← All research spotlights

Research spotlight · Unit 2

Malte Helmert

University of Basel

Research field
Automated planning, heuristic search

Malte Helmert works on automated planning, the part of artificial intelligence in which a program finds a sequence of actions that turns a described situation into a described goal. His planning system Fast Downward, presented in a 2006 paper, translates a task into variables with several values, reads off which variables influence which, and computes its own heuristic from that structure, so that the hard part of A* and its relatives, the estimate of the remaining cost, is derived from the problem by the program itself. Fast Downward won the classical track of the International Planning Competition in 2004 and became the base of many later planners. The unit presents it where it explains that the heuristic is the hard part of informed search.

In the masterclass

Section 4, after the worked examples on A* and the 8-puzzle, shows how Fast Downward computes its estimate of the remaining cost from the structure of a planning task.

The research is presented in Unit 2: Search and Planning, in a research spotlight box that explains the result at the level of the unit and links it to what you have just learned.

The paper

Helmert, M. (2006). The Fast Downward Planning System. Journal of Artificial Intelligence Research, 26, 191‑246. https://doi.org/10.1613/jair.1705

The scientists named here are not contributors to the AI Masterclass, are not affiliated with it, and have not endorsed it. The program presents and explains their published research, with references to the original papers.