Alexandra M. Coddington

A Motivation-based Planning and Execution Framework

AI planning systems tend to be disembodied and are not situated within the environment for which plans are generated, thus losing information concerning the interaction between the system and its environment. This paper argues that such information may potentially be valuable in constraining plan formulation, and presents both an agent- and domainindependent architecture that extends the classical AI planning framework to take into account context, or the interaction between an autonomous situated planning agent and its environment. The paper describes how context constrains the goals an agent might generate, enables those goals to be prioritised, and constrains plan selection.

Many hands make light work? An investigation into behaviourally controlled co-operant autonomous mobile robots

The past ten years has seen a urry of research activity into the behavioural control of autonomous mobile robots. Yet despite this effort, many researchers are of the opinion that behavioural robots are incapable of achieving tasks more complex than simple can collecting, box pushing, herding or moving in formation. If such robots are to gain industrial credibility, these criticisms must be addressed. To focus the research we have studied the application of multiple mobile robots to a complex nuclear plant decommissioning problem. We argue that it is possible for multiple mobile robots to co-operatively perform a complex task provided that solutions to a number of key issues are incorporated into a behavioural control architecture. These include: behaviour con ict resolution, behaviour adaptation and behaviour scheduling. We have designed behavioural control methods to address these issues and our work has resulted in the creation of a behaviour synthesis architecture (BSA) which has been implemented on two real mobile robots. The application of the BSA to our complex industrial task is detailed and the results from the work are presented.

Integrating motivations with planning

This paper presents two models of goal generation which enable a motivated autonomous agent to generate goals in response to changes in its underlying drives or motivations, while it is both planning and executing. A Mars rover domain is used to illustrate the two models: the first model involves goals being generated explicitly in response to changes to the agents motivations where such goals are then provided to a planner, while the second model involves encoding the motivations and the goals that they may generate as part of the planners domain model. Results from experiments on integrating the models with different planners suggest that while they may bring the benefits of autonomy that we seek, they also introduce more complexity into the planning problem.

A continuous planning framework with durative actions

This paper describes a continuous planning framework to be used by a planning agent situated within an environment, which reasons about goals with priorities and deadlines, and actions with duration. The framework assumes that goals may be generated continuously, which requires the interleaving of planning and execution. Constraints upon time may mean it is not possible for all goals to be achieved-as a consequence the planning agent must be able to prioritise its goals. A crucial component of this framework is a temporal manager which enables the planner to reason about whether or not there is sufficient time available to achieve all goals, and to calculate deadlines for actions and outstanding subgoals. The main contribution of this paper is an examination of the way in which the partial order planning paradigm could be extended to reason with PDDL2.1 level 3 durative actions.