Temporal Planning with Constraints

Foundations of Automatic Planning - The Classical Approach and Beyond

Volume

Papers from the 1993 AAAI Spring Symposium

Contents

We present ZENO, an implemented, partialorder planner that handles simultaneous actions occurring over extended time intervals. The action language supports a large subset of KRSL, including metric constraints, deadline goals, and synergistic (additive) effects. We believe the algorithm is both sound and complete. The chief technical contributions are (1) token reduction breaks complex goals into pieces that can be systematically supported, (2) incremental algorithms from operations research determine the consistency of temporal and other metric constraints, and (3) lazy evaluation and the Mean Value Theorem are exploited to cope with actions involving continuous change.