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

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

Controlling Underwater Robots with Electronic Nervous Systems

We are developing robot controllers based on biomimetic design principles. The goal is to realise the adaptive capabilities of the animal models in natural environments. We report feasibility studies of a hybrid architecture that instantiates a command and coordinating level with computed discrete-time map-based (DTM) neuronal networks and the central pattern generators with analogue VLSI (Very Large Scale Integration) electronic neuron (aVLSI) networks. DTM networks are realised using neurons based on a 1-D or 2-D Map with two additional parameters that define silent, spiking and bursting regimes. Electronic neurons (ENs) based on Hindmarsh–Rose (HR) dynamics can be instantiated in analogue VLSI and exhibit similar behaviour to those based on discrete components. We have constructed locomotor central pattern generators (CPGs) with aVLSI networks that can be modulated to select different behaviours on the basis of selective command input. The two technologies can be fused by interfacing the signals from the DTM circuits directly to the aVLSI CPGs. Using DTMs, we have been able to simulate complex sensory fusion for rheotaxic behaviour based on both hydrodynamic and optical flow senses. We will illustrate aspects of controllers for ambulatory biomimetic robots. These studies indicate that it is feasible to fabricate an electronic nervous system controller integrating both aVLSI CPGs and layered DTM exteroceptive reflexes.

Biomimetic approaches to the control of underwater walking machines

We have developed a biomimetic robot based on the American lobster. The robot is designed to achieve the performance advantages of the animal model by adopting biomechanical features and neurobiological control principles. Three types of controllers are described. The first is a state machine based on the connectivity and dynamics of the lobster central pattern generator (CPG). The state machine controls myomorphic actuators based on shape memory alloys (SMAs) and responds to environmental perturbation through sensors that employ a labelled-line code. The controller supports a library of action patterns and exteroceptive reflexes to mediate tactile navigation, obstacle negotiation and adaptation to surge. We are extending this controller to neuronal network-based models. A second type of leg CPG is based on synaptic networks of electronic neurons and has been adapted to control the SMA actuated leg. A brain is being developed using layered reflexes based on discrete time map-based neurons.