We describe the ucpop partial order planning algorithm which handles a subset of Pednault's ADL action representation. In particular, ucpop operates with actions that have conditional effects, universally quantified preconditions and effects, and with universally quantified goals. We prove ucpop is both sound and complete for this representation and describe a practical implementation that succeeds on all of Pednault's and McDermott's examples, including the infamous "Yale Stacking Problem" [McDermott 1991].

Many real world applications require systems with both reasoning and sensing/acting capabilities. However, most often, these systems do not work properly, i.e. they fail to execute actions and rarely perceive the external world correctly. No action, even if apparently simple, is guaranteed to succeed and, therefore, no planning can be "sound" (with respect to the real world) without taking into account failure. In this paper, we present a theory of planning that provides (1) a language that allows us to express failure; (2) a declarative formal semantics for this language; (3) a logic for reasoning about (possibly failing) plans. 1 Failure Many real world applications require systems with both reasoning and sensing/acting capabilities. Reasoning allows systems to achieve high level goals. Acting and sensing allows them to work in a complex and unpredictable external environment. Most often, systems sensing and acting in the world do not work properly, i.e. they fail to exe...

l purpose resolution theorem prover, several extensions and modifications have been developed. They led to efficiency gains and increased the expressive power of the underlying logic; most prominent examples being Kowalski's version of the situational calculus [21] and Manna and Waldinger's fluent theory [22], respectively. Kowalski's approach provides a representational solution of the frame problem by drastically reducing the number of required frame axioms. Plans are generated using a Horn clause resolution procedure guided by several strategies. Fluent theory was the first approach which allowed for a domain axiomatisation using flexible function symbols and with that formed the basis for the generation of recursive plans. These were produced by Manna and Waldinger's resolution-based tableau calculus which also provides induction rules. The paradigm, which underlies these approaches as well as most of the more recent ones, is that of

Recent research in planning is more and more focusing on planning systems working in "real world" domains. These systems need to act in, sense and represent the real world. Furthermore, no action, even if apparently simple, is guaranteed to succeed and, therefore, no planning can be "sound" (with respect to the real world) without taking into account possible failures. This is mainly due to the intrinsic complexity of reality. A planning language is therefore required to represent explicitly failures, sensing tasks, planning tasks, and task combinations. In this paper, we propose a planning language (called L) which addresses the above features. L allows representing the basic planning activities, the control structures and the basic operations to deal with failures. As a consequence, a uniform representation is used to describe both acting/sensing in the external world and basic planning activities. In this paper, we give the syntax and the semantics of L. Furthermore we also give some examples from an application (the project MAIA) which uses L as planning language. 2 1

The interleaving of planning with execution is a basic approach to planning with incomplete information. This paper investigates the use of this approach in stationary environments. We first define a computational framework where this approach can be investigated. We then show how the interleaving of planning with execution can be used for reducing planning with incomplete information to planning with complete information and constraint satisfaction. In addition, we show a restriction on our model where this reduction yields a polynomial procedure for planning with incomplete information. Finally, we discuss belief-based conditional planning; an approach which bridges part of the gap between planning while executing and general conditional planning. 1 Introduction Consider a robot in a maze. The robot is initially located in the entrance point of the maze, while its objective is to reach the exit point. The robot can move along the walls of the maze and can assume its movements won'...