CTP: A new constraint-based formalism for conditional, temporal planning (original) (raw)

A temporal planning system for time-optimal planning

Progress in Artificial Intelligence, 2001

Dealing with temporality on actions presents an important challenge to AI planning. Unlike Graphplan-based planners which alternate levels of propositions and actions in a regular way, introducing temporality on actions unbalance this symmetry. This paper presents TPSYS, a Temporal Planning SYStem, which arises as an attempt to combine the ideas of Graphplan and TGP to solve temporal planning problems more efficiently. TPSYS is based on a three-stage process. The first stage, a preprocessing stage, facilitates the management of constraints on duration of actions. The second stage expands a temporal graph and obtains the set of temporal levels at which propositions and actions appear. The third stage, the plan extraction, obtains the plan of minimal duration by finding a flow of actions through the temporal graph. The experiments show the utility of our system for dealing with temporal planning problems.

Planning with Problems Requiring Temporal Coordination

2008

We present the first planner capable of reasoning with both the full semantics of PDDL2.1 (level 3) temporal planning and with numeric resources. Our planner, CRIKEY3, employs heuristic forward search, using the start-and-end semantics of PDDL2.1 to manage temporal actions. The planning phase is interleaved with a scheduling phase, using a Simple Temporal Network, in order to ensure that temporal constraints are met. To guide search, we introduce a new temporal variant of the Relaxed Planning Graph heuristic that is capable of reasoning with the features of this class of domains, along with the Timed Initial Literals of PDDL2.2. CRIKEY3 extends the state-of-the-art in handling the full temporal expressive power of PDDL2.1, including numeric temporal domains.

Constraint Models for Sequential Planning

Lecture Notes in Computer Science, 2009

Constraint satisfaction techniques are commonly used for solving scheduling problems, still they are rare in AI planning. Although there are several attempts to apply constraint satisfaction for solving AI planning problems, these techniques never became predominant in planning; and they never reached the success of, for example, SATbased planners. In this paper we argue that existing constraint models for classical AI planning are not fully using the power of constraint satisfaction; thus we propose a reformulation, which significantly improves their efficiency.

Temporal Planning while the Clock Ticks

2018

One of the original motivations for domain-independent planning was to generate plans that would then be executed in the environment. However, most existing planners ignore the passage of time during planning. While this can work well when absolute time does not play a role, this approach can lead to plans failing when there are external timing constraints, such as deadlines. In this paper, we describe a new approach for time-sensitive temporal planning. Our planner is aware of the fact that plan execution will start only once planning finishes, and incorporates this information into its decision making, in order to focus the search on branches that are more likely to lead to plans that will be feasible when the planner finishes.

Using Temporal Knowledge in a Constraint-Based Planner

Incorporating domain-specic temporal knowledge into a constraint-based planner can signican tly decrease the computation time. This paper describes a method for compiling and embedding user- specied temporal information into the constraint problem translation of a planning problem. We provide empirical results for our framework implemented within the Blackbox (9) planning system; the data indicate that using temporal domain information accelerates planning. A classical planning problem is usually specied as sets of objects, actions, propositional pred- icates to describe the relationships of the objects in the world, a description of the initial state of the world, and a description of the desired nal state. The actions and predicates make up the planning domain, while a specic set of objects, an initial state, and a nal state constitute a particular planning problem. Actions act on objects; when taken, an action may change what is true about the world through its ee cts. Further, a...

Fast temporal planning in a Graphplan framework

AIPS Workshop on Planning in Temporal Domains, 2002

Graphplan (Blum & Furst 1995) has been successfully extended to plan with actions with durations (Smith & Weld 1999)(Garrido, Onaindıa, & Barber 2001; Garrido, Fox, & Long 2001). Existing approaches treat durative actions as spanning several layers in the plan graph, with fact layers corresponding to points in the flow of time. A simple model of time is used which prohibits much of the concurrency available for exploitation in an interesting problem. In this paper we describe an alternative approach, in which the fact layers of a ...

Planning with Inaccurate Temporal Rules

2012 IEEE 24th International Conference on Tools with Artificial Intelligence, 2012

We use a temporal pattern model called Temporal Interval Tree Associative Rules (Tita rules). This pattern model has been introduced in a previous work. The model can express uncertainty, temporal inaccuracy, the usual time point operators, synchronicity, incomplete orders, chaining, disjunctive time constraints and temporal negation. This pattern model is initially designed to be used for temporal learning. In this paper, we use Tita rules as world description models for a Planning and Scheduling task. We present an efficient temporal planning algorithm able to deal with uncertainty, temporal inaccuracy, discontinuous (or disjunctive) time constraints and predictable but imprecisely time located exogenous events. We evaluate our technique by joining a learning algorithm and our planning algorithm into a simple reactive cognitive architecture that we apply on with virtual robot.

Constraint-based temporal planning

1988

This paper deals with the application of notions from "planning" and the "representation of temporal information" in an animation system to simulate human task performance. Specifically, a model was developed in which the representation and manipulation of temporal information forms the basis of a planning system. This model has been implemented as part of an animation system in which the goal was to enable a natural, clear language for the specification of relevant temporal constraints along with a planner that would manipulate these constraints, ultimately resulting in exact timing parameters to be passed to the low-level animation routines. The first part of the paper describes the essential role of temporal planning in an animation system that models human task performance. The second part of the paper goes on to explain how a wide variety of temporal constraints can be "compiled" into a set of low-level "simple constraints" through the use of "dummy intervals" and "fuzzy constraints." The third part of the paper describes how a definite "plan" of events can be generated based on an analogical "spring system."

Temporal planning with mutual exclusion reasoning

1999

Many planning domains require a richer notion of time in which actions can overlap and have di erent durations. The key to fast performance in classical planners (e.g., Graphplan, ipp, and Blackbox) has been the use of a disjunctive representation with powerful mutual exclusion reasoning. This paper presents tgp, a new algorithm for temporal planning. tgp operates by incrementally expanding a compact planning graph representation that handles actions of di ering duration. The key to tgp performance is tight m utual exclusion reasoning which is based on an expressive language for bounding mutexes and includes mutexes between actions and propositions. Our experiments demonstrate that mutual exclusion reasoning remains valuable in a rich temporal setting.