Planning the project management way: Efficient planning by effective integration of causal and resource reasoning in RealPlan

Citation
B. Srivastava et al., Planning the project management way: Efficient planning by effective integration of causal and resource reasoning in RealPlan, ARTIF INTEL, 131(1-2), 2001, pp. 73-134
Citations number
55
Language
INGLESE
art.tipo
Article
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
ARTIFICIAL INTELLIGENCE
ISSN journal
0004-3702 → ACNP
Volume
131
Issue
1-2
Year of publication
2001
Pages
73 - 134
Database
ISI
SICI code
0004-3702(200109)131:1-2<73:PTPMWE>2.0.ZU;2-E
Abstract
In most real-world reasoning problems, planning and scheduling phases are l oosely coupled. For example, in project planning, the user comes up with a task list and schedules it with a scheduling tool like Microsoft Project. O ne can view automated planning in a similar way in which there is an action selection phase where actions are selected and ordered to reach the desire d goals, and a resource allocation phase where enough resources are assigne d to ensure the successful execution of the chosen actions. On the other ha nd, most existing automated planners studied in Artificial Intelligence do not exploit this loose-coupling and perform both action selection and resou rce assignment employing the same algorithm. The current work shows that th e above strategy severely curtails the scale-up potential of existing state of the art planners which can be overcome by leveraging the loose coupling . Specifically, a novel planning framework called RealPlan is developed in wh ich resource allocation is de-coupled from planning and is handled in a sep arate scheduling phase. The scheduling problem with discrete resources is r epresented as a Constraint Satisfaction Problem (CSP) problem, and the plan ner and scheduler interact either in a master-slave manner or in a peer-pee r relationship. In the former, the scheduler simply tries to assign resourc es to the abstract causal plan passed to it by the planner and returns succ ess. In the latter, a more sophisticated "multi-module dependency directed backtracking" approach is used where the failure explanation in the schedul er is translated back to the planner and serves as a nogood to direct plann er search. RealPlan not only preserves both the correctness as well as the quality (measured in length) of the plan but also improves efficiency. More over, the failure-driven learning of constraints can serve as an elegant an d effective approach for integrating planning and scheduling systems. Beyon d the context of planner efficiency, the current work can be viewed as an i mportant step towards merging planning with real-world problem solving wher e plan failure during execution can be resolved by undertaking only necessa ry resource re-allocation and not complete re-planning. (C) 2001 Elsevier S cience B.V. All rights reserved.