SciTePress - Affiliation Profile

Affiliation

Select a Subject
Refine Ontology Subject

University of Southern Denmark, Denmark

Context (please use complete words only)

Field

Sort by

55 Papers

An Integrated Object Detection and Tracking Framework for Mobile Robots

Topics: Mobile Robots and Autonomous Systems; Perception and Awareness; Engineering Applications on Robotics and Automation

Authors: William Juel , Frederik Haarslev , Norbert Krüger and Leon Bodenhagen

Affiliation: SDU Robotics, University of Southern Denmark, Campusvej 55, Odense C, Denmark

Digital Object Identifier: 10.5220/0009888405130520

Proceedings of ICINCO 2020, ISBN: 978-989-758-442-8

Quick Abstract

In this paper, we propose an end-to-end-solution to the problem of multi-object tracking on a mobile robot. The tracking system consists of a process where we project 2D multi-object detections to the robots base frame, using RGB-D sensor data. These detections are then transformed to the map frame using a localization algorithm. This system predicts trajectories of humans and objects in the environment of the robot and can be adapted to work with any detector and track from multiple cameras. The system can then be used to build a temporally consistent costmap to improve navigation strategies.

Improving Semantic Similarity of Words by Retrofitting Word Vectors in Sense Level

Topics: Neural Networks; Natural Language Processing

Authors: Rui Zhang , Peter Schneider-Kamp and Arthur Zimek

Affiliation: Mathematics & Computer Science, University of Southern Denmark, Campusvej 55, Odense, Denmark

Digital Object Identifier: 10.5220/0008953001080119

Proceedings of ICAART 2020, ISBN: 978-989-758-395-7

Quick Abstract

This paper presents an approach for retrofitting pre-trained word representations into sense level representations to improve semantic distinction of words. We use semantic relations as positive and negative examples to refine the results of a pre-trained model instead of integrating them into the objective functions used during training. We experimentally evaluate our approach on two word similarity tasks by retrofitting six datasets generated from three widely used techniques for word representation using two different strategies. Our approach significantly and consistently outperforms three state-of-the-art retrofitting approaches.

Optimal Path Planning for Drone Inspections of Linear Infrastructures

Topics: Spatial Modeling and Reasoning; Storage, Indexing and Searching; Geographic Information Retrieval; Energy Information Systems

Authors: Golizheh Mehrooz and Peter Schneider-Kamp

Affiliation: Dept. of Mathematics & Computer Science, University of Southern Denmark, Denmark

Digital Object Identifier: 10.5220/0009846703260336

Proceedings of GISTAM 2020, ISBN: 978-989-758-425-1

Quick Abstract

Autonomous Beyond Visual Line of Sight (BVLOS) flights represent a huge opportunity in the drone industry due to their ability to monitor larger areas. Autonomous navigation and path planning are essential capabilities for BVLOS flights. In this paper, we introduce the routing component of a path planning system for inspecting linear infrastructures. We explore both a direct algorithm and a transformation algorithm. The direct algorithm is an extension of A* to allow limited routing through air as well as the use of non-logic intersections. The transformation algorithm pre-computes a graph that include edges for routing through air and nodes for non-logic intersections. We implemented both algorithms for routing along a particular type of linear infrastructure, power lines, and validated them through an empirical evaluation at three different scales: the Danish power grid, the French power grid, and the entire European power grid. The test results show that the transformation algorithm allows for sub-second routing performance for a small-to-medium sized power grid. Larger power grids can be routed in less than five seconds, and even an optimal route of more than six thousand kilometers along linear infrastructures from Portugal to Sweden via Russia is found in less than half a minute. All algorithms have been implemented and are available as an open-source Python package for Linear-infrastructure Mission Control (LiMiC).

Real-time On-board Detection of Components and Faults in an Autonomous UAV System for Power Line Inspection

Topics: Convolutional Neural Networks (CNN); Meta-Learning and Deep Networks; IoT and Smart Devices; Image Classification; Object Detection

Authors: Naeem Ayoub and Peter Schneider-Kamp

Affiliation: Department of Mathematics and Computer Science, University of Southern Denmark, Odense, Denmmark

Digital Object Identifier: 10.5220/0009826700680075

Proceedings of DeLTA 2020, ISBN: 978-989-758-441-1

Quick Abstract

The inspection of power line components is periodically conducted by specialized companies to identify possible faults and assess the state of the critical infrastructure. UAV-systems represent an emerging technological alternative in this field, with the promise of safer, more efficient, and less costly inspections. In the Drones4Energy project, we work toward a vision-based beyond-visual-line-of-sight (BVLOS) power line inspection architecture for automatically and autonomously detecting components and faults in real-time on board of the UAV. In this paper, we present the first step towards the vision system of this architecture. We train Deep Neural Networks (DNNs) and tune them for reliability under different conditions such as variations in camera used, lighting, angles, and background. For the purpose of real-time on-board implementation of the architecture, experimental evaluations and comparisons are performed on different hardware such as Raspberry Pi 4, Nvidia Jetson Nano, Nvidia Jetson TX2, and Nvidia Jetson AGX Xavier. The use of such Single Board Devices (SBDs) is an integral part of the design of the proposed power line inspection architecture. Our experimental results demonstrate that the proposed approach can be effective and efficient for fully-automatic real-time on-board visual power line inspection.

Robotic Finger Design Workflow for Adaptable Industrial Assembly Tasks

Topics: Knowledge-Based Systems; Intelligent Robots

Authors: Adam Wolniakowski ^{1

;

2} , Anders Lindvig ² , Nicolai Iversen ³ , Norbert Krüger ² and Aljaž Kramberger ²

Affiliations: ¹ Faculty of Electrical Engineering, Bialystok University of Technology, Bialystok, Poland , ² The Maersk Mc-Kinney Moller Institute, University of Southern Denmark, Odense, Denmark , ³ The Maersk Mc-Kinney Moller Institute, SDU UAS Center, University of Southern Denmark, Odense, Denmark

Digital Object Identifier: 10.5220/0010020100690076

Proceedings of ROBOVIS 2020, ISBN: 978-989-758-479-4

Quick Abstract

In this work, we introduce a web-based system connected to a simulation framework that can be used to facilitate the design of industrial fingers. We provide an overview of the state of the art and of the currently used manual gripper finger design methods prevailing in the industry. With a concrete use case we demonstrate the advantages in terms of quality and saved time for designing the fingers when utilizing our presented framework compared to a common manual method of designing the gripper fingers.

Synthetic Ground Truth for Presegmentation of Known Objects for Effortless Pose Estimation

Topics: Segmentation and Grouping; Object Detection and Localization; Deep Learning for Visual Understanding

Authors: Frederik Haarslev , William Juel , Norbert Krüger and Leon Bodenhagen

Affiliation: SDU Robotics, Maersk Mc-Kinney Molle Institute, University of Southern Denmark, Campusvej 55, 5230 Odense, Denmark

Digital Object Identifier: 10.5220/0009163904820489

Proceedings of VISAPP 2020, ISBN: 978-989-758-402-2

Quick Abstract

We present a method for generating synthetic ground truth for training segmentation networks for presegmenting point clouds in pose estimation problems. Our method replaces global pose estimation algorithms such as RANSAC which requires manual fine-tuning with a robust CNN, without having to hand-label segmentation masks for the given object. The data is generated by blending cropped images of the objects with arbitrary backgrounds. We test the method in two scenarios, and show that networks trained on the generated data segments the objects with high accuracy, allowing them to be used in a pose estimation pipeline.

Wilson Score Kernel Density Estimation for Bernoulli Trials

Topics: Optimization Algorithms; Industrial Automation and Robotics; Engineering Applications on Robotics and Automation

Authors: Lars Sørensen , Simon Mathiesen , Dirk Kraft and Henrik Petersen

Affiliation: SDU Robotics, University of Southern Denmark, Campusvej, Odense, Denmark

Digital Object Identifier: 10.5220/0009816503050313

Proceedings of ICINCO 2020, ISBN: 978-989-758-442-8

Quick Abstract

We propose a new function estimator, called Wilson Score Kernel Density Estimation, that allows to estimate a mean probability and the surrounding confidence interval for parameterized processes with binomially distributed outcomes. Our estimator combines the advantages of kernel smoothing, from Kernel Density Estimation, and robustness to low number of samples, from Wilson Score. This allows for more robust and data efficient estimates compared to the individual use of these two estimators. While our estimator is generally applicable for processes with binomially distributed outcomes, we will present it in the context of iterative optimization. Here we first show the advantage of our estimator on a mathematically well defined problem, and then apply our estimator to an industrial automation process.

A Mixed Neural Network and Support Vector Machine Model for Tender Creation in the European Union TED Database

Topics: e-Government; Tools and Technology for Knowledge Management; Intelligent Information Systems

Authors: Sangramsing Kayte and Peter Schneider-Kamp

Affiliation: Department of Mathematics and Computer Science (IMADA), University of Southern Denmark (SDU), Odense M, 5230 and Denmark

Digital Object Identifier: 10.5220/0008362701390145

Proceedings of KMIS 2019, ISBN: 978-989-758-382-7

Quick Abstract

This research article proposes a new method of automatized text generation and subsequent classification of the European Union (EU) Tender Electronic Daily (TED) text documents into predefined technological categories of the dataset. The TED dataset provides information about the respective tenders includes features like Name of project, Title, Description, Types of contract, Common procurement vocabulary (CPV) code, and Additional CPV codes. The dataset is obtained from the SIMAP-Information system for the European public procurement website, which is comprised of tenders described in XML files. The dataset was preprocessed using tokenization, removal of stop words, removal of punctuation marks etc. We implemented a neural machine learning model based on Long Short-Term Memory (LSTM) nodes for text generation and subsequent code classification. Text generation means that given a single line or just two or three words of the title, the model generates the sequence of a whole sentence. After generating the title, the model predicts the main applicable CPV code for that title. The LSTM model reaches an accuracy of 97% for the text generation and 95% for code classification using Support Vector Machine(SVM). This experiment is a first step towards developing a system that based on TED data is able to auto-generate and code classify tender documents, easing the process of creating and disseminating tender information to TED and ultimately relevant vendors. The development and automation of this system will future vision and understand current undergoing projects and the deliveries by a SIMAP-Information system for European public procurement tenders organisation based on the tenders published by it.

Related Ontology Subjects/Areas/Topics (12)

Applications	Government	Society, e-Business and e-Government
Artificial Intelligence	Intelligent Information Systems	Symbolic Systems
e-Business	Knowledge Management and Information Sharing	Tools and Technology for Knowledge Management
Enterprise Information Systems	Knowledge-Based Systems	Web Information Systems and Technologies

Bayesian Optimization of 3D Feature Parameters for 6D Pose Estimation

Topics: Machine Learning Technologies for Vision; Shape Representation and Matching; Features Extraction; Object Detection and Localization

Authors: Frederik Hagelskjær , Norbert Krüger and Anders Buch

Affiliation: Maersk Mc-Kinney Moller Institute, University of Southern Denmark, Odense and Denmark

Digital Object Identifier: 10.5220/0007568801350142

Proceedings of VISAPP 2019, ISBN: 978-989-758-354-4

Quick Abstract

6D pose estimation using local features has shown state-of-the-art performance for object recognition and pose estimation from 3D data in a number of benchmarks. However, this method requires extensive knowledge and elaborate parameter tuning to obtain optimal performances. In this paper, we propose an optimization method able to determine feature parameters automatically, providing improved point matches to a robust pose estimation algorithm. Using labeled data, our method measures the performance of the current parameter setting using a scoring function based on both true and false positive detections. Combined with a Bayesian optimization strategy, we achieve automatic tuning using few labeled examples. Experiments were performed on two recent RGB-D benchmark datasets. The results show significant improvements by tuning an existing algorithm, with state-of-art performance.

Related Ontology Subjects/Areas/Topics (4)

Computer Vision, Visualization and Computer Graphics		Image and Video Analysis		Shape Representation and Matching
Features Extraction

Impact of Social Welfare Methods on Multi-objective Resource Allocation in Energy Systems

Topics: Optimization Techniques for Efficient Energy Consumption; Evolutionary Algorithms in Energy Applications; Optimization approaches

Authors: Aisha Umair ¹ , Anders Clausen ¹ , Yves Demazeau ² and Bo Jørgensen ¹

Affiliations: ¹ Center for Energy Informatics, University of Southern Denmark, Campusvej 55, Odense and Denmark , ² Laboratoire d’Informatique de Grenoble, Batiment IMAG, 700 avenue Centrale, Grenoble and France

Digital Object Identifier: 10.5220/0007761101790186

Proceedings of SMARTGREENS 2019, ISBN: 978-989-758-373-5

Quick Abstract

Multi-agent resource allocation refers to the distribution of resources among agents. Resource allocation can be particularly challenging if the agents have conflicting objectives over multiple interdependent issues. In such cases, multi-objective optimization methods can be used to find an optimal allocation of resources, that maximizes social welfare. Social welfare refers to the welfare of the entire society of agents and therefore considered as a suitable metric for assessing the overall system performance in multi-agent resource allocation. In this paper we study and discuss different notions of social welfare and investigate their impact on the optimization outcome specifically for the problems comprising multiple conflicting objectives with interdependent issues. To this end, we implement and apply different notions of social welfare to a real-world, complex problem, where a resource domain is responsible for making allocation of energy resources to multiple energy intensive consumers (Commercial Greenhouse Growers). The problem is modeled as a multi-objective optimization context. Our results show how different social welfare methods affect the optimization outcome and result in different socially optimal resource allocations, depending on the behavior we expect from the system.

Related Ontology Subjects/Areas/Topics (4)

Energy and Economy		Evolutionary Algorithms in Energy Applications		Optimization Techniques for Efficient Energy Consumption
Energy-Aware Systems and Technologies

Topic(s):

Filter listed results, by

Publication date

From:

To:

Author

Locate Author in list below

Data pager
Data pager
1 2 3 4 5 6 7 8 9 10 ... Page 1 of 11, items 1 to 10 of 102.

Add Author(s) to filter:

Norbert Kruger (20)

Dirk Kraft (11)

Bo Joergensen (8)

Uffe Kock Wiil (7)

Anders Buch (6)

Leon Bodenhagen (6)

Peter Schneider-Kamp (5)

Emre Baseski (3)

Frederik Haarslev (3)

Lars Johnsen (3)

Conference

Locate Conference in list below

Data pager
Data pager
1 2 Page 1 of 2, items 1 to 10 of 20.

Add Conference(s) to filter:

VISAPP(17)

KMIS(5)

ICINCO(5)

CSEDU(4)

SMARTGREENS(3)

ICEIS(3)

HEALTHINF(2)

ICAART(2)

ICSOFT-EA(2)

ICSOFT-PT(1)

Title:
ISBN:
Year:
Acronym:
Subject:

Title:
Author:
Affiliation:
Subject:

Name:
Affiliation:
Country:
Conference:
Subject:

Name:
Country:
Subject: