CLG Contingent planner


Albore, A., Palacios, H., Geffner, H.: Fast and Informed Action Selection for Planning with Sensing. In: Borrajo, D., Castillo, L., and Corchado, J.Manuel Current Topics in Artificial Intelligence, 12th Conference of the Spanish Association for Artificial Intelligence (CAEPIA). Selected Papers. Springer, Salamanca, Spain (2007). Paper CAEPIA 2007 (144.71 KB)
Albore, A., Palacios, H., Geffner, H.: A Translation-based approach to Contingent Planning. Proc. of Int. Joint Conf. on Artificial Intelligence (IJCAI-09). , Pasadena, USA (2009). Paper IJCAI (169.69 KB)

 A contingent planning problem P, which is a  non-deterministic search problem in belief space, is compiled into a non-deterministic problem X(P) in state space, whose literals represent beliefs over P. The planner CLG (Closed-Loop Greedy Planner) accepts a contingent planning problem P with non-deterministic actions, and solves it by using the translation X(P) for keeping track of the beliefs, and a suitable strenghtening relaxation X+(P) for selecting the actions to apply next.