Showing papers presented at "Computer Aided Systems Theory in 2019"

Lane Detection and Classification Using Cascaded CNNs

Abstract: Lane detection is extremely important for autonomous vehicles. For this reason, many approaches use lane boundary information to locate the vehicle inside the street, or to integrate GPS-based localization. As many other computer vision based tasks, convolutional neural networks (CNNs) represent the state-of-the-art technology to indentify lane boundaries. However, the position of the lane boundaries w.r.t. the vehicle may not suffice for a reliable positioning, as for path planning or localization information regarding lane types may also be needed. In this work, we present an end-to-end system for lane boundary identification, clustering and classification, based on two cascaded neural networks, that runs in real-time. To build the system, 14336 lane boundaries instances of the TuSimple dataset for lane detection have been labelled using 8 different classes. Our dataset and the code for inference are available online.

IEC 61499 Runtime Environments: A State of the Art Comparison

Abstract: Networked automation devices, as needed for Industry 4.0 or Cyber Physical Production Systems, demand for new programming languages like the one defined in the IEC 61499 standard. IEC 61499 was originally released in 2005. Since then, different runtime environments—academic and commercial—surfaced: They partly differ in their execution semantics and behavior, and in the features they offer, e.g. Multitasking, Real-time performance, or Dynamic Reconfiguration. Users who want to apply this standard to their problem have to choose the right tool. This paper compares a selection of IEC 61499 runtime environments and outlines topics for further research.

An Integrated Decision-Making Framework for Shuttle Bus Selection Using DEMATEL and MULTIMOORA Methods

Abstract: In the present study we integrate the Decision Making Trial and Evaluation Laboratory (DEMATEL) and the Multi-Objective Optimization by Ratio Analysis and Full Multiplicative Form (MULTIMOORA) methods to form a Multi-Criteria Decision Making process, in which we seek the solution for the problem of ranking the suitable options for vehicles types to be used in a shuttle bus fleet that serves the internal transport of a university campus.

Autonomous Vehicle Architecture for High Automation

Abstract: This work presents the new research platform of the Intelligent System Lab (LSI), a fully functional self driven vehicle that is used by this laboratory to research in the autonomous vehicles field. There exist many research works in perception, path planning and control, but software and hardware architecture is hardly mentioned. This paper is focused on the architecture design of this vehicle, focusing on making it very flexible and to allow testing and validating state of the art algorithms. Furthermore, all the elements needed for that automation are detailed: the modules and algorithms used and how they are connected between them, the sensors it uses, its locations and all the additional elements needed to set up an autonomous vehicle (low level control design, computers, network, ...). This platform is tested and showed in a public demonstration of an international conference, proving its flexible architecture and good performance.

Modeling of Smart Contracts in Blockchain Solution for Renewable Energy Grid

Abstract: The paper presents the manner of smart contracts’ modeling in blockchain solution. The authors illustrate the modeling approach using the example of a renewable energy system. The paper proposes extension of Unified Modeling Language (UML) in form of UML Profile for Smart Contracts, appropriate for smart contract design. Moreover, the standard smart contract design and implementation method, in Corda environment, has been made more flexible. The authors present static aspect of newly designed Smart Contract Design Pattern.

CNNs for Fine-Grained Car Model Classification

Abstract: This paper describes an end-to-end training methodology for CNN-based fine-grained vehicle model classification. The method relies exclusively on images, without using complicated architectures. No extra annotations, pose normalization or part localization are needed. Different full CNN-based models are trained and validated using CompCars [31] dataset, for a total of 431 different car models. We obtained a top-1 validation accuracy of 97.62% which substantially outperforms previous works.

A Mixed Integer Linear Programming Formulation for Green Vehicle Routing Problem: Case for Shuttle Services

Abstract: In the present study, we focus on a two-stage Green Vehicle Routing Problem that we have formulated to investigate the shuttle routing plan for a relatively small and specific case. We seek in the latter stage the optimal speed profile for a route that is determined according to the shortest paths in the former stage. Our findings from the case, where two shuttles lines serve the internal area of a university campus, reveal that the two-stage formulation we propose is an efficient alternative to be utilized especially for practical applications in real life, where there is a smooth traffic with low speeds and no significant effect of the topography.

Kernel Adaptive Filters: A Panacea for Self-interference Cancellation in Mobile Communication Transceivers?

Abstract: In frequency division duplex transceivers, the non-ideal analog duplexer has only a limited stop-band attenuation, and therefore a part of the transmit signal leaks into the receive path. Although operating on a different frequency band, non-ideal effects in the receive path cause different kinds of self-interferences, which can have a higher power level than the actual wanted receive signal. A possible way to tackle this problem are adaptive filters. These approaches are mostly model based, and for each kind of interference a different algorithm is needed. Kernel adaptive filtering offers the possibility to deal with different sorts of interferences with the same algorithm. In this work, we investigate the capabilities of kernel adaptive filtering to cancel especially the second-order intermodulation distortion (IMD2) and the transmitter (Tx)-harmonics interference.

Ensemble Learning for Heartbeat Classification Using Adaptive Orthogonal Transformations

Abstract: In this work, we are focusing on the problem of heartbeat classification in electrocardiogram (ECG) signals. First we develop a patient-specific feature extraction scheme by using adaptive orthogonal transformations based on wavelets, B-splines, Hermite and rational functions. The so-called variable projection provides the general framework to find the optimal nonlinear parameters of these transformations. After extracting the features, we train a support vector machine (SVM) for each model whose outputs are combined via ensemble learning techniques. In the experiments, we achieved an accuracy of \(94.2\%\) on the PhysioNet MIT-BIH Arrhythmia Database that shows the potential of the proposed signal models in arrhythmia detection.

Hash-Based Tree Similarity and Simplification in Genetic Programming for Symbolic Regression

Abstract: We introduce in this paper a runtime-efficient tree hashing algorithm for the identification of isomorphic subtrees, with two important applications in genetic programming for symbolic regression: fast, online calculation of population diversity and algebraic simplification of symbolic expression trees. Based on this hashing approach, we propose a simple diversity-preservation mechanism with promising results on a collection of symbolic regression benchmark problems.

Verification of Architectural Views Model 1+5 Applicability

Abstract: The paper presents architectural views model 1+5 which has been proposed for designing integration solutions of collaborating software systems. The author has introduced modeling extensions of Unified Modeling Language (UML) in form of UML profiles. The author has proposed an Integration flow diagram which is special form of UML activity diagram. The diagram arranges mediation mechanisms from UML Profile for Integration Flows into an integration flow. The paper presents transformations of model-to-model and model-to-code types which automate design of integration platform. The 1+5 was successfully applied in Service-Oriented Architecture. The approach reveals its potential in Domain-Driven Design, Micro-services and blockchain solutions.

Bluetooth low energy technology applied to indoor positioning systems: an overview

Abstract: Indoor Positioning Systems (IPS) are an alternative to Global Positioning System (GPS) in those environments where its signal is attenuated. This is one of the main reasons why IPS has been the subject of much research over the last two decades and where different technologies and methods have been used. Among the technologies used in IPS are those that use radio frequency (RF) signals, such as Bluetooth Low Energy (BLE) or Wi-Fi. BLE is widely used in ubiquitous computing and in many applications of the Internet of Things (IoT) mainly due to its low power consumption and because it can provide advanced services to users. The aim of this paper is to provide an overview of the state of the art of BLE-based IPS including its main methods and algorithms.

Some Remarks on the Prendiville Model in the Presence of Jumps

Abstract: The inhomogeneous Prendiville process in the presence of catastrophes at a generic state of the space of the states is considered. The transition probabilities and the moments are determined in closed form for the homogeneous case and when the intensities of the involved processes have the same time dependence.

White Box vs. Black Box Modeling: On the Performance of Deep Learning, Random Forests, and Symbolic Regression in Solving Regression Problems

Abstract: Black box machine learning techniques are methods that produce models which are functions of the inputs and produce outputs, where the internal functioning of the model is either hidden or too complicated to be analyzed. White box modeling, on the contrary, produces models whose structure is not hidden, but can be analyzed in detail. In this paper we analyze the performance of several modern black box as well as white box machine learning methods. We use them for solving several regression and classification problems, namely a set of benchmark problems of the PBML test suite, a medical data set, and a proteomics data set. Test results show that there is no method that is clearly better than the others on the benchmark data sets, on the medical data set symbolic regression is able to find the best classifiers, and on the proteomics data set the black box modeling methods clearly find better prediction models.

VNS and PBIG as Optimization Cores in a Cooperative Optimization Approach for Distributing Service Points

Abstract: We present a cooperative optimization approach for distributing service points in a geographical area with the example of setting up charging stations for electric vehicles. Instead of estimating customer demands upfront, customers are incorporated directly into the optimization process. The method iteratively generates solution candidates that are presented to customers for evaluation. In order to reduce the number of solutions presented to the customers, a surrogate objective function is trained by the customers’ feedback. This surrogate function is then used by an optimization core for generating new improved solutions. In this paper we investigate two different metaheuristics, a variable neighborhood search (VNS) and a population based iterated greedy algorithm (PBIG) as core of the optimization. The metaheuristics are compared in experiments using artificial benchmark scenarios with idealized simulated user behavior.

Identification of Dynamical Systems Using Symbolic Regression.

Abstract: We describe a method for the identification of models for dynamical systems from observational data. The method is based on the concept of symbolic regression and uses genetic programming to evolve a system of ordinary differential equations (ODE).

ROS-Based Architecture for Multiple Unmanned Vehicles (UXVs) Formation

Abstract: A formation is recognized in different animals as a result of the cooperative behavior among its members, where each member maintains a specific distance and orientation with respect to the others in movement. In the robotics world, a group of robots is defined as Multi-Robot System (MRS), which are used in many applications, such as search and rescue or reconnaissance and surveillance. In these applications, MRS require robots to self-organize to solve complex tasks with better overall performance, in terms of maximizing spatial/temporal coverage or minimizing mission completion time. Consequently, this work presents a robust, heterogeneous and scalable architecture based on ROS for the formation of multiple robots, supported by a dynamic leader and virtual grid algorithms. The algorithm runs on all robots, so they select the leader based on an objective cost function. The leading robot then obtains the formation positions of the other robots and assigns them to the desired position. The assignment of positions is optimized and the proposed architecture has been validated using a simulation environment, which can be easily exported to real robots.

On the Successive Passage Times of Certain One-Dimensional Diffusions

Abstract: We study the distribution of the nth-passage time of a one-dimensional diffusion, obtained by a space or time transformation of Brownian motion, through a constant barrier a. Some explicit examples are reported.

Analyzing Network-Wide Effects of Cooperative Adaptive Cruise Control Without Traffic Signal Control at Intersections

Abstract: In this paper, a problem to jointly optimize the performances of vehicular traffic flow, i.e., the total time spent, the number of stop-and-go movements, and the total emissions, is handled to investigate and discuss the effectiveness of the penetration rates of cooperatively controlled vehicles in mixed traffic. A simulation-based solution is sought over the SUMO micro-simulation environment considering three hypothetical road networks and varying demand profiles.

Analysis on Pedestrian Green Time Period: Preliminary Findings from a Case Study

Abstract: In the present study, we concentrate on the problem of optimization and the usage of green times for pedestrian traffic at crosswalks. We conduct a simulation based study that is calibrated with the field data collected. We propose a frame for the signal controllers in an adaptive fashion. Result from micro-simulation strengthens the finding that a trade-off between the pedestrian travel times and vehicles’ delays has to be made.

Concept for a Technical Infrastructure for Management of Predictive Models in Industrial Applications

Abstract: With the increasing number of created and deployed prediction models and the complexity of machine learning workflows we require so called model management systems to support data scientists in their tasks. In this work we describe our technological concept for such a model management system. This concept includes versioned storage of data, support for different machine learning algorithms, fine tuning of models, subsequent deployment of models and monitoring of model performance after deployment. We describe this concept with a close focus on model lifecycle requirements stemming from our industry application cases, but generalize key features that are relevant for all applications of machine learning.

A Model-Based Learning Approach for Controlling the Energy Flows of a Residential Household Using Genetic Programming to Perform Symbolic Regression

Abstract: In recent years, renewable energy resources have become increasingly important. Due to the fluctuating and changing environment, these energy sources are not permanently available. At certain times, e.g. a photovoltaic (PV) power plant can only generate little or no electricity at all. This is why energy management systems (EMS), which store, use and distribute the available energy as optimally as possible, have been strongly promoted and further developed recently.

Surrogate-Assisted Fitness Landscape Analysis for Computationally Expensive Optimization

Abstract: Exploratory fitness landscape analysis (FLA) is a category of techniques that try to capture knowledge about a black-box optimization problem. This is achieved by assigning features to a certain problem instance utilizing only information obtained by evaluating the black-box. This knowledge can be used to obtain new domain knowledge but more often the intended use is to automatically find an appropriate heuristic optimization algorithm [9]. FLA-based algorithm selection and parametrization hinges on the idea, that, while no optimization algorithm can be the optimal choice for all black-box problems, algorithms are expected to work similarly well on problems with similar statistical characteristics [8, 15].

Decision Diagram Based Limited Discrepancy Search for a Job Sequencing Problem.

Abstract: We consider the Price-Collecting Job Sequencing with One Common and Multiple Secondary Resources problem. The task is to feasibly schedule a subset of jobs from a given larger set. Each job needs two resources: a common resource for a part of the job’s execution time and a secondary resource for the whole execution time. Furthermore each job has one or more time windows and an associated prize. In addition to previous work, we also consider precedence constraints on the jobs. We aim to maximize the total prize over the actually scheduled jobs. To solve large instances heuristically we propose a hybrid of limited discrepancy search and beam search approach that utilize a relaxed decision diagram. We could show that the use of a relaxed decision diagram substantially speed-up the computation times of the search approach.

Surrogate-Assisted Multi-Objective Parameter Optimization for Production Planning Systems

Abstract: Efficient global optimization is, even after over two decades of research, still considered as one of the best approaches to surrogate-assisted optimization. In this paper, material requirements planning parameters are optimized and two different versions of EGO, implemented as optimization networks in HeuristicLab, are applied and compared. The first version resembles a more standardized version of EGO, where all steps of the algorithm, i.e. expensive evaluation, model building and optimizing expected improvement, are executed synchronously in sequential order. The second version differs in two aspects: (i) instead of a single objective, two objectives are optimized and (ii) all steps of the algorithm are executed asynchronously. The latter leads to faster algorithm execution, since model building and solution evaluations can be done in parallel and do not block each other. Comparisons are done in terms of achieved solution quality and consumed runtime. The results show that the multi-objective, asynchronous optimization network can compete with the single-objective, synchronous version and outperforms the latter in terms of runtime.

Concept Drift Detection with Variable Interaction Networks.

Abstract: The current development of today’s production industry towards seamless sensor-based monitoring is paving the way for concepts such as Predictive Maintenance. By this means, the condition of plants and products in future production lines will be continuously analyzed with the objective to predict any kind of breakdown and trigger preventing actions proactively. Such ambitious predictions are commonly performed with support of machine learning algorithms. In this work, we utilize these algorithms to model complex systems, such as production plants, by focussing on their variable interactions. The core of this contribution is a sliding window based algorithm, designed to detect changes of the identified interactions, which might indicate beginning malfunctions in the context of a monitored production plant. Besides a detailed description of the algorithm, we present results from experiments with a synthetic dynamical system, simulating stable and drifting system behavior.

Investigating the Dynamic Block Relocation Problem.

Abstract: The dynamic block relocation problem is a variant of the BRP where the initial configuration and retrieval priorities are known but are subject to change during the implementation of an optimized solution. This paper investigates two kinds of potential changes. The exchange of assigned priorities between two blocks and the arrival of new blocks. For both kind of events we present algorithms that can adjust an existing solution to the changed situation. These algorithms are combined with a branch and bound based solver to enable online optimization with look-ahead. Our experiments show that the algorithms enable finding better solutions in a shorter time after a event occurs.

Simulation-Based Design and Evaluation of a Smart Energy Manager.

Abstract: In this work we describe an advanced development environment for energy management systems using a combination of a Modelica-based simulation tool for multi-physics systems and a controller implemented in the Python scripting language, exchanging information via the FMI (Functional Mockup Interface) standard. As an example, we present the development of a simple but robust Electric Vehicle (EV) charging controller for a smart home with a Photo Voltaic system. The performance of the controller is evaluated using conventional criteria like annual energy costs and peak load plus two newly developed customer satisfaction indicator (CSI) functions.

Lotfi Zadeh: Fuzzy Sets and Systems

Abstract: We give an outline of the history of the origin of the theory of Fuzzy Sets and Systems in the life of its founder Lotfi A. Zadeh. We present roots of this theory in the developments of computers, systems and information theory.

Hand Gesture Recognition Using Computer Vision Applied to Colombian Sign Language

Abstract: In this document we describe a hand gesture classification system of the Colombian Sign Language for both dynamic and static signs, based on Computer Vision and Machine learning. The proposed processes sequence is divided in four stages: acquisition of RGB-D image, extraction of the blob closest to the sensor, detection and validation of the hand, and classification of the sign entered. The results obtained are for multi-class classifiers with a self-captured dataset of 3.600 samples. As a conclusion we found that the best choice for descriptor-classifier according to sign type are HOG-SVM for static signs with an accuracy of \(98\%\), and SVM classifier besides the trajectory-based descriptor with an accuracy of \(94\%\).