This paper describes an implementation of the 3T robot architecture which has been under development for the last eightyears. The architecture uses three levels of abstraction and description languages whichare compatible between levels. The makeup of the architecture helps to coordinate planful activities with real-time behaviors for dealing with dynamic environments. In recent years, other architectures have been created with similar attributes but two features distinguish the 3T architecture: 1) a variety of useful software tools have been created to help implement this architecture on multiple real robots;, and 2) this architecture, or parts of it, have been implemented on a varietyofvery different robot systems using different processors, operating systems, effectors and sensor suites.

This paper describes the New Millennium Remote Agent #NMRA# architecture for autonomous spacecraft control systems. This architecture integrates traditional real-time monitoring and control with constraintbased planning and scheduling, robust multi-threaded execution, and model-based diagnosis and recon#guration.

All intelligence relies on search --- for example, the search for an intelligent agent's next action. Search is only likely to succeed in resource-bounded agents if they have already been biased towards finding the right answer. In artificial agents, the primary source of bias is engineering. This dissertation

Abstract not found

The Simple Temporal Network formalism permits significant flexibility in specifying the occurrence time of events in temporal plans. However, to retain this flexibility during execution, there is a need to propagate the actual execution times of past events so that the occurrence windows of future events are adjusted appropriately. Unfortunately, this may run afoul of tight real-time control requirements that dictate extreme efficiency. The performance may be improved by restricting the propagation. However, a fast, locally propagating, execution controller may incorrectly execute a consistent plan. To resolve this dilemma, we identify a class of dispatchable networks that are guaranteed to execute correctly under local propagation. We show that every consistent temporal plan can be reformulated as an equivalent dispatchable network, and we present an algorithm that constructs such a network. Moreover, the constructed network is shown to have a minimum number of edges among all such n...

This paper has been accepted to the Journal of Logic and Computation, and should

Effective task-level control is critical for robots that are to engage in purposeful activity in realworld environments. This paper describes PRSLite, a task-level controller grounded in a procedural knowledge approach to action description. The controller embodies much of the philosophy that underlies the Procedural Reasoning System (PRS) but in a minimalist fashion. Several features of PRS-Lite distinguish it from its predecessor, including a richer goal semantics and a generalized control regime. Both of these features are critical for supporting the management of continuous processes employed in current-generation robots. PRS-Lite has been used extensively as a task-level controller for a robot whose underlying behaviors are implemented as fuzzy rules. Tasks to which it has been applied include vision-based tracking, autonomous exploration, and complex delivery scenarios. Introduction In recent years, there has been a convergence of design methodologies for certain aspects of rob...

The goal of the Atlas project is to increase the opportunities for students to construct their own knowledge by conversing (in typed form) with a natural language-based ITS. Our previous research (Freedman et al., 2000; Freedman, 2000; Ros'e, 2000) has produced reusable components and tools for facilitating the development of domain specific tutorial dialogue systems. The domain independent Atlas system (Freedman et al., 2000) provides a general purpose planning engine and robust input understanding component that can be used to augment any tutoring system with dialogue capabilities. Natural language dialogue, which involves language understanding, planning, and language generation, offers a number of attractive features for intelligent tutoring systems. First, providing the opportunity for natural language input allows the system to assess student understanding based on the direct evidence of the content of student explanations rather than based on the indirect evidence of the problem solving mistakes the students make. Natural language dialogue provides a context in which the system can tailor its presentation of material more directly to the students' needs, for example by addressing student misconceptions immediately as they arise in conversation. Secondly, it gives students the opportunity to gain experience using the language of the domain they are learning. In contrast to using short menu interfaces, requiring students to type an answer triggers recall memory as opposed to recognition memory. Natural language dialogue makes it possible to build a more sophisticated type of tutor. In addition to the hints that many current tutors give, we can extend the tutor's repertoire to include the types of remediation subdialogues seen in natural human tutor...

We are interested in solving real-world planning problems and, to that end, argue for the use of domain knowledge in planning. We believe that the field must develop methods capable of using rich knowledge models in order to make planning tools useful for complex problems. We discuss the suitability of current planning paradigms for solving these problems. In particular, we compare knowledge-rich approaches such as hierarchical task network (HTN) planning to minimal-knowledge methods such as STRIPS-based planners and disjunctive planners (DPs). We argue that the former methods have advantages such as scalability, expressiveness, continuous plan modification during execution, and the ability to interact with humans. However, these planners also have limitations, such as requiring complete domain models and failing to model uncertainty, that often make them inadequate for real-world problems. In this paper, we define the terms knowledge-based (KB) and primitive-action (PA) planning, and argue for the use of KB planning as a paradigm for solving real-world problems. We next summarize some of the characteristics of real-world problems that we are interested in addressing. Several current real-world planning applications are described, focusing on the ways in which knowledge is brought to bear on the planning problem. We describe some existing KB approaches, and then discuss additional capabilities, beyond those available in existing systems, that are needed. Finally, we draw an analogy from the current focus of the planning community on disjunctive planners to the experiences of the machine learning community over the past decade.

this paper, we adopt Newell's perspective [51] of a "knowledge level"' analysis rather than addressing this topic at the "symbol level". Practical plan management requires a "knowledge rich" model [69] that facilitates efficient reasoning given the demands of the surrounding environment. According to our medical interest, approaches dealing with time handling, context, and incomplete information about the world's states and the effects of actions (dynamically changing environments) are most important.