Diploma/Master's/Bachelor's Theses and Job Offers
Here's the general procedure of a thesis at the Knowledge-Based Systems Group:
- Thesis Work
- Final Presentation
Recent work in the field of game AI has been leaning towards creating agents able to play games and compete with human professionals. There exists several approaches to build such agents. Whether they are built by hand or by hard coding or using some learning techniques such as Reinforcement Learning (RL) or Genetic Programming (GP). However, Such approaches are based purely on observations.
The Cluster of Excellence "Internet of Production" is a huge interdisciplinary project with more than 25 institutes at the RWTH. The KBSG is part of that cluster and is involved in decision support for production processes. That includes modelling and planning of processes, and using AI techniques to optimise these processes.
The KBSG is looking for a new HiWi to assist in a research project on Optimizing the Performance of Robot Fleets in Production Logistics Scenarios Using SMT Solving
The Winograd Schema Challenge (WSC) has been proposed as an alternative to the Turing Test for measuring a machine’s intelligence by letting it solve pronoun resolution problems that cannot be tackled by statistical analysis alone, but require commonsense, everyday background knowledge and some form of deeper "understanding" of the question. WSCs are thus hard to solve for machines, but easy for humans.
Many solutions so far are based on machine learning and natural language processing, and achieve results that are hardly better than guessing. Moreover, most knowledge-based approaches to the WSC have been purely theoretical. The goal of this thesis was to develop and implement a knowledge-based WSC solver. In particular, a logic of conditional beliefs called BO is employed that is capable of dealing with incomplete or even inconsistent information (which commonsense knowledge often is). It does so by formalising the observation that humans often reason by picturing different contingencies of what the world could be like, and then choose to believe what is thought to be most plausible. Relevant commonsense background information furthermore is obtained from the ConceptNet semantic network and translated into BO, and processed by the Limbo reasoner.
Humanity has crossed the line from being a rural to urban species since 2007. For the first time in history, more people live in cities and urban areas than in the countryside. Starting in the developed ations, where the urbanisation process has been significantly decelerated in the meantime, urbanisation has especially increased in Asia and South America as well as in Africa to a substantial extent in the second half of the last century. Processes of urbanisation have a negative influence on the availability and quality of water resources. Especially in developing and emerging countries, the hydrological and hydrogeological setting of each region is deteriorated through the growing urbanization processes. Often the hydrological and hydrogeological basis of an area is strongly affected by rocesses of urbanisation in these countries. Changes of the structure of urban development going along with the urbanisation will not be without consequences for the environment and water resources.
In frame of this background the Department of Engineering Geology and Hydrogeology of the RWTH Aachen University analysis the interaction between high speed urbanisation/mega-urbanisation and water resources in China and India. In context to this we want to develop a knowledge-based planning and simulation framework.
The goal of this thesis was to integrate Answer Set Programming (ASP) into a Golog system in order to obtain an agent framework that is capable of efficient non-monotonic reasoning with introspection.
The aim in General Game Playing (GGP) is to create programs called agents that are able to play a yet unknown game after they are given the rules. These programs must thus be as intelligent and as independent as possible to solve problems on their own. A similar research field is represented by Golog and the Situation-Calculus that are well-studied languages and allow reasoning about dynamic domains and diversified sets of problems.
All agents must be able to reason about actions. Hence one approach is to use the Situation-Calculus and Golog to represent games and to express strategies in such games. By introducing Golog to GGP we allow results and methods from both areas to be applied in GGP and vice versa. Accordingly we present an exemplary Golog setup and how to realize it in order to build a General Game Player. For this purpose we discuss how one can overcome the difficulties of translating a formalized game description to its corresponding Golog representation and show how simulations of games can be utilized with methods from Programming by Demonstration (PbD) in order to dynamically develop and maintain executable game strategies during runtime.
We show that our proof of concept agent GologPlayer is able to play turn-based games, supports most Game Description Language (GDL) features and show that it is able to beat randomly playing agents with a success rate of about 70%.