David E. Wilkins

Planning and reacting in uncertain and dynamic environments

Abstract Agents situated in dynamic and uncertain environments require several capabilities for successful operation. Such agents must monitor the world and respond appropriately to important events. The agents should be able to accept goals, synthesize complex plans for achieving those goals, and execute the plans while continuing to be responsive to changes in the world. As events render some current activities obsolete, the agents should be able to modify their plans while continuing activities unaffected by those events. The Cypress system is a domain-independent framework for defining persistent agents with this full range of behaviour. Cypress has been used for several demanding applications, including military operations, real-time tracking, and fault diagnosis.

Can AI planners solve practical problems

While there has been recent interest in research on planning and reasoning about actions, nearly all research results have been theoretical. We know of no previous examples of a planning system that has made a significant impact on a problem of practical importance. One of the primary goals during the development of the SIPE‐2 planning system has been the balancing of efficiency with expressiveness and flexibility. With a major new extension, SIPE‐2 has begun to address practical problems. This paper describes this new extension and the new applications of the planner. One of these applications is the problem of producing products from raw materials on process lines under production and resource constraints. This is a problem of commercial importance and SiPE‐2s application to it is described in some detail.

A Call for Knowledge- Based Planning

� 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 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 planning to minimal-knowledge methods such as STRIPS-based planners and disjunctive planners. 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 article, we define the terms knowledge-based and primitive-action planning and argue for the use of knowledge-based 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 knowledge-based 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.

A Common Knowledge Representation for Plan Generation and Reactive Execution

Abstract : This paper describes the ACT formalism, which is designed to encode the knowledge required to support both the generation of complex plans and reactive execution of those plans in dynamic environments. ACT is a heuristically adequate representation that is useful in practical applications. It serves as an interlingua for Artificial Intelligence technologies in planning and reactive control. The design of the formalism is discussed and its use in practical applications is demonstrated. These applications show that the ACT representational constructs have reasonable computational properties and also are adequately expressive.

Recovering from execution errors in SIPE

In real‐world domains (e.g., a mobile robot environment), things do not always proceed as planned, so it is important to develop better execution‐monitoring techniques and replanning capabilities. This paper describes these capabilities in the SIPE (System for Interactive Planning and Execution Monitoring) planning system. The motivation behind SIPE is to place enough limitations on the representation so that planning can be done efficiently, while retaining sufficient power to still be useful. This work assumes that new information given to the execution monitor is in the form of predicates, thus avoiding the difficult problem of how to generate these predicates from information provided by sensors.

Interactive execution monitoring of agent teams

There is an increasing need for automated support for humans monitoring the activity of distributed teams of cooperating agents, both human and machine. We characterize the domain-independent challenges posed by this problem, and describe how properties of domains influence the challenges and their solutions. We will concentrate on dynamic, data-rich domains where humans are ultimately responsible for team behavior. Thus, the automated aid should interactively support effective and timely decision making by the human. We present a domain-independent categorization of the types of alerts a plan-based monitoring system might issue to a user, where each type generally requires different monitoring techniques. We describe a monitoring framework for integrating many domain-specific and task-specific monitoring techniques and then using the concept of value of an alert to avoid operator overload. We use this framework to describe an execution monitoring approach we have used to implement Execution Assistants (EAs) in two different dynamic, data-rich, real-world domains to assist a human in monitoring team behavior. One domain (Army small unit operations) has hundreds of mobile, geographically distributed agents, a combination of humans, robots, and vehicles. The other domain (teams of unmanned ground and air vehicles) has a handful of cooperating robots. Both domains involve unpredictable adversaries in the vicinity. Our approach customizes monitoring behavior for each specific task, plan, and situation, as well as for user preferences. Our EAs alert the human controller when reported events threaten plan execution or physically threaten team members. Alerts were generated in a timely manner without inundating the user with too many alerts (less than 10% of alerts are unwanted, as judged by domain experts).

The Grasper-CL graph management system

Grasper-CL is a system for manipulating and displaying graphs, and for building graph-based user interfaces for application programs. It is implemented in COMMON LISP and CLIM, and has been proven by use in a number of applications. Grasper-CL includes several advances in graph drawing. It contains a graph abstract datatype plus a comprehensive and novel language of operations on that datatype. The appearance of Grasper-CL graphs can be tailored by a wide variety of shape parameters that allow the application to customize the display of nodes and edges for different domains. Default values for shape parameters can be established at several levels. Grasper-CL employs a toolbox approach to graph layout: the system contains a suite of graph layout algorithms that can be applied individually, or in combination to produce hierarchical graph layouts. The system also contains an interactive graph browser.

Airlift mission monitoring and dynamic rescheduling

We describe the Flight Manager Assistant (FMA), a prototype system, designed to support real-time management of airlift operations at the USAF Air Mobility Command (AMC). In current practice, AMC flight managers are assigned to manage individual air missions. They tend to be overburdened with associated data monitoring and constraint checking, and generally react to detected problems in a local, myopic fashion. Consequently, decisions taken for one mission can often have deleterious effects on others. FMA combines two key capabilities for overcoming these problems: (1) intelligent monitoring of incoming information (for example, weather, airport operations, aircraft status) and recognition of those situations that require corrective action and (2) dynamic rescheduling of missions in response to detected problems, both to understand the global implications of changed circumstances and to determine appropriate rescheduling actions. FMA builds on two existing technologies: an execution-monitoring framework previously applied to small-unit operations and control of robots, and a dynamic scheduling tool that is transitioning into operational use in AMCs Tanker/Airlift Control Center. FMAs dynamic-mediation module provides for collaborative mission management by different planning and execution offices by structuring communication for decision making.

Reasoning about Locations in Theory and Practice

Locational reasoning plays an important role in many applications of AI problem‐solving systems, yet has remained a relatively unexplored area of research. This paper addresses both theoretical and practical issues relevant to reasoning about locations. We define several theories of location designed for use in various settings, along with a sound and complete belief revision calculus for each that maintains a STRIPS‐style database of locational facts. Techniques for the efficient operationalization of the belief revision rules in planning frameworks are presented. These techniques were developed during application of the location theories to several large‐scale planning tasks within the Sipe planning framework.

Casual reasoning in planning

Reasoning about actions necessarily involves tracking the truth of assertions about the world over time. The SIPE planning system retains the efficiency of the STRIPS assumption for this while enhancing expressive power by allowing the specification of a causl theory. Separation of knowledge about causality from knowledge about actions relieves operators of much of their representational burden and allows them to be applicable in a wide range of contexts. The implementation of causal theories is described, together with examples and evaluations of the systems expressive power and efficiency.