Reasoning across scenarios in planning under uncertainty

Planning under uncertainty requires the adoption of assumptions about the current and future states of the world, and the preparation of conditional plans based on these assumptions. In any realistic domain, however, there will be an exponential explosion in the number of conditional plans required. One approach to this problem is to articulate a set of scenarios, which together are representative of the possible and/or likely futures. Doing this then creates the challenge of reasoning across the scenarios to decide a course of action. We present an argumentation-based formalism for representing different assumptions in a scenario framework and for reasoning across the resulting scenarios.