Conformant planning is the problem of nding a sequence of actions that is guaranteed to achieve the goal for any possible initial state and nondeterministic behavior of the planning domain. In this paper we present a new approach to conformant planning. We propose an algorithm that returns the set of all conformant plans of minimal length if the problem admits a solution, otherwise it returns with failure. Our work is based on the planning via model checking paradigm, and relies on symbolic techniques such as Binary Decision Diagrams to compactly represent and eciently analyze the planning domain. The algorithm, called cmbp, has been implemented in the mbp planner. cmbp is strictly more expressive than the state of the art conformant planner cgp. Furthermore, an experimental evaluation suggests that cmbp is able to deal with uncertainties more eciently than cgp.

uncertainty. However, this work generally assumes an extremely simple model of action that does not consider continuous time and resources. These assumptions are not reasonable for a Mars rover, which must cope with uncertainty about the duration of tasks, the energy required, the data storage necessary, and its current position and orientation.

For a given problem, the optimal Markov policy over a finite horizon is a conditional plan containing a potentially large number of branches. However, there are applications where it is desirable to strictly limit the number of decision points and branches in a plan. This raises the question of how one goes about finding optimal plans containing only a limited number of branches. In this paper, we present an any-time algorithm for optimal k-contingency planning. It is the first optimal algorithm for limited contingency planning that is not an explicit enumeration of possible contingent plans. By modelling the problem as a partially observable Markov decision process, it implements the Bellman optimality principle and prunes the solution space. We present experimental results of applying this algorithm to some simple test cases.

this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Advanced Research Projects Agency or the U.S. Government

Conformant planning is the problem of finding a sequence of actions that is guaranteed to achieve the goal for any possible initial state and nondeterministic behavior of the planning domain. In this paper we present a new algorithm for conformant planning, returning conformant plans of minimal length if the problem admits a solution, otherwise returning with failure. This work extends previous work on conformant planning, based on model checking techniques. The algorithm presented in this paper takes full advantage of the symbolic representation based on Binary Decision Diagrams (BDDs). In particular, it proceeds forwards, from the initial states towards the goal. It is thus able to exploit forward computation steps, which some times can be significantly more efficient than the corresponding backwards. Furthermore, it implements a new simplification technique which also significantly improves the performance, by containing the combinatorial explosion of the number of explored plans due to the use of a serial encoding. Experimental results suggest that our conformant planner CMBP, besides being strictly more expressive than the state of the art conformant planner CGP, is also able to deal with uncertainties more efficiently.