Spiking neural network for controlling the artificial muscles of a humanoid robotic arm

One of the main functions of the human nervous system is the muscles control. The complexity of this function increases for hand and fingers control because of the high diversity and accuracy of the possible motions. Starting from the control mechanisms of the natural muscle we developed a structure of spiking neurons implemented in PCB hardware that is able to drive the elbow of a human-like robotic arm. In order to increase the biological plausibility of the designed robotic arm driving system, the artificial neural network controls artificial muscles that are implemented with Flexinol actuator wires. From our knowledge the control of the actuator wires using spiking neural networks was not performed previously. The results show that the excitatory spiking neurons are able to flex and straighten the artificial elbow by stimulating two antagonistic actuator wires. Moreover, by using inhibitory neurons that modulate the neural excitatory activity the arm mobile segment can be driven to specific positions. Our final goal is the development of a biologically plausible neural structure that is able to easily drive the hand and fingers of a human-like robotic arm by controlling the artificial muscles implemented with actuator wires. This will give important clues for future developments of spiking neural networks that drive robotic arms for performing complex tasks.