AN IMPLEMENTATION EFFICIENT LEARNING ALGORITHM FOR ADAPTIVE-CONTROL USING ASSOCIATIVE CONTENT-ADDRESSABLE MEMORY

Three modifications to the Boxes-ASE/ACE reinforcement learning improv es implementation efficiency and performance. A state history queue (S HQ) eliminates computations for temporally insignificant states. A dyn amic link table only allocates control memory to states the system tra verses. CMAC state association uses previous learning to decrease trai ning time. Simulations show a 4-fold improvement in learning. The SHQ in a hardware implementation of the pole-cart balancer reduces computa tion time 11-fold.