Beth and Alan are sitting at a Fifth Avenue outdoors restaurant in Manhattan. Alan is telling Beth an exciting story about his vacation in Nice. Alan presents the story through gesture and speech. Then Beth's arm starts moving and her neck stiffens. We, the viewers, know that she's surprised to see an elephant in the middle of Manhattan, and that in 460 milliseconds her arm and hand motion will turn into a welldefined deictic gesture, her eyebrows will rise, and her mouth will open with surprise, at which point Alan will most certainly recognize the signs and look over at the elephant. But right now, at t-minus-460 milliseconds, Beth's gesture is barely recognizable as a communicative action, so Alan doesn't know for sure. And thus, before that all happens, in the next 460 milliseconds, Alan has to decide what to do about Beth's behavior. Should he stop telling his story? Or should he go on, in case Beth is simply adjusting her jacket? Decisions like these a

This paper presents a computational model of real-time task-oriented dialogue skills. The architecture, termed Ymir, bridges between multimodal perception and multimodal action and supports the creation of autonomous computer characters that afford full-duplex, real-time face-to-face interaction with a human. Ymir has been prototyped in software, and a humanoid created, called Gandalf, capable of fluid multimodal dialogue. Ymir demonstrates several new ideas in the creation of communicative computer agents, including perceptual integration of multimodal events, distributed planning and decision making, an explicit handling of real-time, and layered input analysis and motor control with human characteristics. This paper describes the architecture and explains its main elements. Examples ofimplementation and performance are given, and the architectures limitations and possibilities are discussed.

This paper presents a type of scheduling blackboard called whiteboards. Blackboards can simplify construction of systems with large numbers of heterogeneous components requiring a high number of fine-grained interactions. An increase in systems integration, for example in humanoid robotics and intelligent environments, has called for better solutions to support multi-module integration. Whiteboards extend the blackboard model in a number of significant ways that allow them to fill this role. Chief among their features are: an explicit temporal model; quality of service; both publish-subscribe and querying for data; both discrete and streaming data using the same API; explicit data wrappers; programming language independence; as well as a number of solutions to practical issues for improving development effort and runtime performance. Whiteboards consist of (i) a general-purpose message type format, (ii) ontologically-defined message and data stream types, and (iii) specifications for routing between system components. Whiteboards thus provide a development tool especially relevant for simulations of complex natural systems where symbolic data meets raw numerical data; systems with illdefined boundaries between sub-systems; and systems where the number of component states and interactions is considered to be relatively large. Examples include computer vision, speech recognition and robotics, ecosystems and biological systems. This paper describes the main constituents of whiteboards and their use.

An increase in systems integration, for example in humanoid robotics and intelligent environments, has called for better solutions to support multi-module integration. Blackboards can simplify construction of systems with large numbers of heterogeneous components requiring a high number of fine-grained interactions. In this paper we describe the use of a scheduling blackboard used for developing interactive robots. Our blackboards both simplify and extend the blackboard model in a number of ways. Chief among them are: An explicit temporal model; quality of service; publish-subscribe; queries; discrete messaging; streaming data; programming language independence; as well as a number of solutions to practical issues for improving development effort and runtime performance. Whiteboards present a compelling case for the use of scheduling blackboards in robotics, where multiple functionalities and modules need to be integrated into a coherent whole.

This paper compares two solutions for human-like perception using two different modular “plug-and-play” frameworks, CAVIAR (List et al, 2005) and Psyclone (Thórisson et al, 2004, 2005a). Each uses a central point of configuration and requires the modules to be autodescriptive, auto-critical and auto-regulative (Crowley and Reignier, 2003) for fully autonomous configuration of processing and dataflow. This allows new modules to be added to or removed from the system with minimal reconfiguration. CAVIAR uses a centralised global controller (Bins et al, 2005) whereas Psyclone supports a fully distributed control architecture. We implemented a computer vision-based human behaviour tracker for public scenes in the two frameworks. CAVIAR’s global controller uses offline learned knowledge to regulate module parameters and select between competing results whereas in Psyclone dynamic multi-level control modules adjust parameters, data and process flow. Each framework results in two very different solutions to control issues such as dataflow regulation and module substitution. However, we found that both frameworks allow easy incremental development of modular architectures with increasingly complex functionality. Their main differences lie in runtime efficiency and module interface semantics.

Abstract. Innovation is a central element of economic development. Understanding knowledge – its organization and especially its dynamics in a market – becomes therefore the main challenge when explaining economic development in general, and the competitiveness and growth of firms and industries in particular. Past research has generally treated knowledge as a monolithic object rather than a composite dynamic phenomenon. In this paper we present work on a new fine-grain, dynamic, morphogenic model of knowledge that is easy to manage, interpret and extend. This knowledge model is embedded a larger market simulation where selected elements of an economy, including employees, companies, banks and consumers, are modeled at multiple levels of abstraction, from agents to monolithic entities. We present data from early runs of the system, showing predictable results in baseline conditions and product innovation effects using the knowledge representation. The results show the model’s excellent potential to address questions about emergent phenomena related to knowledge evolution, knowledge transfer and knowledge use in market innovation.

A number of present-day problems act to hold back progress in the field of artificial intelligence (A.I.), both theoretical and pragmatic. Among the most serious pragmatic issues has to do with integration and large-scale systems construction, as much recent work on humanoids and interactive robots has

Abstract. One of the many benefits of agent-based modeling is the ability to develop modules in parallel, with teams focusing on isolated modules with well-defined interfaces. This also presents a challenge, however: Integrating a system with a large number of modules with complex interactions, developed by many people, is a significant challenge. Constructionist Design Methodology (CDM) is an approach for building highly modular systems of many interacting components. Originally proposed for use in artificial intelligence research, CDM’s strength lies in simplifying the modeling of complex, multifunctional systems that require architectural evolution of tangled data flow and control hierarchies. We have adapted CDM for the creation of agent-based simulations, resulting in a new version called CDM-S, and used it in the development of a family of agent-based market simulations where selected elements of an economy, including employees, companies, banks and consumers, are modeled at multiple levels of abstraction, from specific knowledge of single individuals to monolithic consumer groups. The systems have been built by a total of 15 Master’s students over a period of 10 weeks in two consecutive periods. Here we describe the CDM-S and present data on the application of