VEHITS 2020 Abstracts

Area 1 - Connected Vehicles

Full Papers

Paper Nr:	26
Title:	Cooperative Maneuvers of Highly Automated Vehicles at Urban Intersections: A Game-theoretic Approach
Authors:	Björn Koopmann, Stefan Puch, Günter Ehmen and Martin Fränzle
Abstract:	In this paper, we propose an approach how connected and highly automated vehicles can perform cooperative maneuvers such as lane changes and left-turns at urban intersections where they have to deal with human-operated vehicles and vulnerable road users such as cyclists and pedestrians in so-called mixed traffic. In order to support cooperative maneuvers the urban intersection is equipped with an intelligent controller which has access to different sensors along the intersection to detect and predict the behavior of the traffic participants involved. Since the intersection controller cannot directly control all road users and – not least due to the legal situation – driving decisions must always be made by the vehicle controller itself, we focus on a decentralized control paradigm. In this context, connected and highly automated vehicles use some carefully selected game theory concepts to make the best possible and clear decisions about cooperative maneuvers. The aim is to improve traffic efficiency while maintaining road safety at the same time. Our first results obtained with a prototypical implementation of the approach in a traffic simulation are promising.
Download

Short Papers

Paper Nr:	12
Title:	A Vehicle Telematics Service for Driving Style Detection: Implementation and Privacy Challenges
Authors:	Christian Kaiser, Alexander Stocker, Andreas Festl, Marija D. Petrovic, Efi Papatheocharous, Anders Wallberg, Gonzalo Ezquerro, Jordi O. Orbe, Tom Szilagyi and Michael Fellmann
Abstract:	Connected mobility is not only a future market, but also holds great innovation potential. The analysis of vehicle telematics data in the cloud enables novel data-driven services for several stakeholders, e.g. a mobile application for the driver to obtain his driving style. This inevitably leads to privacy concerns and the question why and when are users willing to share driving telematic data, which we addressed in an empirical study. The paper presents an implementation of a data-driven service based on vehicle telematics data and discusses how privacy issues can be tackled. For the data-driven service, the most interesting steps along the vehicle data value chain are described in detail, firstly (i) vehicle telematics data collection, secondly, (ii) the wireless data transfer to a cloud platform, and thirdly, (iii) pre-processing and data analysis to evaluate the drivers’ driving style and analyse the driving risk. Finally, (iv) a smartphone application for drivers presents driving style and driving risk data on the smartphone in an interactive way, so that the driver can work on improving both, which has a positive effect on driving and road safety.
Download

Paper Nr:	30
Title:	Evaluating the Dedicated Short-range Communication for Connected Vehicles against Network Security Attacks
Authors:	Tu Le, Ingy Elsayed-Aly, Weizhao Jin, Seunghan Ryu, Guy Verrier, Tamjid Al Rahat, B. B. Park and Yuan Tian
Abstract:	According to the National Highway Traffic Safety Administration, there are more than 5 million road crashes every year in the U.S. More than 90 people die in car crashes every day. Even though the number of people surviving crashes has increased significantly thanks to safety features, such as airbags and anti-lock brakes, many people experience permanent injuries. The U.S. Department of Transportation introduced connected vehicle technologies, which enables vehicles to “talk” to each other and exchange important data on the roads, with the goal of preventing crashes from happening in the first place. With the rapid development of autonomous driving technology, vehicles in the near future will be able to operate completely without human drivers, increasing the need of reliable connected vehicle technologies. Due to the safety-critical characteristics of autonomous vehicles, it is important to evaluate the technologies extensively prior to deployment to ensure the safety of drivers, passengers, and pedestrians. In this paper, we evaluate the safety of Dedicated Short-Range Communication (DSRC), which is a popular low-latency wireless communication technology specifically designed for connected vehicles. We present three real-world network security attacks and conduct experiments on real DSRC-supported modules. Our results show that DSRC is vulnerable to these dangerous attacks and such attacks can be easily implemented by adversaries without significant resources. Based on our evaluation, we also discuss potential countermeasures to better improve the security and safety of DSRC and connected vehicles.
Download

Paper Nr:	75
Title:	Constructing Tool-based Security Test Sequences for Vehicles as High-tech Data-rich Systems
Authors:	Alexandr Vasenev, Stelios Karagiannis and Roland Mathijssen
Abstract:	Vehicles, as a prime example of high-tech systems, get increasingly connected and data-centric with the need to process personally identifiable information. Often, companies that develop such systems act as integrators and need to comply to adequate data protection requirements. For instance, GDPR requires securing personal data. Yet, testing security of data (including, but not limited to personal data) is challenging. Penetration testing often starts from the outside of the system and take place at the end of the development lifecycle. This may be insufficient to adequately test for potential errors hidden within system boundaries. Having methods to design, execute, and reuse (automated) security test cases on a ‘white-box’ system is desirable. This positioning paper proposes an approach to design tool-based security test sequences. We structurally approach high-level data storing, processing, and communicating functionality in connection to the system boundary. We suggest to use pen-testing tools and sequences for testing the functionality of the vehicle’s (sub)system, before test-enabling interfaces are removed. This paper intends to contribute to discussions how to test layered defense implementations. The proposed approach is undergoing extensions and validations.
Download

Paper Nr:	64
Title:	Towards Data-driven Services in Vehicles
Authors:	Milan Koch, Hao Wang, Robert Bürgel and Thomas Bäck
Abstract:	Numerous recent studies show the prosperous future of data-driven business models. Some key challenges have to be dealt with when moving towards the development of data-driven car services. In this paper, a new data-driven customer service is proposed for the settlement of vehicle low speed accidents. Beyond that, we present a more general approach towards the development of data-driven car services. We point out its main challenges and suggest a method for developing new customer-oriented data-driven services. This approach illustrates key points in developing a practical service, from a technical and business related perspective. Such data-driven services are developed mostly on a small number of initial test data, which results often in a limited prediction performance. Therefore, based on an optimized CRISP-DM approach, we propose a methodology for developing initial prediction models with limited test data and stabilizing the models with newly gained data after deployment by online learning. On-board and off-board services are discussed with the result that especially off-board running services offer a large potential for future data-driven business models in a digital ecosystem. The flexibility of such an ecosystem depends on the degree of the integration of the vehicle in the ecosystem - in other words, the car needs to be enabled to deliver data on demand according to GDPR and to any applicable regional law and in cooperation with the customer. The presented method, together with the ecosystem, enables fast developments of various data-driven services.
Download

Area 2 - Intelligent Transport Systems and Infrastructure

Full Papers

Paper Nr:	11
Title:	How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments
Authors:	Sören Schleibaum, Maike Greve, Tim-Benjamin Lembcke, Amos Azaria, Jelena Fiosina, Noam Hazon, Lutz Kolbe, Sarit Kraus, Jörg P. Müller and Mark Vollrath
Abstract:	Ridesharing can significantly reduce individual passenger transport and thus greenhouse gas emissions generated by traffic. Although ridesharing offers great potential, it is not yet popular enough to be seen as an important contribution to solving the aforementioned problems. Our hypothesis suggests that we need to make the assignment mechanism of ridesharing systems more human-centric and comprehensible in order to popularise ridesharing. Therefore, we investigate factors that influence the choice of users and their satisfaction with the assigned ride. Most of today’s ridesharing assignment algorithms focus solely on features such as time, distance and price. Contrarily, this paper examines additional factors that influence customer decisions to increase their satisfaction. Therefore, we first conduct a literature study to identify previous preferences relevant for ridesharing from a research perspective. Subsequently, we extract the relevant preferences for an assignment process. From these we secondly conduct a survey. Last, we analyse the obtained survey data and order the preferences based on their importance for participants overall and among demographic subgroups.
Download

Paper Nr:	20
Title:	Machine Learning based Video Processing for Real-time Near-Miss Detection
Authors:	Xiaohui Huang, Tania Banerjee, Ke Chen, Naga S. Varanasi, Anand Rangarajan and Sanjay Ranka
Abstract:	Video-based sensors are ubiquitous and are therefore indispensable in understanding traffic behavior at intersections. Deriving near-misses from large scale video processing is extremely useful in assessing the level of safety of intersections. In this paper, we develop real-time or near real-time algorithms for detecting near-misses for intersection video collected using fisheye cameras. We propose a novel method consisting of the following steps: 1) extracting objects and multiple object tracking features using convolutional neural networks; 2) densely mapping object coordinates to an overhead map; 3) learning to detect near-misses by new distance measures and temporal motion. The experimental results demonstrate the effectiveness of our approach with real-time performance at 40 fps and high specificity.
Download

Paper Nr:	63
Title:	A Validation Study of the Fadhloun-Rakha Car-following Model
Authors:	Karim Fadhloun, Hesham Rakha, Amara Loulizi and Jinghui Wang
Abstract:	The research presented in this paper investigates and validates the performance of a new car-following model (the Fadhloun-Rakha (FR) model). The FR model incorporates the key components of the Rakha-Pasumarthy-Adjerid (RPA) model in that it uses the same steady-state formulation, respects vehicle dynamics, and uses very similar collision-avoidance strategies to ensure safe following distances between vehicles. The main contributions of the FR model over the RPA model are the following: (1) it explicitly models the driver throttle and brake pedal input; (2) it captures driver variability; (3) it allows for shorter than steady-state following distances when following faster leading vehicles; (4) it offers a much smoother acceleration profiles; and (5) it explicitly captures driver perception and control inaccuracies and errors. In this paper, a naturalistic driving dataset is used to validate the FR model. Furthermore, the model performance is compared to that of five widely used car-following models, namely: the Wiedemann model, the Frietzsche model, the Gipps model, the RPA model and the Intelligent Driver Model (IDM). A comparative analysis between the different model outputs is used to determine the performance of each model in terms of its ability to replicate the empirically observed driver/vehicle behavior. Through quantitative and qualitative evaluations, the proposed FR model is demonstrated to significantly decrease the modeling error when compared to the five aforementioned models and to generate trajectories that are highly consistent with empirically observed driver following behavior.
Download

Short Papers

Paper Nr:	27
Title:	The Comparison of 3D and 2D Measurement Techniques Used for the Analysis of Vehicle Deformation
Authors:	Pavlína Moravcová, Kateřina Bucsuházy, Robert Zůvala, Martin Bilík and Albert Bradáč
Abstract:	As one of the main assumptions for the accident analysis has been detailed information about vehicle deformation. The precise deformation depth allows to quantify deformation energy and related impact speed. The aim of this paper has been the comparison of two selected methods used for the determination of deformation depth. For the purpose of this paper were selected top-view photography as basic and cheap method and 3D scanning as modern and advanced method. Different vehicles and 2 basic types of damage - frontal and side impact - were chosen for the analysis. Also, the different range of vehicle deformation depth were selected. On the basis of obtained results is possible to determine the applicability of these methods, their advantages and limitations.
Download

Paper Nr:	29
Title:	A Data Driven Approach to Derive Traffic Intersection Geography using High Resolution Controller Logs
Authors:	Dhruv Mahajan, Tania Banerjee, Yashaswi Karnati, An. Rangarajan and Sanjay Ranka
Abstract:	Current traffic signal controllers are capable of recording events (signal events; vehicle arrival and departure) at very high resolutions (usually, 10Hz). The high resolution data rates enable the computation and study of various (granular) measures of effectiveness. However, without knowing the location of specific detectors on an intersection and the phases they are mapped to, a number of measures of effectiveness (of signal performance) cannot be evaluated. These mappings may not be available or up to date for many practical reasons (e.g., old infrastructure, mappings not machine readable, maintenance or addition of new lanes, etc.). In this paper, we develop an inference engine to map detectors to phases and distinguish between the stop bar and advance detectors, or in other words, infer the location of the loop detectors with reference to the intersection.
Download

Paper Nr:	34
Title:	A Lightweight Virtualisation Platform for Cooperative, Connected and Automated Mobility
Authors:	Fabian Gand, Ilenia Fronza, Nabil El Ioini, Hamid R. Barzegar, Van T. Le and Claus Pahl
Abstract:	Digital mobility systems such as autonomous cars or traffic management build on connectivity and automated cooperation. In order to facilitate various use cases such as vehicle manoeuvres, infotainment support or state share functions, a distributed layered computation and communication infrastructure is needed that connects vehicles and other devices through mobile networks, linking them to edge and cloud services. Of particular relevance are lightweight clustered infrastructures close to the edge of the network that provide nonetheless sufficient compute, storage and networking capabilities. Clusters consisting of single-board devices are used in a variety of these use cases. In most cases, data that is accumulated on the devices has to be sent to remote cloud hubs for processing. However, with the hardware capabilities of these controllers continuously increasing, it is now possible to directly process data on these edge cluster. This concept is known as Edge Computing. We propose an edge computing architecture for cooperative, connected and automated mobility that relies on industry-standard technologies such as the MQTT protocol for communication, Prometheus for monitoring and Docker swarm in conjunction with openFaas for deploying containerized services.
Download

Paper Nr:	39
Title:	Video-based Machine Learning System for Commodity Classification
Authors:	Pan He, Aotian Wu, Xiaohui Huang, Anand Rangarajan and Sanjay Ranka
Abstract:	The cost of video cameras is decreasing rapidly while their resolution is improving. This makes them useful for a number of transportation applications. In this paper, we present an approach to commodity classification from surveillance videos by utilizing text information of logos on trucks. A new real-world benchmark dataset is collected and annotated accordingly that covers over 4,000 truck images. Our approach is evaluated on video data collected in collaboration with the state transportation entity. Results on this dataset indicate that our proposed approach achieved promising performance. This, along with prior work on trailer classification, can be effectively used for automatically deriving the commodity classification for trucks moving on highways using video collection and processing.
Download

Paper Nr:	40
Title:	Towards a Better Management of Emergency Evacuation using Pareto Min Cost Max Flow Approach
Authors:	Sreeja Kamishetty and Praveen Paruchuri
Abstract:	Events during an emergency unfold in an unpredictable fashion which makes management of traffic during emergencies pretty challenging. Furthermore, some vehicles would need to be evacuated faster than others e.g., emergency vehicles or large vehicles carrying a lot more people. The Prioritized Routing Assistant for Flow of Traffic (PRAFT) enables prioritized routing during emergencies. However, the PRAFT solution does not compute multiple plans that can help handle better dynamic nature of emergencies. PRAFT maps the prioritized routing problem to the Minimum-Cost Maximum-Flow (MCMF) problem, hence its solution can accommodate maximum flow while routing vehicles based on priority (maps higher priority vehicles to better quality routes (i.e., ones with minimum cost)). We build upon the PRAFT solution to make the following contributions: (a) Develop a Pareto Minimum-Cost Maximum-Flow (Pareto-MCMF) algorithm which can compute all the possible MCMF solutions. (b) Through a series of experiments performed using the well known traffic simulator SUMO, we could show that all the solutions generated by Pareto-MCMF indeed have properties similar to a MCMF solution thus providing multiple high quality options for traffic police to pick from depending on the situation.
Download

Paper Nr:	51
Title:	A Key Performance Optimization Agent-based Approach for Public Transport Regulation
Authors:	Nabil Morri, Sameh Hadouaj and Lamjed Ben Said
Abstract:	Today’s, an efficient and reliable public transport system becomes essential to assist cities in their wealth creation. However, public transportation systems are highly complex because of the modes involved, the multitude of origins and destinations, and the amount and variety of traffic. They have to cope with dynamic environments where many complex and random phenomena appear and disturb the traffic network. To ensure a good quality service, perturbations caused by these phenomena must be detected and treated within an acceptable time frame via the use of a control system. The control process should rely on many criteria related to the traffic management of public transport: Key Performance Indicators. In this paper, we introduce a Regulation Support System of Public Transport (RSSPT) that detects and regulates the traffic perturbation of multimodal public transportation. The system uses optimization techniques to solve the control problem. We based our regulation support system on a multi-agent approach to cope with the distributed nature of the public transportation system. To validate our model, we conducted tests by simulating perturbation scenarios in a real traffic network. A comparison between real data and the obtained results shows an improvement in the quality service.
Download

Paper Nr:	52
Title:	Automatic Train Operation: History and Open Questions
Authors:	Aleš Lieskovský, Ivo Myslivec and Michal Žemlička
Abstract:	The paper presents the concept of automatic train operation. We give here short description of its functionality and remember some points from its history. There is an overview of various future development as well as proposals for improvement of some existing, especially mainline, solutions. There are presented also some observations from decades of practical use of automatic train operation in the Czech Republic. Selected challenges are presented and discussed.
Download

Paper Nr:	69
Title:	Taxi Demand Prediction based on LSTM with Residuals and Multi-head Attention
Authors:	Chih-Jung Hsu and Hung-Hsuan Chen
Abstract:	This paper presents a simple yet effective framework to accurately predict the taxi demands of different regions in a city in the near future. This framework is based on a deep-learning structure with residual connections in the LSTM layers and the attention mechanism. We found that adding residuals accelerates optimization and that adding the attention mechanism makes the model better predict the taxi demands, especially when the demand fluctuates greatly in the peak hours and off-peak hours. We conducted extensive experiments by comparing the proposed models to the time-series model (ARIMA), traditional supervised learning model (ridge regression), strong machine learning model that won many Kaggle competitions (Gradient Boosted Decision Tree implemented in the XGBoost library), and deep learning models (LSTM and DMVST-Net) on two real and open-source datasets. Experimental results show that the proposed models outperform the baselines for most cases. We believe the greatest improvement comes from the attention mechanism, which helps distinguish the demands in the peak hours and off-peak hours. Additionally, the proposed model runs 10% to 40%-times faster than the other deep-learning-based models. We applied the models to participate in a taxi demand prediction challenge and won second place out of hundreds of teams.
Download

Paper Nr:	36
Title:	Towards a Comprehensive Model for the Impact of Traffic Patterns on Air Pollution
Authors:	Caterina Balzotti, Maya Briani, Barbara de Filippo and Benedetto Piccoli
Abstract:	The impact of vehicular traffic on society is huge and multifaceted, including economic, social, health and environmental aspects. The problems is complex and hard to model since it requires to consider traffic patterns, air pollutant emissions, and the chemical reactions and dynamics of pollutants in the low atmosphere. This paper aims at exploring a comprehensive simulation tool ranging from vehicular traffic all the way to environmental impact. As first step in this direction, we couple a traffic second-order model, tuned on NGSIM data, with an nitrogen oxides (NOx) emission model and a set of equations for some of the main chemical reactions behind ozone (O3) production.
Download

Paper Nr:	48
Title:	Regional Module of Intelligent Transportation System: Algorithms and Information Infrastructure
Authors:	Anton Sysoev and Elena Khabibullina
Abstract:	Due to the increasing number of personal transportation vehicles and cargo transportation it is reasonable to implement intelligent transportation systems based on adaptive algorithms to deliver the effective control of traffic flows within the highspeed transportation corridors connecting different countries. The presented paper proposed the concept of the regional intelligent transportation system module which could be extented into regions taking into account its specific features. Presented approaches are considered on the data on real-time traffic flow parameters collected from different heterogeneous data sources. The nature of the data and its structure underlie the data warehouse model.
Download

Area 3 - Intelligent Vehicle Technologies

Full Papers

Paper Nr:	9
Title:	Aware and Intelligent Infrastructure for Action Intention Recognition of Cars and Bicycles
Authors:	Cristofer Englund
Abstract:	Action intention recognition is becoming increasingly important in the road vehicle automation domain. Autonomous vehicles must be aware of their surroundings if we are to build safe and efficient transport systems. This paper explores methods for predicting the action intentions of road users based on an aware and intelligent 3D camera-based sensor system. The collected data contains trajectories of two different scenarios. The first one includes bicyclists and the second cars that are driving in a road approaching an intersection where they are either turning or continuing straight. The data acquisition system is used to collect trajectories of the road users that are used as input for models trained to predict the action intention of the road users.
Download

Paper Nr:	14
Title:	3D Object Detection from LiDAR Data using Distance Dependent Feature Extraction
Authors:	Guus Engels, Nerea Aranjuelo, Ignacio Arganda-Carreras, Marcos Nieto and Oihana Otaegui
Abstract:	This paper presents a new approach to 3D object detection that leverages the properties of the data obtained by a LiDAR sensor. State-of-the-art detectors use neural network architectures based on assumptions valid for camera images. However, point clouds obtained from LiDAR data are fundamentally different. Most detectors use shared filter kernels to extract features which do not take into account the range dependent nature of the point cloud features. To show this, different detectors are trained on two splits of the KITTI dataset: close range (points up to 25 meters from LiDAR) and long-range. Top view images are generated from point clouds as input for the networks. Combined results outperform the baseline network trained on the full dataset with a single backbone. Additional research compares the effect of using different input features when converting the point cloud to image. The results indicate that the network focuses on the shape and structure of the objects, rather than exact values of the input. This work proposes an improvement for 3D object detectors by taking into account the properties of LiDAR point clouds over distance. Results show that training separate networks for close-range and long-range objects boosts performance for all KITTI benchmark difficulties.
Download

Paper Nr:	17
Title:	Classification of Driver Intentions at Roundabouts
Authors:	Moritz Sackmann, Henrik Bey, Ulrich Hofmann and Jörn Thielecke
Abstract:	Classification of other drivers’ intentions is an important requirement for automated driving. We present two methods to estimate whether a driver leaves a roundabout. The first, like many other approaches to this problem, requires training data of the specific roundabout to extract typical behavior patterns. Afterwards, these patterns are used for classification of other drivers’ intentions. The second approach generates typical behavior patterns from a precise map. Consequently, no training data is required and classification can be performed on arbitrary roundabouts as long as a map is available. Experimental evaluation on a real world dataset of 266 trajectories shows that the performance of the map-based approach is comparable to the data-driven approach. The classification result can be used in a later stage for behavior planning of automated vehicles or driver assistance systems.
Download

Paper Nr:	18
Title:	Tutorial on Sampling-based POMDP-planning for Automated Driving
Authors:	Henrik Bey, Maximilian Tratz, Moritz Sackmann, Alexander Lange and Jörn Thielecke
Abstract:	Behavior planning of automated vehicles entails many uncertainties. Partially Observable Markov Decision Processes (POMDP) are a mathematical framework suited for formulating the arising sequential decision problems. Solving POMDPs used to be intractable except for overly simplified examples, especially when execution time is of importance. Recent sampling-based solvers alleviated this problem by searching not for the exact but rather an approximated solution, and made POMDPs usable for many real-world applications. One of these algorithms is the Adaptive Belief Tree (ABT) algorithm which will be analyzed in this work. The scenario under consideration is an uncertain obstacle in the way of an automated vehicle. Following this example, the setup of POMDP and ABT is derived and the impact of important parameters is assessed in simulation. As such, this work provides a hands-on tutorial, giving insights and hints on how to overcome the pitfalls in using sampling-based POMDP solvers.
Download

Paper Nr:	19
Title:	Detection and Recognition of Arrow Traffic Signals using a Two-stage Neural Network Structure
Authors:	Tien-Wen Yeh and Huei-Yung Lin
Abstract:	This paper develops a traffic light detection and recognition system based on convolutional neural networks for Taiwan road scenes. A two-stage approach is proposed with first detecting the traffic light position, followed by the light state recognition. It is specifically designed to identify the challenging arrow signal lights in many urban traffic scenes. In the detection stage, the map information and two cameras with different focal lengths are used to detect the traffic lights at different distances. In the recognition stage, a new method combining object detection and classification is proposed to deal with various light state classes in Taiwan road scenes. Furthermore, an end-to-end network with shared feature maps is implemented to reduce the computation time. Experiments are carried out on the public LISA dataset and our own dataset collected from two routes with urban traffic scenes.
Download

Paper Nr:	28
Title:	Weather Effects on Obstacle Detection for Autonomous Car
Authors:	Rui Song, Jon Wetherall, Simon Maskell and Jason F. Ralph
Abstract:	Adverse weather conditions have become a critical issue when developing autonomous vehicles and driver assistance systems. Training and testing autonomous vehicles in a simulation environment before deploying them into the market have many benefits due to lower costs and fewer risks. However, there are only a few works about weather influences on sensors in the simulated environment. A more systematic study of weather effects on the sensors used on autonomous cars is required. This paper presents a multi-sensor simulation environment under different weather conditions and examines the influence on environmental perception and obstacle detection for autonomous cars. The simulation system is being developed as part of a collaborative project entitled: Artificial Learning Environment for Autonomous Driving (ALEAD). The system incorporates a suite of sensors typically used for autonomous cars. Each sensor model has been developed to be as realistic as possible – incorporating physical defects and other artefacts found in real sensors. The influence of weather on these sensors has been simulated based on experimental data. The multi-sensor system has been tested under different simulated weather conditions and analysed to determine the effect on detection of a dynamic obstacle and a road lane in a 3D environment.
Download

Paper Nr:	32
Title:	User-adaptive Eyelid Aperture Estimation for Blink Detection in Driver Monitoring Systems
Authors:	Juan D. Ortega, Marcos Nieto, Luis Salgado and Oihana Otaegui
Abstract:	This paper presents a new method for eyelid aperture estimation, suitable to be used in Driver Monitoring Systems (DMS) to measure blink patterns such as microsleeps and any other metric that assess the fatigue level of the driver. The method has been designed to work real-time and in continuous operation, by introducing a novel online Exponential Weighted Moving Average (EWMA)-based Bayesian estimation process, which ensures dynamic adaptability to drivers with different physiognomy features, and also to changes due to physiological states (e.g. drowsiness). Our method has been implemented in the framework of a DMS, to take advantage of existing facial landmark detection and tracking mechanisms, and to provide real-time functionality for driving platforms (such as the NVIDIA Drive PX 2). The method is evaluated against a large labelled dataset, and compared to baseline and previous existing methods, showing an excellent balance between adaptability, performance, and robustness.
Download

Paper Nr:	46
Title:	Pedestrian Head and Body Pose Estimation with CNN in the Context of Automated Driving
Authors:	Michaela Steinhoff and Daniel Göhring
Abstract:	The challenge of determining pedestrians head poses in camera images is a topic that has already been researched extensively. With the ever-increasing level of automation in the field of Advanced Driver Assistance Systems, a robust head orientation detection is becoming more and more important for pedestrian safety. The fact that this topic is still relevant, however, indicates the complexity of this task. Recently, trained classifiers for discretized head poses have recorded the best results. But large databases, which are essential for an appropriate training of neural networks meeting the special requirements of automatic driving, can hardly be found. Therefore, this paper presents a framework with which reference measurements of head and upper body poses for the generation of training data can be carried out. This data is used to train a convolutional neural network for classifying head and upper body poses. The result is extended in a semi-supervised manner which optimizes and generalizes the detector, so that it is applicable to the prediction of pedestrian intention.
Download

Paper Nr:	53
Title:	Auto-Sapiens Autonomous Driving Vehicle
Authors:	Maicol Laurenza, Gianluca Pepe and Antonio Carcaterra
Abstract:	This paper presents the Auto-Sapiens project, an autonomous driving car developed by the Mechatronics and Vehicle Dynamics Lab, at Sapienza University of Rome. Auto-Sapiens is a technological platform to test and improve innovative control algorithms. The car platform is a standard car (Smart ForTwo) equipped with throttle, brake, steering actuators and different sensors for attitude identification and environment reconstruction. The first experiments of the Auto-Sapiens car test a new obstacle avoidance. The vehicle, controlled by an optimal variational feedback control, recently developed by the authors, includes the nonlinearities inherent in the car dynamics for better performances. Results show the effectiveness of the system in terms of safety and robustness of the avoidance maneuvers.
Download

Paper Nr:	55
Title:	An Optimization-based Strategy for Shared Autonomous Vehicle Fleet Repositioning
Authors:	Felipe de Souza, Krishna M. Gurumurthy, Joshua Auld and Kara M. Kockelman
Abstract:	With the emergence of autonomous technology, shared autonomous vehicles (SAVs) will potentially be the prevalent transportation mode for urban mobility. On one hand, relying on SAV fleets can provide several operational benefits. On the other hand, SAVs can increase travel distance and add congestion due to unoccupied trips such as pickup and repositioning trips. One important aspect for a SAV fleet’s success is to serve the incoming requests at reasonably low waiting time. This is achieved by an adequate fleet size that is spatially distributed thoughtfully so that incoming requests can be served by a nearby vehicle. Unfortunately, it is challenging to keep a satisfactory spatial distribution of vehicles due to imbalances in the origin and destination patterns of incoming requests. This paper focuses on the impact of SAV relocation on traveler wait times using a novel optimization-based algorithm for repositioning. POLARIS, an agent-based tool, is used for a case study of Bloomington, Illinois to quantify the benefits of allowing SAV repositioning. On average, the wait times were around 20% lower with repositioning for all adequate fleet sizes. SAVs were available more uniformly across the region’s zones, and proportional to trip-making at different times of day. In addition, enabling repositioning led to a higher share of demands being served. These benefits, however, are achieved at the expense of 6% added vehicles miles traveled.
Download

Paper Nr:	62
Title:	Linked Real and Virtual Test Environment for Distributed C-ITS-Applications
Authors:	Michael Klöppel-Gersdorf and Thomas Otto
Abstract:	The development and test of automated and connected driving, based on vehicle-to-infrastructure (V2I) communication, is essential for C-ITS pilot implementations. Even today, the deployment and test of these services is a great challenge. Multiple connections, interfaces as well as interactions between several entities make it difficult to find and eliminate malfunction of cooperative components. The ranges and boundaries of drive and test scenarios make debugging during test drives in a real traffic environment substantially difficult, because it requires reproducible conditions. The solution of the above mentioned problem is a linked real and virtual test environment for distributed C-ITS-Applications under test. A microscopic simulation of traffic scenarios running on a test environment computer is combined with a real signal control device including traffic lights, real roadside unit and a real on-board unit for the communication between infrastructure (traffic lights) and vehicle (on-board unit). This hardware and software-in-the-loop (HiL/SiL) approach enables the use of reproducible drive and test scenarios for testing C-ITS-Applications based on an interaction of different traffic conditions, traffic light devices and vehicles.
Download

Paper Nr:	81
Title:	Petri Net-based Smart Parking Information System
Authors:	Omar Makke and Oleg Gusikhin
Abstract:	In this paper, we propose a Petri Net digital twin solution for smart parking information system to track the occupancy of a parking space while respecting the privacy of the drivers. An edge computing device is deployed to process camera images, and a Petri Net model is generated from the event logs and tracks the occupancy of the parking structure. This type of solution can be enhanced to any desirable level of accuracy. The paper provides preliminary analytics for the parking dynamics in a period of three months. This analysis clearly demonstrates the tangible benefits of the parking information system.
Download

Paper Nr:	83
Title:	What Cooperation Costs: Quality of Communication and Cooperation Costs for Cooperative Vehicular Maneuvering in Large-scale Scenarios
Authors:	Daniel Bischoff, Florian Schiegg, Tobias Meuser, Dieter Schuller, Nils Dycke and Ralf Steinmetz
Abstract:	With the rise of vehicles on the road, Cooperative Vehicular Maneuvering (CVM) is a crucial prospect to increase the efficiency of future vehicular traffic. Recent work proposes promising approaches for CVM using Vehicle-to-Everything (V2X) communication to increase traffic efficiency but evaluates its performance with only a few vehicles involved and without considering realistic radio propagation channel models. CVM relies on high quality of communication to coordinate cooperative maneuvers and increases traffic efficiency primarily under heavy vehicular traffic load, which also challenges the V2X quality of communication in terms of channel load and reliability. In this paper, we propose a novel computational efficient CVM planning algorithm specially designed for large-scale scenarios considering a realistic radio propagation channel model and analyze the quality of communication and cooperation cost of CVM using ad hoc communication technology. For our urban intersection scenario, we show that imperfect communication limits the earliest start of cooperation to 150m and increases the average Age of Information (AoI) of CVM messages up to 400ms, which motivates the need for more advanced V2X dissemination strategies.
Download

Paper Nr:	87
Title:	Practical Depth Estimation with Image Segmentation and Serial U-Nets
Authors:	Kyle J. Cantrell, Craig D. Miller and Carlos W. Morato
Abstract:	Knowledge of environmental depth is required for successful autonomous vehicle navigation and VSLAM. Current autonomous vehicles utilize range-finding solutions such as LIDAR, RADAR, and SONAR that suffer drawbacks in both cost and accuracy. Vision-based systems offer the promise of cost-effective, accurate, and passive depth estimation to compete with existing sensor technologies. Existing research has shown that it is possible to estimate depth from 2D monocular vision cameras using convolutional neural networks. Recent advances suggest that depth estimate accuracy can be improved when networks used for supplementary tasks such as semantic segmentation are incorporated into the network architecture. A novel Serial U-Net (NU-Net) architecture is introduced as a modular, ensembling technique for combining the learned features from N-many U-Nets into a single pixel-by-pixel output. Serial U-Nets are proposed to combine the benefits of semantic segmentation and transfer learning for improved depth estimation accuracy. The performance of Serial U-Net architectures are characterized by evaluation on the NYU Depth V2 benchmark dataset and by measuring depth inference times. Autonomous vehicle navigation can substantially benefit by leveraging the latest in depth estimation and deep learning.
Download

Short Papers

Paper Nr:	8
Title:	Online Driving Behavior Scoring using Wheel Speeds
Authors:	Marian Waltereit, Peter Zdankin, Viktor Matkovic, Maximilian Uphoff and Torben Weis
Abstract:	We present an online scoring algorithm for measuring driving behavior using wheel speeds only. Such an algorithm can be used to provide drivers with feedback about their driving behavior while driving in order to reduce aggressive driving, which is a primary cause of traffic accidents. Our algorithm uses a minimal data set already available through the built-in wheel speed sensors of contemporary cars. Due to the small amount of data used and the low computational complexity, our algorithm can easily be deployed on single-board computers. With real driving experiments in a controlled and an uncontrolled environment, we demonstrate the suitability of our scoring algorithm for identifying aggressive driving and assessing the driving behavior.
Download

Paper Nr:	15
Title:	Real-time Spatial-temporal Context Approach for 3D Object Detection using LiDAR
Authors:	K. C. Kumar and Samir Al-Stouhi
Abstract:	This paper proposes a real-time spatial-temporal context approach for BEV object detection and classification using LiDAR point-clouds. Current state-of-art BEV object-detection approaches focused mainly on single-frame point-clouds while the temporal factor is rarely exploited. In current approach, we aggregate 3D LiDAR point clouds over time to produce a 4D tensor, which is then fed to a one-shot fully convolutional detector to predict oriented 3D object bounding-box information along with object class. Four different techniques are evaluated to incorporate the temporal dimension; a) joint training b) CLSTM c) non-local context network (NLCN) d) spatial-temporal context network (STCN). The experiments are conducted on large-scale Argoverse dataset and results shows that by using NLCN and STCN, mAP accuracy is increased by a large margin over single frame 3D object detector and YOLO4D 3D object detection with our approach running at a speed of 28fps.
Download

Paper Nr:	38
Title:	Dynamic Control System Design for Autonomous Car
Authors:	Shoaib Azam, Farzeen Munir and Moongu Jeon
Abstract:	The autonomous vehicle requires higher standards of safety to maneuver in a complex environment. We focus on control of the self-driving vehicle that includes the longitudinal and lateral dynamics of the vehicle. In this work, we have developed a customized controller for our KIA Soul self-driving car. The customized controller implements the PID control for throttle, brake, and steering so that the vehicle follows the desired velocity profile, which enables a comfortable and safe ride. Besides, we have also catered the lateral dynamic model with two approaches: pure pursuit and model predictive control. An extensive analysis is performed between pure pursuit and its adversary model predictive control for the efficacy of the lateral model.
Download

Paper Nr:	41
Title:	Advanced Curve Speed Warning System using Standard GPS Technology and Road-level Mapping Information
Authors:	Shahnewaz Chowdhury, Muhammad Faizan and M. I. Hayee
Abstract:	Lane departure and advance curve warning are critical among several Advanced Driver-Assistance Systems (ADAS) functions, which have significant potential to reduce crashes. Generally, lane departure and advance curve speed warning systems either use different image processing techniques or GPS technology with digital maps of lane-level resolution. However, these systems are expensive to implement as well as have some limitations such as harsh weather or irregular lane markings can negatively influence their performance. Previously we proposed a lane departure detection which uses a standard GPS receiver without any lane-level resolution maps. Now, we have added another feature in this algorithm to detect an upcoming curve in advance and warn the driver about its advisory speed at a safe distance so that driver can adjust vehicle speed accordingly before reaching the curve. We have implemented our algorithm in a prototype system and demonstrated in the field. We have performed extensive field tests and the test results show that each time vehicle approaches a curve, our algorithm issues a warning and correctly determines the advisory speed for the curve to warn the driver at a safe distance before the curve starts.
Download

Paper Nr:	43
Title:	Are Consumers Ready to Adopt Highly Automated Passenger Vehicles? Results from a Cross-national Survey in Europe
Authors:	Ilias Panagiotopoulos, George Dimitrakopoulos, Gabrielė Keraitė and Urte Steikuniene
Abstract:	Automated vehicles are currently being developed by major car manufacturers planning to be available in market diffusion the next years. This disruptive technology is expected to provide an alternative type of transportation services by positively affecting road safety, traffic congestion, more individual comfort and convenience for drivers/users. However, besides the aforementioned societal benefits, researches on the predictors influencing individuals’ attitudes and willingness to adopt automated vehicles in the future are crucial requirements for their successful diffusion in international market. In this way, the current study aims to investigate the factors that may hinder or facilitate consumers’ acceptance and adoption of Highly Automated Passenger Vehicles (HAPVs). A research model through extending the original Unified Theory of Acceptance and Use of Technology (UTAUT) was developed and accordingly an online survey was conducted among the general public in Europe; 811 valid answers were collected and analyzed. The results indicate that the constructs of perceived driving enjoyment, perceived financial cost, perceived reliability/trust, social influence and performance expectancy were all useful predictors of behavioural intentions to drive/use HAPVs. The findings derived from this study will contribute to car manufacturers towards HAPVs in order not only to develop better driving automation technology systems for them, but also to develop proper implementation strategies that will lead to widespread deployment in international market.
Download

Paper Nr:	47
Title:	Multiple Path Prediction for Traffic Scenes using LSTMs and Mixture Density Models
Authors:	Jaime B. Fernandez, Suzanne Little and Noel E. O’connor
Abstract:	This work presents an analysis of predicting multiple future paths of moving objects in traffic scenes by leveraging Long Short-Term Memory architectures (LSTMs) and Mixture Density Networks (MDNs) in a single-shot manner. Path prediction allows estimating the future positions of objects. This is useful in important applications such as security monitoring systems, Autonomous Driver Assistance Systems and assistive technologies. Normal approaches use observed positions (tracklets) of objects in video frames to predict their future paths as a sequence of position values. This can be treated as a time series. LSTMs have achieved good performance when dealing with time series. However, LSTMs have the limitation of only predicting a single path per tracklet. Path prediction is not a deterministic task and requires predicting with a level of uncertainty. Predicting multiple paths instead of a single one is therefore a more realistic manner of approaching this task. In this work, predicting a set of future paths with associated uncertainty was archived by combining LSTMs and MDNs. The evaluation was made on the KITTI and the CityFlow datasets on three type of objects, four prediction horizons and two different points of view (image coordinates and birds-eye view).
Download

Paper Nr:	50
Title:	An Automatic Scenario Generator for Validation of Automated Valet Parking Systems
Authors:	Andrea Tagliavini, Donato Ferraro, Tomasz Kloda and Paolo Burgio
Abstract:	A primary goal of self-driving car manufacturers is to create an autonomous car system that is clearly and demonstrably safer than an average human-controlled car. The real-world tests are expensive, time-consuming and potentially dangerous. The virtual simulation is therefore required. The autonomous driving valet parking is expected to be the first commercially available automated driving function without a human driver at the wheel (SAE Level 4). Although many simulation solutions for the automotive market already exist, none of them features the parking environments. In this paper, we propose a new software virtual scenario generator for the parking sites. The tool populates the synthetics parking maps with objects and actions related to these environments: the cars driving from the drop-off point towards the vacant slots and the randomly placed parked cars, each with a given probability of exiting its slot. The generated scenarios are in the OpenSCENARIO format and are fully simulated in the Virtual Test Drive simulator.
Download

Paper Nr:	60
Title:	Driving Fast but Safe: On Enforcing Operational Limits of a NMPC System
Authors:	Adam Gotlib, Krzysztof Jóskowiak, Piotr Libera, Marcel Kaliński, Jakub Bednarek and Maciej Majek
Abstract:	In this paper, we present a novel approach to Model Predictive Control that allows to explore the largest possible portion of the state–space when still using a low–computational–complexity vehicle model. By introducing additional constraint for acceleration magnitude we are able to stay within the limits where the model gives accurate predictions, while driving with high velocity. This effects a behavior similar to one of a professional racing driver, as the controller is able to balance speed and curvature of the vehicle at any point in time.
Download

Paper Nr:	65
Title:	Integrating Multiscale Deformable Part Models and Convolutional Networks for Pedestrian Detection
Authors:	Wen-Hui Chen, Chi-Wei Kuan and Chuan-Cho Chiang
Abstract:	Pedestrian detection has many real-world applications, such as advanced driver assistance systems, security surveillance, and traffic control, etc. One of the pedestrian detection challenges is the presence of occlusion. In this study, a jointly learned approach using multiscale deformable part models (DPM) and convolutional neural networks (CNN) is presented to improve the detection accuracy of partially occluded pedestrians. Deep convolutional networks provide a framework that allows hierarchical feature extraction. The DPM is used to characterize non-rigid objects on the histogram of oriented gradients (HoG) feature maps. Scores of the root and parts filters derived from the DPM are used as deformable information to help improve the detection performance. Experimental results show that the proposed jointly learned model can effectively reduce the miss rate of CNN-based object detection models tested on the Caltech pedestrian dataset.
Download

Paper Nr:	72
Title:	Trajectory Simulation Tool for Assessment of Active Vehicle Safety Systems
Authors:	Chinmay S. Patil, Taehyun Shim, Jemyoung Ryu and Seunghwan Chung
Abstract:	Advanced Driver Assist Systems (ADAS) have been widely employed in the automotive industry to improve vehicle safety and to reduce the driver’s workload. In addition, there are increasing efforts toward autonomous driving vehicles using enhanced ADAS technologies. For effective ADAS development, it is critical to test and validate these systems. This paper presents a vehicle simulation tool that can be used for various ADAS vehicle test scenarios in which it can generate vehicle trajectories and speed profiles that satisfy user defined test conditions. The proposed simulation tool is useful to design a test scenario in the simulation environment before the physical test. Thus, it can significantly reduce the time needed for the proper test scenario development.
Download

Paper Nr:	74
Title:	Validity Analysis of Simulation-based Testing concerning Free-space Detection in Autonomous Driving
Authors:	Fabio Reway, Maikol Drechsler, Diogo Wachtel and Werner Huber
Abstract:	Automated vehicles must perceive their environment and accordingly plan a safe trajectory to navigate. Camera sensors and image processing algorithms have been extensively used to detect free-space, which is an unoccupied area where a car can safely drive through. To reduce the effort and costs of real test drives, simulation has been increasingly used in the automotive industry to test such systems. In this work, an algorithm for free-space detection is evaluated across real and virtual domains under different environment conditions: daytime, night time and fog. For this purpose, an algorithm is implemented to ease the process of creating ground-truth data for this kind of test. Based on the evaluation of predictions against ground-truth, the test results from the real test scenario are compared with its corresponding virtual twin to analyze the validity of simulation-based testing of a free-space detection algorithm.
Download

Paper Nr:	88
Title:	Efficient and Selective Upload of Data from Connected Vehicles
Authors:	Zaryab Khan and Christian Prehofer
Abstract:	Vehicles are evolving into a connected sensing platform, generating enormous amounts data about themselves and their surroundings. In this work, we focus on the efficient data collection for connected vehicles, exploiting the fact that the context data of cars on the same road is often redundant. This is for instance relevant for applications which need roadside data for map updating. We propose a vehicular data dissemination architecture with a central coordination scheme to avoid redundant uploads. It also uses roadside WiFi hotspots opportunistically. To evaluate the benefits, we use the SUMO simulator to benchmark our results against a baseline solution, showing improvements of factor 10 up to 20.
Download

Paper Nr:	10
Title:	Sweat Detection with Thermal Imaging for Automated Climate Control and Individual Thermal Comfort in Vehicles
Authors:	Diana Schif, Ulrich T. Schwarz and Holger Forst
Abstract:	In addition to autonomous driving, the automation of comfort functions is currently one of the development focuses of the automotive industry. In particular, the automation of the climate function is considered, as manual operation often leads to distraction from the driving task. In order to implement this automation, various data about the vehicle interior and the occupants are needed. Besides interior temperature, gender or air speed, the sweat status of the occupants is relevant. In this work it is examined to what extent the sweat status can be detected with the help of a thermal imaging camera. The aim is to show if it is possible to distinguish the status not sweating, shortly before sweating and sweating using thermal imaging. For this the part of the thermal image showing the forehead is analyzed. More specifically, the difference between minimum and maximum temperature is compared for the different sweat statuses. At an ambient air temperature of 21 °C the thermal comfort level and sweat status of 20 subjects is inquired and skin temperature is measured by a thermal camera during sport activity. Results indicate that there is a significant difference (p < 0.05) between status not sweating and shortly before sweating and also between status not sweating and sweating. Sweat can therefore be detected with the help of thermal imaging cameras. This result provides important input for automated air conditioning. If sweat is detected for one or more occupants, then with the climate control a corresponding regulation can take place to dry the sweat and to prevent further sweating.
Download

Paper Nr:	16
Title:	Development and Implementation of a Concept for the Meta Description of Highway Driving Scenarios with Focus on Interactions of Road Users
Authors:	Raphael Pfeffer, Jingyu He and Sax Eric
Abstract:	Nowadays reducing the individual risk for advanced driver assistant systems (ADAS) and automated driving while guaranteeing the overall safety on the highway remains a big challenge. The identification of corner test cases and driving scenarios is key in the development process but is still not entirely solved. In the past, many contributed to a unified scenario definition but often with different application focus. In this paper, we develop a new scenario meta model based on existing definitions serving a development and test process where the test data is captured in real (test) drives and its contained scenarios are derived. We present the novelty of our scenario model describing the behaviour of dynamic objects in highway situations and show a first application of our model and results calculating the uniqueness of scenarios using auto-encoders.
Download

Paper Nr:	24
Title:	Fuzzy Alarm System based on Human-centered Approach
Authors:	Elena Magán, Agapito Ledezma, Paz Sesmero and Araceli Sanchis
Abstract:	This paper presents an Advanced Driver Assistance System (ADAS), based on a fuzzy logic decision support system and developed by using a multi-agent system. The ADAS is designed so that it can detect dangerous situations on urban environments and alert the driver about them if necessary. For that, it collects data from the car, the car’s surroundings and the driver, and represents the information as an OWL ontology. Then, a fuzzy logic inference system uses this information to evaluate whether there is danger or not. The system can detect 9 dangerous situations by using a repository of 14 fuzzy rules, based on a previous work and expanded on this one. Although with limitations, the results show that the ADAS can alert the driver when the driver is in a dangerous situation.
Download

Paper Nr:	61
Title:	Map Attribute Validation using Historic Floating Car Data and Anomaly Detection Techniques
Authors:	Carl Esselborn, Leo Misera, Michael Eckert, Marc Holzäpfel and Eric Sax
Abstract:	Map data is commonly used as input for Advanced Driver Assistance Systems (ADAS) and Automated Driving (AD) functions. While most hardware and software components are not changed after releasing the system to the customer, map data are often updated on a regular basis. Since the map information can have a significant influence on the function’s behavior, we identified the need to be able to evaluate the function’s performance with updated map data. In this work, we propose a novel approach for map data regression tests in order to evaluate specific map features using a database of historic floating car data (FCD) as a reference. We use anomaly detection methods to identify situations in which floating car data and map data do not fit together. As proof of concept, we applied this approach to a specific use case finding yield signs in the map, which are currently not present in the real world. For this anomaly detection task, the autoencoder shows a high precision of 90% while maintaining an estimated recall of 45%.
Download

Paper Nr:	66
Title:	Integration of Voice Assistant and SmartDeviceLink to Control Vehicle Ambient Environment
Authors:	Ayush Shah and Anastacia Gusikhin
Abstract:	Over the past few years, the popularity of personal voice assistants has grown, particularly for use in the vehicle environment where voice is a preferred mode of interface to minimize driving distractions. Amazon Alexa, one of the most popular voice assistants, has been integrated in many vehicle brands. While existing Alexa car applications provide vehicle occupants with access to a multitude of voice skills in infotainment and smart home control, these applications lack the capability to manage the vehicle’s ambient environment. This paper discusses an efficient and effective integration of Amazon Alexa with vehicle climate control, potentially augmented with brought-in devices, using SmartDeviceLink API. The paper overviews the architecture, Alexa skill development, and examples of dialogue. We also present the results of a customer evaluation of the presented system and directions for future research and development toward Ambient Intelligence.
Download

Paper Nr:	67
Title:	Safety and Dependability of Autonomous Systems in Container Terminals: Challenges and Research Directions
Authors:	Eetu Heikkilä, Timo Malm, Risto Tiusanen and Toni Ahonen
Abstract:	Increasing use of autonomous machine systems is a major trend in port logistics, especially in container handling. Over the past decades, large seaports have automated parts of their operations. Currently, also smaller ports are looking to apply automated and autonomous solutions. This is expected to increase efficiency and safety, but also to introduce new mixed-traffic situations between humans, manual machines and machines of different levels of autonomy. This is likely to introduce safety risks and dependability challenges for system development and operation. In this paper, we discuss selected key challenges that need to be solved to ensure that autonomous container handling solutions can be implemented safely and profitably. We also present topical research directions that are planned and ongoing to solve these challenges.
Download

Paper Nr:	68
Title:	Towards Digitalized and Automated Work Processes in Port Environments
Authors:	Toni Ahonen, Helena Kortelainen and Antti Rantala
Abstract:	Requirements for safety and productivity in container terminal processes are the key drivers for automation and digitalized processes. While automation development has so far been focused on large terminal environments and introduction of smart port concepts and utilization of digital technologies are to large extent limited to megaports and forerunners, there is an increasing interest in implementing automation and digitalization at smaller ports in profitable manner. Current paper discusses the enablers and barriers of automation and digital services at such ports, presents a framework for customer value in the contexts and outlines focal development targets.
Download

Paper Nr:	73
Title:	Case Study: Regulation of Noise Produced by a Rotary-screw Propulsion Unit in an All-terrain Vehicle
Authors:	Umar Vahidov, Alexander Belyaev, Vladimir Makarov, Dmitriy Mokerov and Yuri Molev
Abstract:	The study presents methods developed to calculate permissible level of acoustic radiation produced by a rotary-screw propulsion unit on ice. The study is based on the papers of the researchers who studied acoustic waves generated by construction and road vehicles. The authors of the study applied the aforementioned theories to the case of interaction between a rotary-screw propulsion unit and ice. The paper provides general measuring methods and evaluates how every type of interaction between propulsion unit components and ice affects overall level of generated acoustic pressure. The results and conclusions obtained during the research can be used to help manufacturers select the parameters of the rotary-screw propulsion unit which contribute to reduction of noise inside the cabin of an all-terrain vehicle.
Download

Paper Nr:	89
Title:	Experimental Theoretical Study of the Mobile Robotic System Movement with Caterpillar-modular Propulsion on the Beach Line Terrain
Authors:	Alexander Belyaev, Alexey Papunin, Evgeny Zharkov, Alexey Vasiliev, Vladimir Belyakov and Vladimir Makarov
Abstract:	This article presents the data for mobile robotic system motion modeling with caterpillar-modular propulsion on the sand support base. The study provides the basics of the development of the calculation model in Adams Tracked Vehicle amid mass and geometric chassis parameters and characteristics of nonrigid soil. The study presents the 3D views of the model created. The study provides the fragments of curvilinear motion. The study provides graphs of behavior moments for chassis beads, as well as shows the total resistance to motion on sandy beach. The mean of the moment on one bead during linear motion amounted to 172 Nm, during curvilinear 195 and 217 Nm respectively for backward and overleaping chassis beads. The mean resistance to motion during linear motion amounted to 1606 N, during curvilinear to 1943 N. To validate the results of the modeling we have conducted experimental studies.
Download

Area 4 - Data Analytics

Full Papers

Paper Nr:	3
Title:	Use Case of Quay Crane Container Handling Operations Monitoring using ICT to Detect Abnormalities in Operator Actions
Authors:	Sergej Jakovlev, Tomas Eglynas, Mindaugas Jusis, Saulius Gudas, Valdas Jankunas and Miroslav Voznak
Abstract:	This paper presents the initial research findings from the Klaipeda port monitoring action related to Blue economy development initiative in the Baltic Sea. Use case study demonstrates the possibility to address the problem of information system deployment in harsh industrial environment to gather valuable statistical knowledge. Custom made monitoring and data transmission units were developed to utilize the best practice of engineering to solve real problems of Klaipeda Port. Several key operations and parameters were monitored during the research, including containers spreader movements, physical characteristics of the cables, metal constructions. Initial results suggested that crane operators’ involvement in the control of the cargo movement produced incorrect control patterns (joystick movements) that delayed port operations. Each control movement of the joystick needs to have a direct real-time feedback from the spreader (actual movement of the cargo). Feedback control functionality will allow adjusting the spreader movement according to the operator and will decrease the cargo transportation time during constant breaks.
Download

Paper Nr:	56
Title:	A Visual Analytics System for Processed Videos from Traffic Intersections
Authors:	Ke Chen, Tania Banerjee, Xiaohui Huang, Anand Rangarajan and Sanjay Ranka
Abstract:	Traffic intersections are the most crucial areas that determine the efficiency of a road network. With the advances in technology, it is now possible to gather real-time data on the performance of an intersection and identify potential inefficiencies. The goal of our work is to develop a visual analytics framework based on videos collected at an intersection using fisheye cameras. The software developed as part of this work is described in detail, along with its utility and usability. The software may be used to filter and display tracks and sort them based on the most frequent signaling phases encountered at an intersection. The software may be used to study anomalous trajectories, such as those that have unusual shapes and those that occur at times that violate the ongoing signal phase. While being useful for analyzing the trajectories at an intersection, the software is also convenient for developers seeking to validate algorithms for the trajectory generation process, object classification, preprocessing, and clustering trajectories.
Download

Short Papers

Paper Nr:	31
Title:	Modelling Commuting Activities for the Simulation of Demand Responsive Transport in Rural Areas
Authors:	Sergei Dytckov, Fabian Lorig, Paul Davidsson, Johan Holmgren and Jan A. Persson
Abstract:	For the provision of efficient and high-quality public transport services in rural areas with a low population density, the introduction of Demand Responsive Transport (DRT) services is reasonable. The optimal design of such services depends on various socio-demographical and environmental factors, which is why the use of simulation is feasible to support planning and decision-making processes. A key challenge for sound simulation results is the generation of realistic demand, i.e., requests for DRT journeys. In this paper, a method for modelling and simulating commuting activities is presented, which is based on statistical real-world data. It is applied to Sjöbo and Tomelilla, two rural municipalities in southern Sweden.
Download

Paper Nr:	57
Title:	Clustering Object Trajectories for Intersection Traffic Analysis
Authors:	Tania Banerjee, Xiaohui Huang, Ke Chen, Anand Rangarajan and Sanjay Ranka
Abstract:	Vehicle and pedestrian traffic at a traffic intersection provide crucial information about the performance of the intersection for safety and throughput. It is possible to discover patterns and outliers on this data by applying data analytics. In this paper, we present a novel clustering algorithm for trajectories that use a new distance measure and a two-level hierarchical clustering approach based on geometric properties of the trajectories and spectral clustering. Trajectory data is augmented with signal phasing and timing information, which gives new insights to the trajectory data. We demonstrate the procedure on a real-life intersection where the prominent patterns for traffic movement are found, and the anomalous trajectories are extracted.
Download

Paper Nr:	58
Title:	Data Mining Algorithms for Traffic Interruption Detection
Authors:	Yashaswi Karnati, Dhruv Mahajan, Anand Rangarajan and Sanjay Ranka
Abstract:	Detection of traffic interruptions (caused by vehicular breakdowns, road accidents etc.) is a critical aspect of managing traffic on urban road networks. This work outlines a semi-supervised strategy to automatically detect traffic interruptions occurring on arteries in urban road networks using high resolution data from widely deployed fixed point sensors (inductive loop detectors). The techniques highlighted in this paper are tested on data collected from detectors installed on more than 300 signalized intersections.
Download

Paper Nr:	59
Title:	Qualitative Feature Assessment for Longitudinal and Lateral Control-features
Authors:	Jacob Langner, Christian Seiffer, Stefan Otten, Kai-Lukas Bauer, Marc Holzäpfel and Eric Sax
Abstract:	Control features take over a multitude of driving tasks in today’s vehicles. The complexity of the underlying software code and control parameters has grown to a staggering size. It is no longer viable to test and evaluate features on a pure feature level while driving through real world traffic. The driving tasks and environmental situations are too manifold to be lumped together undifferentiated. As time and resources during development are scarce, test scopes are limited. However, test coverage and representativity are crucially important and can not be neglected. We propose an approach that enables feature evaluation on a driving task basis and achieves holistic assertions for the maturity level even on small test scopes. The approach is based on recorded road tests and is demonstrated with a brief example.
Download

Paper Nr:	70
Title:	Exploring Spatio-temporal Movements for Intelligent Mobility Services
Authors:	Tobias Grüner, Sören Frey, Jens Nahm and Dirk Reichardt
Abstract:	Mobility services can substantially benefit from incorporating movement behavior information. Models of daily travel routines can facilitate intelligent recommendations of suitable car sharing, ride pooling, or Mobility as a Service (MaaS) offerings, for instance. However, existing approaches that infer regular travel activities from historical location data exhibit several limitations. For example, they often have an insufficient resolution in the spatial and temporal dimension or are restricted to predicting only the next location visit. This paper presents an activity-based approach to model daily travel routines and predict regularities with the help of machine learning (ML). We first extract points of interest (POIs) and corresponding visits from historical location data. Then, regularities for these visits are identified with the help of classification. We validate our work in progress approach using data from voluntary, consenting test subjects (CTS) who agreed to track their movements. They labeled their own data for each activity with corresponding regularity information. We show that POI visits can already be predicted reliably for the first classes of movements.
Download

Paper Nr:	76
Title:	Temporal Network Approach to Explore Bike Sharing Usage Patterns
Authors:	Aizhan Tlebaldinova, Aliya Nugumanova, Yerzhan Baiburin, Zheniskul Zhantassova, Markhaba Karmenova and Andrey Ivanov
Abstract:	The bike-sharing systems have been attracting increase research attention due to their great potential in developing smart and green cities. On the other hand, the mathematical aspects of their design and operation generate a lot of interesting challenges for researchers in the field of modeling, optimization and data mining. The mathematical apparatus that can be used to study bike sharing systems is not limited only to optimization methods, space-time analysis or predictive analytics. In this paper, we use temporal network methodology to identify stable trends and patterns in the operation of the bike sharing system using one of the largest bike-sharing framework CitiBike NYC as an example.
Download

Paper Nr:	85
Title:	Developing a Traffic Congestion Model based on Google Traffic Data: A Case Study in Ecuador
Authors:	Yasmany García-Ramírez
Abstract:	Congestion on urban streets has negative impacts on the urban economy, environment, and lifestyle. Congestion, in developing countries, will increase despite knowing its cons. One way to control or reduce congestion is by sharing traffic information through traffic model congestion. This model includes the estimation of the travel time from the desired place of origin-destination. Speed-flow-density parameters help to calculate travel time. These fundamental parameters could be estimated using Floating Car Data from Google. Therefore, the objective of this research is to calibrate equations for the fundamental parameters with traffic state indicators by Google, relating them to ground truth data. Six density-flow equations and six speed-density equations were calibrated using power and linear curve, and some of them were validated. Other cities can use these equations to build their traffic congestion model. With this model, road users can plan the journey and choice the best route or travel in times of low congestion or uptake of public transport, decongesting the city and saving traffic costs related. This comprehensive research extends the knowledge of how Google traffic information can employ in developing cities.
Download

Paper Nr:	86
Title:	New Traffic Congestion Analysis Method in Developing Countries (India)
Authors:	Tsutomu Tsuboi
Abstract:	This manuscript describes introducing new traffic congestion analysis for developing country in India. In generally, it is challenge to show how traffic congestion occurs, especially in developing countries such as in India because its traffic is consisted of various kinds of transportation like two wheelers, three wheelers, and sometimes animals on the roads. There is a chance to collect real traffic flow data in Ahmedabad of Gujarat states of India since October 2014. The traffic monitoring system there consist of 14 traffic monitoring cameras and the system is capable to monitor traffic density, traffic volume, average vehicle speed, and occupancy at the each location. In this manuscript, there are three types of traffic congestion analysis. One is based on its observation traffic flow, in which it compares daily traffic volume and its average vehicle speed. The second one is based on the judgement of occupancy parameter, which uses as one of traffic congestion parameter in general. The third one is based on estimation from “social loss” calculation which comes from the traffic flow theory but is challenge to analyse in the developing countries. The social loss calculation is proven in the traffic theory but it is difficult to define the traffic demand curve, the social cost curve, and the traffic supply curve. Author shows how to make the practical “social loss” calculation and its validation compared with the actual traffic congestion condition.
Download

Paper Nr:	25
Title:	From a Traditional Bicycle to a Mobile Sensor in the Cities
Authors:	Pedro Nunes, Carlos Nicolau, José P. Santos and António Completo
Abstract:	The present study focuses on the development of a web cloud based connectivity platform to apply in the context of cycling. One of the main challenges in using bicycle as mean of transportation is related to the cyclists safety and risk perception. In response to that problem, we propose an IoT based module, that will be embedded in the bicycle, allowing sensor data collection and real time sending to the connectivity platform. The data will be used to perceive cyclists route choice preferences, give support to stakeholders either in making new policies to promote bike use or to give cyclist suggestions about the most convenient route, depending on his profile. The main objective is to transform a traditional bicycle into a mobile sensor in the city. The connectivity platform will enable several services, so the bicycle can easily be integrated in a free floating bike-sharing environment.
Download

Area 5 - Smart Mobility and Sustainable Transport Services

Full Papers

Paper Nr:	33
Title:	Self-aware Pedestrians Modeling for Testing Autonomous Vehicles in Simulation
Authors:	Qazi H. Jan, Jan A. Kleen and Karsten Berns
Abstract:	With the rise of autonomous vehicles in the urban environment, the focus is also shifted towards autonomous vehicles in pedestrian zones. Pedestrian safety becomes the primary concern in such zones. Autonomous systems for these situations need thorough testing before its deployment in the real-world to ensure safety. Therefore, developing testbeds that resemble the real-world for autonomous vehicles testing in pedestrian zones are highly critical. The proposed work focuses on the modeling of pedestrian behaviors in a simulated environment for realizing autonomous vehicles in the pedestrian zones. The virtual pedestrians are modeled with the self-awareness to avoid static and dynamic obstacles when progressing towards its goal. The goal is also to have a minimum number of parameters to generate various test scenarios with realistic behaving pedestrians for the autonomous systems. The proposed system is evaluated using individual and group of virtual pedestrians. It can be seen from the experiments that simulated pedestrians show trajectories which resemble the trajectories of pedestrians in the real-world for that particular situation.
Download

Paper Nr:	71
Title:	Tournament Selection Algorithm for the Multiple Travelling Salesman Problem
Authors:	Giorgos Polychronis and Spyros Lalis
Abstract:	The multiple Travelling Salesman Problem (mTSP) is a generalization of the classic TSP problem, where the cities in question are visited using a team of salesmen, each one following a different, complementary route. Several algorithms have been proposed to address this problem, based on different heuristics. In this paper, we propose a new algorithm that employs the generic tournament selection heuristic principle, hybridized with a large neighbourhood search method to iteratively evolve new solutions. We describe the proposed algorithm in detail, and compare it with a state-of-the-art algorithm for a wide range of public benchmarks. Our results show that the proposed heuristic manages to produce solutions of the same or better quality at a significantly lower runtime overhead. These improvements hold for Euclidean as well as for general topologies.
Download