A Conserved Neural Circuit-Based Architecture for Ambulatory and Undulatory Biomimetic Robots

摘要: The adaptive capabilities of underwater organisms result from layered exteroceptive reflexes responding to gravity, impediment, and hydrodynamic optical flow. In combination with taxic responses point sources sound or chemicals, these allow reactive autonomy in the most challenging environments. We are developing a new generation lobster lamprey-based robots that operate under control by synaptic networks rather than algorithms. networks, based on command neuron, coordinating central pattern generator architecture, code sensor input as labeled lines activate shape memory alloybased artificial muscles through simple interface couples excitation contraction. have completed robot adapting this sensor, board,andactuatorarchitecturetoanewgenerationofthelobster-basedrobot.The networksareconstructed fromdiscrete timemap-basedneuronsand synapsesand instantiated digital signal processing chip. A board integrates inputs short baseline sonar array (for beacon tracking supervisory control), accelerometer, compass, antennae, optionally chemosensors. Actuator is mediated pulse-width duty cycle coding generated electronic motor neurons comparator power field-effect transistor (FET) system housed low- high-current driver boards. These circular boards stacked tubular hull processor batteries. This can readily mimic