PROBLEM SOLVER FOR MULTI-LAYERED STATE SPACE

摘要

PROBLEM TO BE SOLVED: To prevent repeated movement between a state S1 and a state S2 in finding a path to sub-goal or between a state S3 and a state S4 in finding a path to goal in a multi-layered state space. SOLUTION: An execution mechanism in each of layer agents 1-1, 1-2, and 1-3 of a multi-layered recognition distance learning device characterized in that it is provided with a sub-goal discrimination means 9 and a goal discrimination means 10 which discriminate whether or not the present state is equal to the state at the destination and a sub-goal holding means 11 and a goal discrimination and holding means 12 which reproduce destinations which have been acquired before.