Andrew Garland

Learning hierarchical task models by defining and refining examples

Task models are used in many areas of computer science including planning, intelligent tutoring, plan recognition, interface design, and decision theory. However, developing task models is a significant practical challenge. We present a task model development environment centered around a machine learning engine that infers task models from examples. A novel aspect of the environment is support for a domain expert to refine past examples as he or she develops a clearer understanding of how to model the domain. Collectively, these examples constitute a test suite that the development environment manages in order to verify that changes to the evolving task model do not have unintended consequences.

A plug-in architecture for generating collaborative agent responses

We describe an implemented architecture for programming the responses of collaborative interface agents out of easily composable and reusable plug-in components, and discuss the underlying theoretical and practical issues. The power of the architecture comes primarily from a rich representation of collaborative discourse state, which includes a focus stack and plan tree. The architecture also provides a useful separation between the principles and preferences underlying an agents behavior. We illustrate the use of plug-ins in a complex tutoring agent, which includes plug-ins that diagnose incorrect actions and explain why a step needs to be done. Plug-ins are part of the COLLAGEN agent-building middleware, which has been used by a number of researchers in addition to its developers.

Autonomous Agents that Learn to Better Coordinate

A fundamental difficulty faced by groups of agents that work together is how to efficiently coordinate their efforts. This coordination problem is both ubiquitous and challenging, especially in environments where autonomous agents are motivated by personal goals.Previous AI research on coordination has developed techniques that allow agents to act efficiently from the outset based on common built-in knowledge or to learn to act efficiently when the agents are not autonomous. The research described in this paper builds on those efforts by developing distributed learning techniques that improve coordination among autonomous agents.The techniques presented in this work encompass agents who are heterogeneous, who do not have complete built-in common knowledge, and who cannot coordinate solely by observation. An agent learns from her experiences so that her future behavior more accurately reflects what works (or does not work) in practice. Each agent stores past successes (both planned and unplanned) in their individual casebase. Entries in a casebase are represented as coordinated procedures and are organized around learned expectations about other agents.It is a novel approach for individuals to learn procedures as a means for the group to coordinate more efficiently. Empirical results validate the utility of this approach. Whether or not the agents have initial expertise in solving coordination problems, the distributed learning of the individual agents significantly improves the overall performance of the community, including reducing planning and communication costs.

DiamondHelp: a new interaction design for networked home appliances

Ordinary people already have great difficulty using the advanced features of digitally enhanced household products, and the problem is getting worse as more features are continually being added. This usability problem cannot be solved using only the tiny displays and limited control buttons typically found on home appliances. By using a home network to share a larger and more powerful display, we can provide a new type of collaborative interface in which the product actively helps the user, especially with complex features that are only occasionally used. In this design competition prospectus, we concentrate on the key design principles underlying DiamondHelp. The generic DiamondHelp architecture has been implemented in Java; a prototype live demonstration similar to one of the animated simulations is currently under development.

Collaborative help for networked home products

Ordinary people already have great difficulty using the advanced features of digitally-operated household devices, and the problem is getting worse as more customization and programming features are continually being added. This usability problem cannot be solved using only the tiny displays and limited control buttons typically found on home appliances. We demonstrate how, using home networking to share a larger and more powerful display, we can provide home products with a new type of collaborative interface in which the product actively helps the user, especially with complex features that are only occasionally used.