A genetic based reinforcement neurocontroller for dual arm planar robot

Dual arm manipulators are becoming widely used in industrial applications, however, their control is much more complicated than their single arm counterparts. In this paper we present a novel genetic algorithm based reinforced neurocontroller for the dual arm system. The new controller does not require any knowledge about the dynamics of the system and can be trained offline. A genetic algorithm search and optimize for the best controller in a population of potential neurocontrollers for the system. The simulation results showed an outstanding performance of the neurocontroller over other conventional control methods, the neurocontroller was robust and able to reject noises and disturbances in the measured output variable. The controller also offers great range of flexibility to system parameter changes as it does not depend on the system dynamics.