Norbert Oswald

Cooperative Vision in a Multi-Agent Architecture

We present the concept of cooperative vision and its application to a multi-agent system with special attention to the integration of vision. Cooperative vision can be described as a type of distributed vision, where several agents working in a shared environment are involved. The object recognition task was distributed to several agents in order to demonstrate the concept of cooperative vision. This enables, on the one hand, a verification of objects by several agents and, on the other hand, a localization of spatial positions of other agents. A Bayesian approach is used for the combination of conclusions of several agents. Experiments done so far show significant results with regard to both tasks.

Cooperative object recognition

Abstract This paper introduces an approach to recognize objects in a group of observers. The fundamental idea of this approach is based on the usage of cooperation at object level in order to overcome limitations of typical recognition applications which often lead to ambiguous interpretations. By integrating individual hypotheses which were calculated at spatial distributed viewpoints, the robustness of the recognition results can be increased significantly. Experiments that compare cooperative and individual calculated results confirm the benefit of cooperation in recognition tasks. A qualitative method for appearance-based object recognition is used to build hypotheses for individual recognition. Based on an input image which mainly contains the target object, a translation and scale invariant representation is built. From this, a two-dimensional distribution function is generated and the similarity between any two objects is determined by using non-parametric statistical tests. The fusion of distributed information is done by the use of bayesian networks. Such networks enable both a continuous individual and a cooperative object recognition.

Information Management for Unmanned Systems: Combining DL-Reasoning with Publish/Subscribe

Sharing capabilities and information between collaborating entities by using modem information- and communication-technology is a core principle in complex distributed civil or military mission scenarios. Previous work proved the suitability of Service-oriented Architectures for modelling and sharing the participating entities’ capabilities. Albeit providing a satisfactory model for capabilities sharing, pure service-orientation curtails expressiveness for information exchange as opposed to dedicated data-centric communication principles. In this paper we introduce an Information Management System which combines OWL-Ontologies and automated reasoning with Publish/Subscribe-Systems, providing for a shared but decoupled data model. While confirming existing related research results, we emphasise the novel application and lack of practical experience of using Semantic Web technologies in areas other than originally intended. That is, aiding decision support and software design in the context of a mission scenario for an unmanned system. Experiments within a complex simulation environment show the immediate benefits of a semantic information-management and -dissemination platform: Clear separation of concerns in code and data model, increased service re-usability and extensibility as well as regulation of data flow and respective system behaviour through declarative rules.

An Architectural Framework for Cooperative Civil and Military Mission Scenarios

Today’s mission software requires ever more sophistication, being faced with the challenges of conducting Nnetwork-centric Operations (NCO) and exhibiting mission autonomy. Addressing these challenges, together with improving the engineering process in general, is the main output of the project described in this paper, accomplished at EADS Military Air Systems. The project comprises an architectural concept, a methodology, a development environment, a runtime environment and a simulation connector. The adequacy of the approach was demonstrated by means of a prototypical implementation of a NCO scenario.

PLEXIL-DL: Language and Runtime for Context-Aware Robot Behaviour

Faced with the growing complexity of application scenarios social robots are involved with, the perception of environmental circumstances and the sentient reactions are becoming more and more important abilities. Rather than regarding both abilities in isolation, the entire transformation process, from context-awareness to purposive behaviour, forms a robot’s adaptivity. While attaining context-awareness has received much attention in literature so far, translating it into appropriate actions still lacks a comprehensive approach. In this paper, we present PLEXIL-DL, an expressive language allowing complex context expressions as an integral part of constructs that define sophisticated behavioural reactions. Our approach extends NASA’s PLEXIL language by Description Logic queries, both in syntax and formal semantics. A prototypical implementation of a PLEXIL-DL interpreter shows the basic mechanisms facilitating the robot’s adaptivity through context-awareness.

Object Recognition with Multiple Observers

This paper introduces an approach to recognize objects in a group of observers. The fundamental idea of this approach is based on the usage of cooperation at object level in order to overcome limitations of typical recognition applications which often lead to ambiguous interpretations. By integrating individual hypotheses which were calculated at spatial distributed viewpoints, the robustness of the recognition results can be increased significantly. Experiments confirm the benefit of fusing object hypotheses from multiple observers.

A cooperative architecture to control multi-agent based robots

Internal Structure of a Robot. To control a group of cooperating autonomous mobile robots, we use a multi-agent architecture as introduced in [LBLM98]. The structure of a single robot consists of a set of concurrent Elementary Agents situated on different levels of abstraction. The levels are defined by different response time and are structured as reflexive, tactical, and strategical layer. E.g. in the reflexive layer tasks like object recognition and self-localization are processed.

Konfigurationsbasiertes Verfahren zur schnellen Identifikation komplexer Objekte

Wir prasentieren ein sehr schnelles und robustes Verfahren zur Identifikation hochkomplexer 3D-Objekte, das auf der Konfiguration geometrischer Merkmale basiert. Hierfur fuhren wir einen skalierungs- und translationsinvarianten Konfigurationsraum ein, in dem projektive Verzeichnungen kaum Veranderungen hervorrufen. Dadurch wird nur ein geringer Modellraum zur Beschreibung der 3D-Objekte benotigt. Abhangig von der Anzahl der Merkmale, die zur Beschreibung eines Objekts herangezogen werden, fuhren wir zwei effiziente Matchingverfahren ein, die auf dem Konfigurationsraum aufsetzen. Die Verfahren werden derzeit erfolgreich zur Erkennung mobiler Roboter eingesetzt.

Context-sensitive Plan Execution Language for Adaptive Robot Behaviour

Faced with the growing complexity of application scenarios for autonomous robots, context-awareness and adaptivity are becoming more and more essential abilities to determine environmental circumstances and to adapt to them accordingly. While semantic technologies are widely used for the modelling of a robot’s context, purposive robot behaviour is typically described using plan or plan execution languages which lack explicit semantics for context representations. Context-sensitive adaptive behaviour emerges from the whole transformation process though, from context-awareness to the subsequent execution of associated plans. This comprehensive view of adaptivity lacks sufficient treatment in the field of robotics. Addressing this issue, we have augmented the expressive plan execution language PLEXIL, allowing complex context expressions as Description Logic queries to form an integral part of constructs that define sophisticated behavioural reactions. To demonstrate the symbiosis of context-awareness and plan execution, the enhanced language PLEXIL-DL is employed in the avionics system for a reconnaissance mission of an Unmanned Air Vehicle.

Kooperative Bildverarbeitung und Lokalisierung in einem Team von Robotern

In der vorliegenden Arbeit wird der Begriff der visuellen Kooperation bzw. kooperativen Bildverarbeitung definiert, analysiert und in Bezug gesetzt zum Schichtenmodell der Bildverarbeitung. Wie sich zeigen wird, kann durch die Verteilung der Sichtaufgabe auf mehrere raumlich getrennte Beobachter eines Teams die Zuverlassigkeitsrate bei der Erkennung gesteigert werden. Die Vorteile solch eines verteilten Sehens liegen in der Nutzung von potentiellem Wissen des Teams, in der Analyse von Szenen mit statischen oder dynamischen Objekten sowie in der Generierung robuster Hypothesen. Verteilt berechnete Daten lassen sich jedoch im allgemeinen nur dann kombinieren, wenn sie in Bezug zueinander gesetzt werden konnen. Diese Fahigkeit zur Selbstlokalisierung eines Beobachters, z.B. eines Roboters, bildet damit die Voraussetzung zur visuellen Kooperation. Nach einer kurzen Einfuhrung in die Problemstellung der Selbstlokalisierung schliest dieser Bericht mit Anwendungsbeispielen zu visueller Kooperation aus der high-level Bildverarbeitung vorgestellt.