Real-Time Interface Board for Closed-Loop Robotic Tasks on the SpiNNaker Neural Computing System

摘要: Various custom hardware solutions for simulation of neural circuitry have recently been developed, each focusing on particular aspects such as low power operation, high computation speed, or biologically detailed simulations. The SpiNNaker computing system has developed to simulate large spiking circuits in real-time a network parallel operating microcontrollers, interconnected by high-speed asynchronous interface. A potential application area is autonomous mobile robotics, which would tremendously benefit from on-board simulations networks tens thousands neurons real-time. Currently, the circuit boards provide single Ethernet interface booting, debug, and input output data, results severe bottleneck sensory perception motor control signals. This paper describes small flexible I/O-hardware connect external devices robotic sensors actuators directly fast internal communication infrastructure system. We evaluate performance terms package throughput present simple demonstration closed loop robot interpreting visual data approach most salient stimulus.