Search

Diese Arbeit untersucht die Performance von DeepHAZMAT, einem Modell für Objekterkennung und -segmentierung, und vergleicht es mit aktuellen Modellen wie YOLOv5, YOLOv7 und YOLOv8. Für diese Studie wurde ein Datensatz entwickelt, der eine Fusion der DeepHAZMAT und FireNet Datensätze darstellt. Dieser Datensatz hat das Ziel, ein größeres Set der Anforderungen an die Erstellung eines ROS2-Objekterkennungsmoduls für einen Roboter, der an der RoboCup Rescue League teilnimmt, zu erfüllen. Die Ergebnisse der Studie deuten darauf hin, dass DeepHAZMAT zwar eine neuartige Kombination aus Objekterkennung und -segmentierung bietet, seine Leistung jedoch in stark eingeschränkten Systemen als unzureichend eingestuft wurde. Alternativ lieferten moderne Objekterkennungsmodelle, insbesondere YOLOv7-tiny und YOLOv8n ohne Segmentierung, eine zufriedenstellende Leistung, die für den Einsatz in einem eingeschränkten System geeignet ist. Darüber hinaus stellt die Arbeit ein Proof of Concept für ein ROS2-Objekterkennungsmodul vor, das verarbeitete Bilder von einem Objekterkennungsmodell zu einem ROS2-Topic über einen einfachen Publisher veröffentlicht. Dieses Modul ist ein wichtiger Schritt, um eine bessere Leistung in Roboter-Wettkämpfen wie der RoboCup Rescue League zu ermöglichen. This thesis investigates the performance of DeepHAZMAT, an object detection and segmentation model, benchmarking it against contemporary models such as YOLOv5, YOLOv7, and YOLOv8. A custom dataset, a fusion of the DeepHAZMAT and Firenet datasets, was developed for this study to fulfil better the specific requirements of creating an object detection module in ROS2 for a robot participating in the RoboCup Rescue League. The study's findings suggest that while DeepHAZMAT offers a novel combination of object detection and segmentation, its performance was found to be insufficient in highly constrained systems. Alternatively, utilising modern object detection models, specifically YOLOv7-tiny and YOLOv8n without segmentation, delivered satisfactory performance suitable for deployment within a constrained system. Furthermore, the thesis introduces a proof-of-concept ROS2 object detection module that publishes processed images from an object detection model to a ROS2 topic via a simple publisher, enabling better performance in robot competition scenarios like the RoboCup Rescue League.

Rescue robots are expected to carry out reconnaissance and dexterity operations in unknown environments comprising unstructured obstacles. Although a wide variety of designs and implementations have been presented within the field of rescue robotics, embedding all mobility, dexterity, and reconnaissance capabilities in a single robot remains a challenging problem. This paper explains the design and implementation of Karo, a mobile robot that exhibits a high degree of mobility at the side of maintaining required dexterity and exploration capabilities for urban search and rescue (USAR) missions. We first elicit the system requirements of a standard rescue robot from the frameworks of Rescue Robot League (RRL) of RoboCup and then, propose the conceptual design of Karo by drafting a locomotion and manipulation system. Considering that, this work presents comprehensive design processes along with detail mechanical design of the robot’s platform and its 7-DOF manipulator. Further, we present the design and implementation of the command and control system by discussing the robot’s power system, sensors, and hardware systems. In conjunction with this, we elucidate the way that Karo’s software system and human–robot interface are implemented and employed. Furthermore, we undertake extensive evaluations of Karo’s field performance to investigate whether the principal objective of this work has been satisfied. We demonstrate that Karo has effectively accomplished assigned standardized rescue operations by evaluating all aspects of its capabilities in both RRL’s test suites and training suites of a fire department. Finally, the comprehensiveness of Karo’s capabilities has been verified by drawing quantitative comparisons between Karo’s performance and other leading robots participating in RRL.

In dieser Arbeit wird sich mit der Erstellung eines Sensorikkonzepts welches speziell für den RoboCup Rescue Wettbewerb entworfen wird beschäftigt. Zu Beginn dieser Arbeit werden verschiedene Varianten von Robotern vorgestellt, um einen Überblick über die Vielseitigkeit von autonomen Robotern zu kreieren. Das Sensorikkonzept soll in einen Roboter integriert werden, der im RoboCup Rescue Wettbewerb teilnimmt. Dazu werden die Aufgaben des Wettbewerbs in der Arbeit genauer beschrieben und analysiert. Durch diese Erläuterung der Aufgaben des Wettbewerbs soll auch die Auswahl der notwendigen Sensoren nachvollziehbar gemacht werden. Autonomie spielt eine wichtige Rolle beim RoboCup, deshalb werden diese Funktionen besonders berücksichtigt. Danach wird die Kategorisierung und die grundlegende Methodik zur Signalerzeugung von Sensoren behandelt. Hier wird ein besonderes Augenmerk auf die Methoden, die für das Konzept relevant sind, gelegt. Im selben Kapitel werden auch die Grundlagen der Navigation und Lokalisierung von autonomen Robotern behandelt. Die Auswahl der Sensoren erfolgt nach den Kriterien des RoboCup Rescue Wettbewerbs und dem Versuch, ein gutes Preis-Leistungs-Verhältnis zu erzeugen. Im Kapitel Hardware Konzept werden die ausgewählten Sensoren näher beschrieben und die Implementierung in das Gesamtkonzept erklärt. Ein Anschlussplan wird erstellt, um die Verdrahtung der Sensoren mit der Steuerkarte zu beschreiben und zu erleichtern. Im Kapitel Software Konzept werden leistungsfähige Softwarelösungen vorgestellt, die in Verbindung mit den ausgewählten Sensorikelementen ein hochwertiges Sensorik Konzept vervollständigen. Im Anschluss wird genauer auf die Kosten des Konzepts eingegangen und ein Vergleich zu alternativen Sensoren ermöglicht. This thesis deals with the creation of a sensor concept which is specially designed for the RoboCup Rescue competition. At the beginning of this work, different variants of robots are presented in order to create an overview of the versatility of autonomous robots. The sensor concept is to be integrated into a robot that takes part in the RoboCup Rescue competition. For this purpose, the tasks of the competition are described in the thesis and analyzed in more detail. This explanation of the tasks of the competition should also make the selection of the necessary sensors comprehensible. Autonomy plays an important role in RoboCup, so these functions are given special consideration. Then the categorization and the basic methodology for signal generation from sensors is dealt with. Particular attention is paid to the methods that are relevant to the concept. The same chapter also covers the basics of navigation and localization of autonomous robots. The sensors are selected according to the criteria of the RoboCup Rescue competition and the attempt to create a good price-performance ratio. In the Hardware Concept chapter, the selected sensors are described in more detail and their implementation in the overall concept is explained. A connection diagram of the sensors to the selected controller board is created to describe and facilitate the wiring of the sensors with the control card. In the Software Concept chapter, powerful software solutions are presented which, in conjunction with the selected sensor elements, complete a high-quality sensor concept. The costs of the concept are then discussed in more detail and a comparison with alternative sensors is made possible.

Ziel dieser vorgestellten BA 1 ist es, in einer Gruppe von sieben Studierenden einen, mit den Regeln der „RoboCup Rescue League“ konformen, Rettungsroboter zu konstruieren, welcher der FH Campus Wien als Erstkonzept für diesen Wettbewerb vorgeschlagen werden soll. Es werden aus drei entwickelten Konzepten die besten Ideen ausgewählt, wodurch ein Einstieg in die „RoboCup Rescue League“ erreicht werden kann. Diese Liga dient, um fortschrittliche Lösungen im Bereich Robotik bereitzustellen. Die primäre Aufgabe ist es die Ersthelfer bei Katastropheneinsätzen zu unterstützen. Dabei soll vor allem das Risiko von Ersthelfern selbst minimiert und die Lebenschance der Opfer erhöht werden. Durch die „RoboCup Rescue League“ werden diese Szenarien in einer kontrollierten Wettbewerbsumgebung simuliert, die die kompetentesten Lösungen von den Vorrunden bis hin zum Finale herausfiltert. Unser Konzept mit der Planung eines Rescue-Roboters konnte im Zuge der Arbeit erfolgreich umgesetzt werden. The aim of this presented BA 1 is to construct a rescue robot in a group of seven students, which complies with the requirements of the "RoboCup Rescue League" and which should be proposed to the FH Campus Wien as the first concept for this competition. The best concepts are selected from three developed drafts, through which an entry into the "RoboCup Rescue League" can be achieved. This league serves to provide advanced robotics solutions. The primary task is to support first aiders in situations of crisis. Above all, the risk of first aiders themselves should be minimized and the chance of survival of the victims should be increased. The "RoboCup Rescue League" simulates these scenarios in a controlled competitive environment that filters out the most competent solutions from the preliminary round to the final round. Our concept with the planning of a rescue robot was successfully implemented in the course of this bachelor thesis.

The coordination of cooperative agents involved in rescue missions is an important open research problem. We consider the RoboCup Rescue Simulation (RCS) challenge, where teams of agents perform urban rescue operations. Previous approaches typically cast such problem as separate single-team allocation problems. However, different teams have complementary capabilities, and therefore some kind of inter-team coordination is desirable for high-quality solutions. Our contribution considers inter-team coordination using Max-Sum. We present a methodology that allows teams in RCS to efficiently assess joint allocations. Furthermore, we show how to reduce the algorithm's computational complexity from exponential to polynomial time by using Tractable High Order Potentials. To the best of our knowledge this is the first time where it has been shown that MS can be run in polynomial time in the RCS challenge without relaxing the problem. Experiments with fire brigades and police agents show that teams employing inter-team coordination are significantly more effective than uncoordinated teams. Moreover, the evaluation shows that our BMS and THOPs method achieves up to 2.5 times better results than other state-of-the-art methods. Copyright © 2015, International Foundation for Autonomous Agents and Multiagent Systems., Work funded by projects DAMAS (TIN2013-45732-C4-4-P), COR (TIN2012-38876-C02-01), the Generalitat of Catalunya grant 2009-SGR-1434, and the Ministry of Economy and Competitivity grant BES-2010-030466.

Coordination of agents involved in rescue missions is an important open research problem. We focus on the RoboCup Rescue Simulation (RCS) challenge, where different teams of agents perform urban rescue operations. Previous approaches typically cast such coordination problem as separate single-team allocation problems, and solve them separately. Our first key contribution is to focus on the max-sum approach, which has been successfully applied in this setting. We show that it is possible to reduce the computational complexity associated to max- sum from exponential to polynomial time. Our empirical evaluation shows that, by using our approach, the fire brigades team obtains significantly better results when compared to state-of- the-art approaches. Our second key contribution is a methodology that allows teams in RCS to make joint allocations. Specifically, our approach supports a modular design, where teams are independently modeled and subsequently connected via well-defined coordination points. To the best of our knowledge, this is the first task-assignment approach in the literature that enables teams in RCS to make simultaneous joint allocations. Experiments with fire brigades and police agents show that teams employing inter-team coordination are significantly more effective than uncoordinated teams., Work funded by projects RECEDIT (TIN2009-13591-C02- 02), AT (CSD2007-0022), COR (TIN2012-38876-C02-01), MECER (201250E053), the Generalitat of Catalunya grant 2009-SGR-1434, and the Ministry of Economy and Competitivity grant BES-2010-030466