Showing papers by "Malik Ghallab published in 2002"

Learning how to combine sensory-motor modalities for a robust behavior

Abstract: We are proposing here an approach and a system, called ROBEL, that enables a designer to specify and build a robot supervision system which learns from experience very robust ways of performing a task such as navigate to. The designer specifies a collection of Hierarchical Tasks Networks (HTN) that are complex plans, called modalities, whose primitives are sensory-motor functions. Each modality is a possible combination these functions for achieving the task. The relationship between supervision states and the appropriate modality for pursuing a task is learned through experience as a Markov Decision Process (MDP) which provides a general policy for the task. This MDP is independent of the environment; it characterizes the robot abilities for the task.