Therefore, it is necessary to develop revised and new processes and methods that are tailored to the rail sector. The primary step in scenario-based testing is to set up appropriate test scenarios. A significant challenge faced by the rail industry is the limited availability of measured data from actual railway operations. For this reason, knowledge-based data sources need to be primarily used and considered in the scenario generation process. This paper presents a basic approach to define sufficient quantity of test scenarios for highly automated railway vehicles, using as an example a sensor-supported system for on-sight train operation. The approach uses the system definition of the automated system as input, includes the operational design domain, and considers railway-specific data through formalities and knowledge sources. Scenarios are then systematically derived in three steps: description, combination, and derivation. In the end, a set of testable scenarios is generated that can be used for virtual and real field testing of automated train operations.
Download
Paper Nr: 53
Title:

Connected Vehicle Perception Monitoring: A Runtime Verification Approach for Enhanced Autonomous Driving Safety
Authors:

Redge M. Castelino, Karina Rothemann, Arne Lamm and Axel Hahn
Abstract: Modern autonomous vehicles rely heavily on complex sensor systems for perception tasks, including Advanced Driver Assistance Systems and Autonomous Driving Systems. Accurate sensor perception is essential to ensure the safety of these systems, especially as the level of automation increases. External sensors from the infrastructure or other vehicles can provide useful information to verify the trustworthiness of on-vehicle sensors using V2X communication. This paper presents a novel concept of runtime monitoring to verify the performance of ADS perception systems, taking advantage of the design diversity of connected vehicles and infrastructure based perception sensors in Intelligent Transportation Systems. The proposed approach uses standardised V2X services, such as Collective Perception Service and Location Service from connected participants to estimate a reliable common environment model (CEM) of the driving situation. The established CEM can be used to evaluate the quality of perception of individual road participants during operation, allowing detection and mitigation of system malfunctions of a connected vehicle perception system and enhancing road safety in connected environments. We also discuss open design questions with respect to the perception runtime monitor concept.
Download
Paper Nr: 57
Title:

Beep, Bleep, Oops! A Discussion on the Misuse of Advanced Driver Assistance Systems (ADAS) and the Path Moving Forward
Authors:

Oscar Oviedo-Trespalacios
Abstract: The potential of Advanced Driver Assistance Systems (ADAS) to enhance road safety and driver comfort is significant. However, its realization can be compromised by driver misuse. This paper discusses the misuse of ADAS, defined as the suboptimal, inappropriate, or incorrect utilization of these systems. Such misuse not only diminishes their safety benefits but also poses new risks. In this paper, I argue that misuse encompasses at least three distinct behaviours: Non-Use When Beneficial, Overuse, and Non-Compliant Use. Each presents unique challenges to leveraging ADAS’s full safety capabilities. Through an analysis of these behaviours, the paper aims to shed light on the underlying reasons for ADAS misuse and its implications for road safety and vehicle efficiency. The study underscores the importance of addressing these issues through the development of more effective ADAS technologies, comprehensive education programs, and other interventions tailored to encourage correct usage. By exploring the specific pathways of misuse and their impact on road safety, this research contributes to the broader understanding of how to maximize the benefits of ADAS, ensuring they serve their intended purpose of making roads safer for all users.
Download
Paper Nr: 58
Title:

Simulation-Based Performance Evaluation of 3D Object Detection Methods with Deep Learning for a LiDAR Point Cloud Dataset in a SOTIF-related Use Case
Authors:

Milin Patel and Rolf Jung
Abstract: Safety of the Intended Functionality (SOTIF) addresses sensor performance limitations and deep learning-based object detection insufficiencies to ensure the intended functionality of Automated Driving Systems (ADS). This paper presents a methodology examining the adaptability and performance evaluation of the 3D object detection methods on a LiDAR point cloud dataset generated by simulating a SOTIF-related Use Case. The major contributions of this paper include defining and modeling a SOTIF-related Use Case with 21 diverse weather conditions and generating a LiDAR point cloud dataset suitable for application of 3D object detection methods. The dataset consists of 547 frames, encompassing clear, cloudy, rainy weather conditions, corresponding to different times of the day, including noon, sunset, and night. Employing MMDetection3D and OpenPCDET toolkits, the performance of State-of-the-Art (SOTA) 3D object detection methods is evaluated and compared by testing the pre-trained Deep Learning (DL) models on the generated dataset using Average Precision (AP) and Recall metrics.
Download
Paper Nr: 59
Title:

Heuristic Optimal Meeting Point Algorithm for Car-Sharing in Large Multimodal Road Networks
Authors:

Julien Baudru and Hugues Bersini
Abstract: This article introduces a new version of the car-pooling problem (CPP). This involves defining rendezvous or meeting point in such a way that the travel times of the users are fair, this problem shares similarities with the problem of finding the optimal meeting point (OMP) in a graph. We propose a heuristic algorithm to solve the OMP problem in this new context and compare its results with those of the exact solution algorithm, showing its low error rate and short runtime. Finally, we propose some exploratory directions for future research.
Download
Paper Nr: 63
Title:

Modelling Implicit and Explicit Communication Between Road Users from a Non-Cooperative Game-Theoretic Perspective: An Exploratory Study
Authors:

Isam Bitar, Albert S. Crusat and David Watling
Abstract: Road user interaction is a fertile avenue for communication between road users, be it implicit communication or explicit signals sent with the intent to convey information. To date, most literature on characterising and modelling communication between road users has focussed on cooperative paradigms and concepts of shared goals enforced globally on communicating agents. In this paper, we argue that non-cooperative game theory can be used to characterise and model effective and mutually beneficial communication between road users. We demonstrate that non-cooperative game theory can produce meaningful improvements in payoffs and interaction safety for both the sender and recipient of communication as an emergent phenomenon.
Download
Paper Nr: 64
Title:

Geographical Self-Organizing Map Clustering in Large-Scale Urban Networks for Perimeter Control
Authors:

Maha Elouni, Hesham A. Rakha, Monica Menendez and Hossam M. Abdelghaffar
Abstract: Traffic congestion in urban areas presents a major challenge to efficient transportation systems. Recent advancements in traffic management provide promising solutions, with perimeter control emerging as a technique to tackle network-wide congestion. However, it is crucial to identify geographically connected homogeneously congested areas for effective implementation. This research explores the application of clustering techniques, particularly geographical self-organizing maps (GeoSOM), to identify spatially connected and homogeneously congested areas within transportation networks. While GeoSOM has found applications across various domains, its adaptation to transportation networks for congestion clustering is novel. This study introduces and implements an adaptation of the GeoSOM algorithm tailored for the large-scale urban environment of downtown Los Angeles. Its performance is assessed through a comparative evaluation with two other clustering algorithms, namely DBSCAN and K-means. The results demonstrate that GeoSOM surpasses other clustering algorithms, exhibiting improvements of up to 43% in traffic density variance, up to 61% in the spatial quantization error, and 15% in the quantization error. This finding demonstrates that the proposed clustering algorithm is effective in identifying a spatially homogeneous congested area within a large-scale transportation network.
Download
Paper Nr: 66
Title:

Using V2X-Information for Trajectory Prediction at Urban Intersections
Authors:

Michael Klöppel-Gersdorf and Thomas Otto
Abstract: Crossing an urban intersection is one of the major challenges in automated/autonomous driving. This is due to a manifold of possible interactions with other traffic participants. In this paper, we propose a trajectory prediction service based on historical V2X-information gathered from Cooperative Awareness Messages (CAMs)/Basic Safety Messages (BSMs). The service allows connected vehicles to more easily navigate the intersection by identifying possibly critical encounters, especially with traffic participants which are not covered by the vehicle’s sensors. In comparison with approaches relying on video or other sensor data sources, this has the advantage that Road-Side Units (RSUs), which are used for Vehicle-to-Everything (V2X) communication, are more and more available at public intersections, e.g., due to equipment rollouts all over Europe in projects like C-Roads. The prediction service introduced in this paper is a first step in ongoing research and will act as a baseline for further projects, where additional sensors and also more involved prediction algorithms, e.g., based on neural networks, will be considered.
Download
Paper Nr: 69
Title:

Introducing Flowride® Logger, an Onboard Data Collection Framework for Commercial Automated Vehicles
Authors:

Kerem Par, Ali U. Peker, Reza Ghahremaninejad, Ali Ozcan and Ertan Sahin
Abstract: Rapid growth in the development, implementation, testing and deployment of Automated Vehicles (AV) in recent years highlights the global eagerness toward safer, cost-effective, cleaner and accessible transportation solutions. Reaching the promises requires continuous improvements in the AV software design and the preparation of pipelines to share that progress with the public and road authorities. An AV onboard data recording system plays a crucial role in delivering such information. In this work, we introduce Flowride® logger, a combined continuous and event-based data collection framework for real deployed AVs. Our approach considers two objectives to accomplish: 1- To aid software engineering efforts with real data from the deployed vehicle for data-driven development procedure. 2- To record, store, and share data with third party and road authorities for safety purposes and incident reports. Flowride® logger framework performance was discussed using observation from its implementation on the automated e-ATAK vehicle. This 8-meter electric-powered bus is part of the public transportation fleet of Stavanger, Norway. The experimental results demonstrate the effectiveness of Flowride® logger as a means for data collection from deployed AVs.
Download
Paper Nr: 72
Title:

Challenges of Remote Driving on Public Roads Using 5G Public Networks
Authors:

Adrien Bellanger, Michael Klöppel-Gersdorf, Joerg Holfeld, Lars Natkowski and Thomas Otto
Abstract: Teleoperation, in the form of remotely controlling a vehicle (remote driving), is an important bridging technology until fully autonomous vehicles become available. Currently, there are manifold activities in this area driven by public transport companies, which implement solutions to offer first commercial teleoperation activities on the road. On the other hand, scientific reports of these solutions are hard to come by. In this paper, we propose a potential implementation for remote driving in 5G based public networks. We describe our insights from real world test drives on public roads and discuss possible challenges and suggest solutions.
Download
Paper Nr: 74
Title:

Parameter Adjustments in POMDP-Based Trajectory Planning for Unsignalized Intersections
Authors:

Adam Kollarčík and Zdeněk Hanzálek
Abstract: This paper investigates the problem of trajectory planning for autonomous vehicles at unsignalized intersections, specifically focusing on scenarios where the vehicle lacks the right of way and yet must cross safely. To address this issue, we have employed a method based on the Partially Observable Markov Decision Processes (POMDPs) framework designed for planning under uncertainty. The method utilizes the Adaptive Belief Tree (ABT) algorithm as an approximate solver for the POMDPs. We outline the POMDP formulation, beginning with discretizing the intersection’s topology. Additionally, we present a dynamics model for the prediction of the evolving states of vehicles, such as their position and velocity. Using an observation model, we also describe the connection of those states with the imperfect (noisy) available measurements. Our results confirmed that the method is able to plan collision-free trajectories in a series of simulations utilizing real-world traffic data from aerial footage of two distinct intersections. Furthermore, we studied the impact of parameter adjustments of the ABT algorithm on the method’s performance. This provides guidance in determining reasonable parameter settings, which is valuable for future method applications.
Download
Paper Nr: 15
Title:

Performance Metric for Horn and Brake Automation on Mainline Trains
Authors:

Rustam Tagiew and Christian Klotz
Abstract: This paper argues for the introduction of a mainline rail oriented end-user performance metric for driver-replacing on-board perception systems. Perception at the head of a train is analysed and divided into several subfunctions. This paper presents a preliminary submetric for the obstacle detection subfunction, focusing on false-negatives. To the best of the authors’ knowledge, there is no other such proposal for rail on-board perception systems. A set of submetrics for the subfunctions should facilitate the end-user oriented comparison of perception systems and guide the measurement of human driver performance. It should also be useful for a standardised predictive assessment of the number of accidents for a given perception system in a given operational design domain. In particular, for the proposal of the obstacle detection submetric, practitioners among the readership are invited to provide their feedback and quantitative information to the authors. In addition to the interim feedback, the analysis results of the full feedback will be published later.
Download
Paper Nr: 29
Title:

Improve Bounding Box in Carla Simulator
Authors:

Mohamad M. Chaar, Jamal Raiyn and Galia Weidl
Abstract: The CARLA simulator (Car Learning to Act) serves as a robust platform for testing algorithms and generating datasets in the field of Autonomous Driving (AD). It provides control over various environmental parameters, enabling thorough evaluation. Development bounding boxes are commonly utilized tools in deep learning and play a crucial role in AD applications. The predominant method for data generation in the CARLA Simulator involves identifying and delineating objects of interest, such as vehicles, using bounding boxes. The operation in CARLA entails capturing the coordinates of all objects on the map, which are subsequently aligned with the sensor’s coordinate system at the ego vehicle and then enclosed within bounding boxes relative to the ego vehicle’s perspective. However, this primary approach encounters challenges associated with object detection and bounding box annotation, such as ghost boxes. Although these procedures are generally effective at detecting vehicles and other objects within their direct line of sight, they may also produce false positives by identifying objects that are obscured by obstructions. We have enhanced the primary approach with the objective of filtering out unwanted boxes. Performance analysis indicates that the improved approach has achieved high accuracy.
Download
Paper Nr: 36
Title:

Lifelong Dynamic Timed A* (LTA*) for Fastest Path Retrieval in Congested Road Networks Using Predicted Speeds
Authors:

Kartikey Sondhi, Poulami Dalapati and Saurabh Kumar
Abstract: Efficient transportation systems are crucial for the ever-growing smart cities. With the increasing urbanization and growth in vehicular traffic, congestion has become a significant challenge. This research paper addresses the critical issue of identifying the fastest, least congested path in road transport networks, aiming to enhance overall travel efficiency and reduce the negative impact of traffic congestion. The study employs an improved version of the Lifelong Planning A* (LPA*) that helps find the fastest route between two points in dynamic changing environments. The proposed methodology is called the Lifelong Dynamic Timed A ∗ (LTA∗ ) algorithm with an optimal bound weight factor integrated with it to make the search more guided and efficiently predict optimized traffic paths to provide real-time recommendations. To validate the effectiveness of the de-veloped algorithm, extensive simulations and case studies are conducted on a small area in Washington as well as on Grid Worlds. The experimental results show that LTA*, within accurate weight bounds, always managed to find the fastest path, and in some cases, the time taken was close to half of that produced by A.
Download
Paper Nr: 48
Title:

Real-Time Traffic Prediction Through Stochastic Gradient Descent
Authors:

Yasmine Amor, Lilia Rejeb, Nabil Sahli, Wassim Trojet, Lamjed Ben Said and Ghaleb Hoblos
Abstract: The escalating challenges of urban traffic congestion pose a critical issue that calls for efficient traffic management system solutions. Traffic forecasting stands out as a paramount area of exploration in the field of Intelligent Transportation Systems. Various traditional machine learning techniques have been employed for predicting traffic congestion, often requiring a significant amount of data to train the model. For that reason, historical data are usually used. In this paper, our first concern is to use real-time traffic data. We adopted Stochastic Gradient Descent, an online learning method characterized by its ability to continually adapt to incoming data, facilitating real-time updates and rapid predictions. We studied a network of streets in the city of Muscat, Oman. Our model showed its accuracy through comparisons with actual traffic data.
Download
Paper Nr: 62
Title:

Study of Track Segmentation for Lap Time Optimization
Authors:

Jaroslav Klapálek, Ondřej Benedikt, Michal Sojka and Zdeněk Hanzálek
Abstract: Lap time minimization is of interest in every automotive racing competition. However, finding an optimal racing line is not a trivial task. In this work, we study one particular part of the racing line optimization problem, namely the track segmentation problem. We analyze how different track segmentation methods influence the racing line quality. Further, we present Automated Segmentation based on Curvature (ASC) method, which creates segments adaptively according to the track layout. Using lap time estimation based on a vehicle model, we compare ASC with two other methods from the literature. The preliminary results show that optimization based on ASC is able to outperform the other tested approaches by up to 15 % for the given number of iterations while converging to a good solution 3.88 times faster than the second-best method.
Download

Area 4 - Data Analytics

Full Papers
Paper Nr: 38
Title:

How Should I Measure Vehicle Deformation Depth?
Authors:

Pavlína Moravcová, Robert Zůvala and Kateřina Bucsuházy
Abstract: Determination of deformation energy is an integral part of the accident analysis. Deformation energy could be expressed by parameter EES, which could directly enter the calculation or serve as a control parameter. To determine the EES parameter, it is necessary to know the depth of plastic deformation. There is a lack of standardization in the process of deformation profile determination, because several mathematical models focus on the deformation profile according to established procedures, or the deformation depth is measured along the entire width of the deformation using evenly spaced points. Equal spacing of measurement points can be an unnecessary restriction when documenting traffic accident on accident scene. In the presented article, the differences between equal and non-equal spacing of measurement points and the subsequent influence on the EES calculation are analyzed. Statistical analysis confirmed that equal non-equal distribution of measurement points does not cause significant differences in the determined EES value, so equal spacing is not required. The non-equal spacing could better approximate the deformation profile including subsequent calculation of the EES value, when following certain rules.
Download
Short Papers
Paper Nr: 49
Title:

Bus Routing Optimisation: A Case Study for the Toulouse Metropolitan Area
Authors:

Joan Burgalat, Gaël Pallares, Myriam Foucras and Yohan Dupuis
Abstract: This paper investigates the efficiency evaluation of a public transport using an analysis of Origin-Destination matrices (mOD). The use of a trip-chaining method on the automatically collected transport data provides a realistic and accurate representation of traffic flows characterized by mOD. The introduction of a critical walking distance and an user flow at bus stop allow us to probe possible network configurations and identify the best one in terms of service offer, ecological impact and operational cost. The configurations comparison allows to identify the levers for the transport management. We deploy this methodology on a french case study for the Toulouse Metropole Occitanie region. The main obtained results shows that for a walking distance close to 1000m, the distance per day on a bus line can be optimised by 3km for a time saving close to 20%, representing an annual gain of more than 1ton of CO 2 for a user loss of around 3%. These results suggest that low-cost optimisation of a transport network is possible while maintaining a high-quality, environmentally-friendly service offering.
Download
Paper Nr: 67
Title:

Overview of Publicly-Available Data Sources on Road Traffic Accidents in Russia
Authors:

Alexey Girin, Nikolay Teslya and Nikolay Shilov
Abstract: It is very important to develop measures to prevent traffic accidents to increase road safety. The paper provides identification and description of various publicly-available sources of data related to road traffic accidents in Russia. All the described data could be combined to support an detailed analysis of accidents. A review and classification of risk factors associated with road accidents was conducted, and the data required for accident analysis as well as publicly accessible sources of such data in Russia were described. Additionally, a review of the methods used for analyzing and predicting accidents was undertaken.
Download
Paper Nr: 71
Title:

Towards Scenario Retrieval of Real Driving Data with Large Vision-Language Models
Authors:

Tin S. Sohn, Maximilian Dillitzer, Lukas Ewecker, Tim Brühl, Robin Schwager, Lena Dalke, Philip Elspas, Frank Oechsle and Eric Sax
Abstract: With the adoption of autonomous driving systems and scenario-based testing, there is a growing need for efficient methods to understand and retrieve driving scenarios from vast amounts of real-world driving data. As manual scenario selection is labor-intensive and limited in scalability, this study explores the use of three Large Vision-Language Models, CLIP, BLIP-2, and BakLLaVA, for scenario retrieval. The ability of the models to retrieve relevant scenarios based on natural language queries is evaluated using a diverse benchmark dataset of real-world driving scenarios and a precision metric. Factors such as scene complexity, weather conditions, and different traffic situations are incorporated into the method through the 6-Layer Model to measure the effectiveness of the models across different driving contexts. This study contributes to the understanding of the capabilities and limitations of Large Vision-Language Models in the context of driving scenario retrieval and provides implications for future research directions.
Download

Area 5 - Smart Mobility and Sustainable Transport Services

Full Papers
Paper Nr: 18
Title:

Large-Scale Forecasting of Electric Vehicle Charging Demand Using Global Time Series Modeling
Authors:

Tijmen van Etten, Victoria Degeler and Ding Luo
Abstract: Electric Vehicle (EV) charging demand forecasting holds paramount significance in advancing sustainable transportation systems, particularly as electric vehicle adoption surges globally. Accurate predictions of charging demand are instrumental for optimizing charging infrastructure, energy management, and grid stability. By forecasting the demand for charging, stakeholders can effectively distribute resources, plan ahead for peak usage times, and lay out blueprints for the growth of infrastructure. Furthermore, precise forecasting enables the seamless integration of renewable energy sources into transportation, promoting a cleaner and greener future. In this work, challenges in EV charging demand forecasting are addressed, and an innovative framework tailored for large-scale prediction is proposed. The methodology involves generating individual forecasts for multiple charging stations, enabling a comprehensive evaluation of forecasting models across diverse contexts. The potential of global deep learning models to enhance prediction accuracy by capturing shared patterns across time series is explored. These models exhibit remarkable generalization capabilities, proving effective even in forecasting demand at previously unobserved charging stations. The contributions of this research encompass both methodologies and insights, enriching the realm of accurate EV charging demand forecasting. This work bears significance in fostering the integration of electric vehicles into transportation systems, aligning with the trajectory towards sustainable energy solutions.
Download
Paper Nr: 35
Title:

Multi-Layer Energy Management System for Cost Optimization of Battery Electric Vehicle Fleets
Authors:

Róbinson Medina, Nikos Avramis, Subhajeet Rath, Mohammed M. Hasan, Dai-Duong Tran, Zisis Maleas, Omar Hegazy and Steven Wilkins
Abstract: One of the biggest barriers for a wider adoption of Battery-Electric Vehicles (BEVs) is their relatively higher cost compared to their combustion-based alternatives. A potential solution is to develop Energy Management Systems (EMSs), which make a more efficient use of the vehicle energy, resulting in a cheaper operation. EMSs are commonly composed of algorithms operating at fleet and vehicle layers. For example, at fleet layer one can find eco-routing for optimising the vehicle route, and eco-charging for smart charging. Likewise, at vehicle layer one can find algorithms such as eco-driving for minimizing speed-related losses and eco-comfort for minimizing the thermal-components energy consumption. These eco-functions affect the operational cost of the fleet due to reduction of metrics such as energy consumption and travelling time (which impacts labor costs). This paper presents the development of a multi-layer EMS, which integrates the aforementioned fleet and vehicle-level eco-functions. The paper focuses on the energy and operational cost savings that such a multi-layer EMS can bring to a fleet owner. Simulation results show that the EMS saves on costs produced by travelling time and energy consumption. However, the ideal ratio between these savings ultimately depends on the region, as electricity price and labor costs vary greatly.
Download