Abstract (EN)

Choquet expected utility (CEU) is one ofthe most sophisticated decision criteria usedin decision theory under uncertainty. Itprovides a generalisation of expected utilityenhancing both descriptive and prescriptivepossibilities. In this paper, we investigatethe use of CEU for path-planning under uncertainty with a special focus on robust solutions. We first recall the main features ofthe CEU model and introduce some examples showing its descriptive potential. Thenwe focus on the search for Choquet-optimalpaths in multivalued implicit graphs wherecosts depend on different scenarios. Afterdiscussing complexity issues, we propose twodifferent heuristic search algorithms to solvethe problem. Finally, numerical experimentsare reported, showing the practical efficiencyof the proposed algorithms.