Much effort has been spent on suggesting and implementing new architectures of Multi-Agent Systems. However, we believe the time has come to compare and evaluate these architectures in a more systematic way. Rather than just studying a particular application, we suggest that more general problem domains corresponding to sets of applications should be studied. Similarly, we argue that it is more useful to study the properties of classes of multi-agent system architectures than particular architectures. Also, it is important to evaluate the architectures in several dimensions, both different performance-related attributes, which are domain dependent and more general quality attributes, such as, robustness, modifiability, and scalability. As a case study we investigate the general problem of "dynamic resource allocation" and present four classes of multi-agent system architectures that solve this problem. These classes are discriminated by their degree of distribution of control and degree of synchronization. Finally, we instantiate each of these architecture classes and evaluate, through simulation experiments, how they solve a concrete dynamic resource allocation problem, namely load balancing and overload control of Intelligent Networks.

Based on a classification of artificial societies and the identification of four different types of stakeholders in such societies, we investigate the potential of norm-governed behavior in different types of artificial societies. The basis of the analysis is the preferences of the stakeholders and how they influence the state of the society. A general conclusion drawn is that the more open a society is the more it has to rely on agent owners and designers to achieve norm-governed behavior, whereas in more closed societies the environment designers and owners may control the degree of norm-governed behavior.

The aim of the PsyIntEC project is to explore affective and cognitive modeling of humans in human-robot interaction (HRI) as a basis for behavioral adaptation. To achieve this we have explored human affective perception of relevant modalities in human-human and human-robot interaction on a collaborative problem-solving task using psychophysiological measurements. The experiments conducted have given us valuable insight into the communicational and affective queues interplaying in such interactions from the human perspective. The results indicate that there is an increase in both positive and negative emotions when interacting with robots compared to interacting with another human or solving the task alone, but detailed analysis on shorter time segments is required for the results from all sensors to be conclusive and significant.

Bots for real-time strategy (RTS) games may be very challenging to implement. A bot controls a number of units that will have to navigate in a partially unknown environment, while at the same time avoid each other, search for enemies, and coordinate attacks to fight them down. Potential fields are a technique originating from the area of robotics where it is used in controlling the navigation of robots in dynamic environments. Although attempts have been made to transfer the technology to the gaming sector, assumed problems with efficiency and high costs for implementation have made the industry reluctant to adopt it. We present a multiagent potential field-based bot architecture that is evaluated in two different real-time strategy game settings and compare them, both in terms of performance, and in terms of softer attributes such as configurability with other state of-the-art solutions.We show that the solution is a highly configurable bot that can match the performance standards of traditional RTS bots. Furthermore, we show that our approach deals with Fog of War (imperfect information about the opponent units) surprisingly well.We also show that a multiagent potential field-based bot is highly competitive in a resource gathering scenario.

The Turing Test Track of the Mario AI Championship focused on developing human-like controllers for a clone of the popular game Super Mario Bros. Competitors participated by submitting AI agents that imitate human playing style. This paper presents the rules of the competition, the software used, the voting interface, the scoring procedure, the submitted controllers and the recent results of the competition for the year 2012. We also discuss what can be learnt from this competition in terms of believability in platform games. The discussion is supported by a statistical analysis of behavioural similarities and differences among the agents, and between agents and humans. The paper is co-authored by the organizers of the competition (the first three authors) and the competitors.

Ms. Pac-Man, one of the classic arcade games has recently gained attention in the field of game AI through the yearly competitions of various kinds held at e.g. CIG. We have implemented an Influence Map-based controller for Ms. Pac-Man as well as for the ghosts within the game. We show that it is able to handle a number of various situations through the interesting behaviors emerging through the interplay of the different maps. It is also significantly better than the previous implementations based on similar techniques, such as potential fields.