Bradley J. Clement

Top-down search for coordinating the hierarchical plans of multiple agents

Uncertain and complex environments demand that an agent be able to anticipate the actions of others in order to avoid resource conflicts with them and to realize its goals. Conflicts during plan execution can be avoided by reducing or eliminating interactions by localizing plan effects to particular agents and by merging/coordinating the individual plans of agents by introducing synchronization actions. We describe a method for coordinating plans at abstract levels that takes advantage of hierarchical representations of plan information and that retains the flexibility of plans used in robust plan execution systems such as procedural reasoning systems (PRS). In order to coordinate at abstract levels in plan hierarchies, information about how abstract plans can be refined must be available in order to identify and avoid potential conflicts. We address this by providing procedures for deriving summary information for non-primitive plans that capture the external preconditions and effects of their refinements. We also describe a general search algorithm and an implementation to show how to use this information to coordinate hierarchical plans from the top down to primitive actions.

Continual coordination through shared activities

Interacting agents that interleave planning and execution must reach consensus on their commitments to each other. In domains where agents have varying degrees of interaction and different constraints on communication and computation, agents will require different coordination protocols in order to efficiently reach consensus in real time. We briefly describe a largely unexplored class of real-time, distributed planning problems (inspired by interacting spacecraft missions), new challenges they pose, and a general approach to solving the problems. These problems involve self-interested agents that have infrequent communication but collaborate on joint activities. We describe a Shared Activity Coordination (SHAC) framework that provides a decentralized algorithm for negotiating the scheduling of shared activities over the lifetimes of multiple agents, a soft, real-time approach to reaching consensus during execution with limited communication, and a foundation for customizing protocols for negotiating planner interactions. We apply SHAC to a realistic simulation of interacting Mars missions and illustrate the simplicity of protocol development.

Coordination challenges for autonomous spacecraft

Characterizing multiagent problem spaces requires understanding what constitutes a multiagent problem or where multiagent system technologies should be applied. Here, we explore the multiagent problems involved in distributed spacecraft missions. While past flight projects involved a single spacecraft in isolation, over forty proposed future missions involve multiple coordinated spacecraft. We present characteristics of such missions in terms of properties of the phenomena being measured as well as the rationale for using multiple spacecraft. Then we describe the coordination problems associated with operating different types of missions and identify needed technologies.

Integrating multiagent coordination with reactive plan execution

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. AGENTS’01, May 28-June 1, 2001, Montreal, Quebec, Canada. Copyright 2001 ACM 1-58113-326-X/01/0005 ...15.00. * +

Scheduling high-level tasks among cooperative agents

Scheduling tasks among cooperative agents requires tradeoffs between various factors including task priorities and context-dependent execution times. We have specifically been investigating the space of functions for evaluating alternative distributed task schedules for multi-operator applications. In this paper, we describe some candidate functions and converge on intuitively appealing functions, which we show to lead to equivalent preferences over distributed schedules. We then look at the computational complexity of finding schedules that (approximately) optimize this function. When context switching costs are thrown into the mix moreover, the complexity becomes even more daunting. To address these problems, this paper summarizes our work on forging correspondences between our problems and those studied in operations research. Moreover, we have developed a new hill-climbing strategy for solving these problems, and we show that it performs well within the range of parameter settings that are representative of our application domain.

Automating Deep Space Network scheduling and conflict resolution

The Deep Space Network (DSN) is a central part of NASAs infrastructure for communicating with active space missions, from earth orbit to beyond the solar system. Consisting of more than a dozen major ground antennas at three sites spaced around the globe, it must be carefully scheduled to satisfy the requirements of the various mission users, subject to many constraints. Scheduling the communication services for these missions is a fairly large scale negotiation problem since each mission cannot give up control of its spacecraft. While we will give details of this problem, this paper focuses more on describing an agent that provides scheduling advice to a single user within a collaborative scheduling system being developed. This agent can provide guidance to the user about feasible and optimal solutions to the problem they are working on. We describe our recent work in modeling the complexities of user requirements, and then scheduling and resolving conflicts on that basis.

Using abstraction to coordinate multiple robotic spacecraft

The trend toward multiple-spacecraft missions requires autonomous teams of spacecraft to coordinate their activities when sharing limited resources. The paper describes how an iterative repair planner/scheduler can reason about the activities of multiple spacecraft at abstract levels in order to greatly improve the scheduling of their use of shared resources. By finding consistent schedules at abstract levels, refinement choices can be preserved for use in robust plan execution systems. We present an algorithm for summarizing the metric resource requirements of an abstract activity based on the resource usages of its potential refinements. We find that reasoning about this summary information and that of state constraints can offer exponential improvements in the time to find consistent schedules with an iterative repair planner. We analytically describe the conditions under which these improvements are made and show that sometimes the extra overhead involved does not warrant their use. We apply these techniques within the ASPEN planner/scheduler to a domain where a team of rovers must coordinate their schedules to avoid conflicts over shared resources. Experiments using the ASPEN planner/scheduler in a Mars multi-rover domain support our analyses and compare techniques for controlling decomposition.

Using a multicore processor for rover autonomous science

Multicore processing promises to be a critical component of future spacecraft. It provides immense increases in onboard processing power and provides an environment for directly supporting fault-tolerant computing. This paper discusses using a state-of-the-art multicore processor to efficiently perform image analysis onboard a Mars rover in support of autonomous science activities.

Distributed network scheduling

We investigate missions where communications resources are limited, requiring autonomous planning and execution. Unlike typical networks, spacecraft networks are also suited to automated planning and scheduling because many communications can be planned in advance. Because the network of spacecraft can represent multiple missions, missions will be reluctant to give up control of the spacecraft. Because communication among spacecraft is often intermittent (due to orbital and resource constraints), a spacecraft that can make scheduling decisions autonomously will be more responsive to unexpected events. Thus, a centralized planning system will not be sufficient to enable reactive communications, so we propose a distributed network scheduling system.

K: A wide spectrum language for modeling, programming and analysis

The formal methods community has over the years proposed various formally founded specification languages based on predicate logic and set theory, typically with textual notations. At the same time the model-based engineering community has proposed often less formally founded languages such as UML and SysML, typically with graphical notations. Although the graphical notations have become highly popular in industry, we argue that textual notations can be attractive in many situations. We report on an effort to provide a textual notation for SysML, realized in a language named K. K supports classes, multiple inheritance, predicate logic and set theory. K contains programming constructs, and can thus be considered as a wide-spectrum modeling and programming language. We further explain the translation of a subset of this language to the input language of the SMT-LIB standard, and the application of Z3 for analysis of the generated SMT-LIB formulas. The entire effort is part of a larger effort to develop a general purpose SysML development framework for designing systems, in support of NASAs proposed 2022 mission to Jupiters moon Europa.