BELBIC and Its Industrial Applications: Towards Embedded Neuroemotional Control Codesign
The past few years have witnessed a proliferation of industrial and decision making applications of a novel neurocontroller designated as BELBIC: Brain Emotional Learning Based Intelligent Controller. Based on a proposed open loop descriptive model of the midbrain, where emotional processing is understood to mainly take place, its utilization has been motivated by the belief that most human decisions are made using bounded rationality. Successful control engineering and decision making applications are reviewed, where BEL has been used for satisficing action selection based on artificial emotions. Laboratory and Industrial scale applications are emphasized. Recent results on stability and performance guarantees are also examined. Finally flexible bioinspired SOC and other hardware/software implementations of BELBIC are investigated. It is argued that in order for VLSI implementations of neural networks to be commercially viable, it is crucial to minimize the redesign expenses for their optimal dedicated implementations for any given application on any desired platform. Furthermore, on line applications require recursive learning that need not precede the recall mode. High levels of adaptability, disturbance rejection, and fault tolerance are other important characteristics of the proposed IC.
KeywordsDecision and control neural network artificial emotion model driven architecture SOC implementation
Unable to display preview. Download preview PDF.