Research Projects

Supervision of Autonomous Software Systems.
The main challenge of this project concerns the creation of an adaptive runtime supervision framework that continuously monitors the execution of software systems, evaluates their behavior against the current requirements, and intervenes by deciding which requirements should be added, ignored or weakened. The two main pillars of the envisioned approach are that (i) accurate requirement models can be obtained only at runtime and through learning, and (ii) requirement revision, including requirement approximation, is preferred over system adaptation.

Golden Agents: creative industry and the creation of the Dutch Golden Age.
The main challenge in this software project is to combine Semantic Web technology with multi-agent technology to simulate interactions between different data-sets, and between data-sets and users. Computer agents will not only help researchers in their search through this enormous amount of data from the Golden Age creative industry, but will actually learn from this. Golden Agents will provide their own suggestions to allow researchers ultimately understand the dynamics of the Golden Age. Patterns that explain the success of the Golden Age will possibly lead to relevant insights for the creative industry today.

Agent-based Driver Collaboration and Smart Roads for Traffic Throughput and Road Safety
The main challenge addressed in this proposal is how to improve road traffic, from the perspectives of safety, network performance and ease of driving. Our proposal is to model highways as distributed virtual organizations that continuously monitor the behaviours of vehicles and take necessary measures and sanctions such as redirecting traffic, changing maximum speed, and issuing penalties for violations.

Virtual Organisation
The main problem addressed in this project is how individual software agents can decide whether and in which way it is beneficial to participate in a virtual organization. Our proposal is to provide software agents the ability to represent and reason about the normative system that constitutes the virtual organization.

Programming Cognitive Robotics
The project is concerned with high-level programming of autonomous robots, in particular, how to facilitate the implementation of robots with cognitive architectures in an efficient and effective manner. Our proposal is to extend existing agent programming languages with sensory and action components used for the management of a robot's sensory data and the execution control of its plan/actions.

Coordination and Composition in Multi-Agent Systems
The main problem addressed in this project is how to design and develop executable coordination models at various levels within multi-agent systems and how to integrate these models. The proposal was based on integrated programming languages for exogenous coordination of roles, individual agents, and multi-agent systems.

Virtual characters: Modelling Cognitive Behaviour of Virtual Characters
The problem addressed in this project is how to build virtual characters with believable behaviour by enriching them with mindreading capabilities, i.e., the capability to infer mental states of other agents from their observed behavior. Our proposal was based on abduction mechanism applied to the BDI agents. Assuming that the agent's decision rules are known, we proposed a logical approach to describe and predict the mental state of agents based on their (complete or partial) observable behaviour.

Modularization of 2APL
The main challenge in this project was to extend 2APL, which is a BDI-based agent programming language, with modules. We proposed a modularisation approach to overcome some of the limitations of existing frameworks and unify commonly accepted characteristics of various existing approaches into one single framework. This approach is applied and implemented in the 2APL framework, which is now available on SourceForge (http://apapl.sourceforge.net/).

Dutch Boon Companion
The main problem addressed in this project is how to design and develop companion robots that interact with human users (e.g., children or elderly people) in an effective and realistic manner (e.g., a care-taking robot can entertain elderly people but also act as an alarm system when something unusual is observed). Our proposal was to integrate emotions in the decision models that control the robot's behaviours. Emotions are generally known as a mechanism that makes decisions effective and realistic.

 
PhD Students

Birna van Riemsdijk (completed)
Starting date: 2002
Defense date: 2006
Title: Cognitive Agent Programming: A Semantic Approach

Bas Steunebrink (completed)
Starting date: 2006
Defense date: 2010
Title: The Logical Structure of Emotions

Nick Tinnemeier (completed)
Starting date: 2006
Defense date: 2010
Title: Organizing Agent Organizations: Syntax and Operational Semantics of an Organization-Oriented Programming Language

Lacramioara Astefanoaei (completed)
Staring date: 2006
Defense date: 2011
Title: An Executable Theory of Multi-Agent Systems Refinement

Michal Sindler (completed)
Starting date: 2007
Expected defense date: 2011
Title: Mental State Abduction

Max Knobbout (completed)
Starting date: 2011
Expected defense date: 2015
Title: Virtual Organizations

Pouyan Ziafati (completed)
Starting date: 2011
Expected defense date: 2015
Topic: Programming Cognitive Robotics

Bas Testerink (ongoing)
Starting date: 2012
Expected defense date: 2015
Title: Distributed Normative Organisations

Mahdieh Shadi (ongoing)
Starting date: 2013
Expected defense date: 2017
Title: Collaboration Behavior Enhancement in Co-development Networks