AltAlt: Combining Graphplan and Heuristic State Search

and Kambhampati 2000). This heuristic, along with the problem specification, and the set of ground actions in the final action level of the planning graph structure are fed to a regression state search planner. The regression planner code is adapted from HSP-R (Bonet and Geffner 1999) because it provides an optimized state search engine. The crux of controlling the regression search involves providing a heuristic function that can estimate the relative goodness of the states on the fringe of the current search tree and guide the search in most promising directions. Such heuristics can be tricky to develop. The main problem consists of taking interactions into account. Fortunately, our recent work (Nguyen and Kambhampati 2000) provides an interesting way of leveraging the GRAPHPLAN technology to generate effective heuristics. In the next section, we briefly introduce the default heuristic in ALTALT.