Joscha Bach

Designing Agent Behavior with the Extensible Agent Behavior Specification Language XABSL

Specific behavior description languages prove to be suitable replacements to native programming language like C++ when the number and complexity of behavior patterns of an agent increases. The XML based Extensible Agent Behavior Specification Language (XABSL) also simplifies the process of specifying complex behaviors and supports the design of both very reactive and long term oriented behaviors. XABSL uses hierarchies of behavior modules called options that contain state machines for decision making. In this paper we introduce the architecture behind XABSL, the formalization of that architecture in XML and the software library XabslEngine that runs the formalized behavior on an agent platform. The GermanTeam [9] employed XABSL in the RoboCup Sony Four Legged League competitions in Fukuoka.

Mapping the Landscape of Human-Level Artificial General Intelligence

We present the broad outlines of a roadmap toward human-level artificial general intelligence (henceforth, AGI). We begin by discussing AGI in general, adopting a pragmatic goal for its attainment and a necessary foundation of characteristics and requirements. An initial capability landscape will be presented, drawing on major themes from developmental psychology and illuminated by mathematical, physiological and information processing perspectives. The challenge of identifying appropriate tasks and environments for measuring AGI will be addressed, and seven scenarios will be presented as milestones suggesting a roadmap across the AGI landscape along with directions for future research and collaboration.

Recognizing and Predicting Agent Behavior with Case Based Reasoning

Case Based Reasoning is a feasible approach for recognizing and predicting behavior of agents within the RoboCup domain. Using the method described here, on average 98.4 percent of all situations within a game of virtual robotic soccer have been successfully classified as part of a behavior pattern. Based on the assumption that similar triggering situations lead to similar behavior patterns, a prediction accuracy of up to 0.54 was possible, compared to 0.17 corresponding to random guessing. Significant differences are visible between different teams, which is dependent on the strategic approaches of these teams.

A motivational system for cognitive AI

General Intelligence is not only characterized by the general representation and (relatively) general problem solving capabilities, but also by general motivation. Here, I sketch a framework for an extensible motivational system for cognitive agents, based on research in psychology. It draws on a finite set of pre-defined drives, which relate to needs of the system. Goals are established through reinforcement learning by interacting with an environment.

Mental Models for Robot Control

The paper investigates problems of real time control for complex long term behavior in dynamically changing environments, especially with regard to requirements arising in actual applications. The scenario of robotic soccer (RoboCup) is used for illustration. The proposed Double Pass Architecture avoids some difficulties of layered hybrid architectures. It allows for real time adaptations to new situations even on the higher levels. It implements concepts of bounded rationality, and supports Case Based Reasoning methods.

A Framework for Emergent Emotions, Based on Motivation and Cognitive Modulators

Although traditional appraisal models have been successful tools for describing and formalizing the behavior of emotional agents, they have little to say about the functional realization of affect and emotion within the cognitive processing of these agents. The cognitive architecture MicroPsi addresses emotion and motivation by defining pre-requisites over which affective dynamics and goal-seeking emerge. Here, these pre-requisites are explained in detail, along with a possible approach of using them to model personality traits.

Designing Agents with MicroPsi Node Nets

The MicroPsi agent architecture, which is based mainly on the Psi theory of Dietrich Dorner, describes the interaction of emotion, motivation and cognition of situated agents. The underlying theory has been formulated within the context of psychology but captures numerous aspects of interest for cognitive science, psychological modeling, social simulation and human-machine interaction. MicroPsi is an attempt to formulate this theory in a more abstract and formal way and to make necessary extensions to allow for the implementation of agents for a range of applications. This contribution describes some components of the architecture and introduces a node net formalism, which serves as the primary means of representation, behavior control and modeling of MicroPsi agents.

MicroPsi 2: the next generation of the micropsi framework

The cognitive architecture MicroPsi builds on a framework for simulating agents as neuro-symbolic spreading activation networks. These agents are situated in a simulation environment or fitted with robotic bodies. The current implementation of MicroPsi has been re-implemented from the ground up and is described here.

The AEP Toolkit for Agent Design and Simulation

The design of artificial agents that are meant to model behavioral, cognitive, economic or social structures asks for tools that aid in layout and implementation of agent architectures. To implement agents based on Dorner’s Psi theory of emotion and cognition, our group has introduced a toolkit that assists in designing modular architectures, as well as representational structures, such as semantic networks, control scripts and connectionist structures by means of a graphical editor. At the same time, the framework supports the inclusion of functionality written in a native programming language. This paper gives an overview over the implementation of agents according to Dorner’s theory, and while it also aims at giving an insight into the functioning of these agents (which we call “MicroPsi” agents), its main purpose is the explanation of the use of the toolkit.

AT Humboldt 2000 (Team Description)

Our agent team AT Humboldt 2000 is partly an extension of our former team AT Humboldt 99[2,3]. Again we used a BDI architecture. Especially the world model and some skills where revised. A new timing concept and a completely different architecture for the deliberation component were developed. The actual development was subject of an undergraduate course. Because of problems with the integration of components developed by different work groups, we were forced to start in RoboCup 2000 with a mixed team, consisting mainly of an extended version of the players used in EuRoboCup 2000. Only the goalie used all our new concepts.