The anthroform neural controller: An architecture for spinal circuit emulation

Existing robotic manipulator and controller designs compare unfavorably to the human arm when performing tasks in unstructured environments. Traditionally, "anthropomorphic" designs have focused on replicating only the kinematics of the hu man arm. In contrast, we describe a versatile parallel computing architecture for emulating the spinal circuits of the human nervous system in conjunction with a dynamically realistic actuated arm model. This design is based on the structural constraints of the nervous system, and consists of a special purpose digital bus which implements connections between simulated neuron pools. The processing elements are circuit cards based an the TMS 320C30 digital signal processing chip. The system may be expand ed to 256 processor cards, supporting a total of 1024 computation al nodes that are interconnected every millisecond.