Insect and Crustacean Robot Lit Review

Early draft of a perspective article by Nicholas Szczecinski, Clarus Goldsmith, William Nourse, and Roger Quinn.

Past, present, and future of neuromorphic control of legged locomotion in insect-like robots

1. Introduction

Nature has often presented solutions to engineering problems since the time of Da Vinci (Brioist, 2020). One such engineering problem is creating legged robots with the agility and payload of animals but without the expiration or stubbornness of their biological counterparts. Despite the appeal of natural examples for engineering solutions, engineers have often lacked the neuromechanical knowledge, computation, components and fabrication processes to fully implement them. However, with advancing technologies in engineering and biology, animal-like locomotion may be more within reach than ever before. With such tools at our disposal, we advocate for a “biology first” approach to the control of locomotion, in which details in the animal’s nervous system and mechanics are implemented whenever possible. We believe that including more biological details will lead to more animal-like performance. We summarize the history of insect-inspired, hexapedal robot legged locomotion, show how closer adherence to biological principles and neural details have improved performance, and postulate what relatively recent breakthroughs have the greatest potential to push the field forward.

Readers seeking reviews on broader topics of neural control in animals beyond insects, central pattern generating networks, and other robots that walk, swim, or fly would benefit from related review and perspective articles (Central pattern generators in arthropods (Mulloney and Smarandache, 2010); central pattern generators in insect walking (Bidaye, Bockemühl and Büschges, 2017); Principles of central pattern generators (Marder and Bucher, 2001); Legged locomotion in vertebrates, invertebrates, and robots (Buschmann et al., 2015); Legged locomotion in hexapod robots (Manoonpong et al., 2021); Bioinspired control of vertebrate-inspired robots (Ijspeert, 2014); Insect-inspired control of robot navigation (Webb, 2020)).

2. Background

2.1 Foundational biological principles

Throughout the second half of the twentieth century, many organizing principles of the nervous system were discovered through experiments on invertebrate animals. Many of these phenomena supported a view of the arthropod nervous system as highly distributed, in which coordinated behaviors emerged from a combination of mechanical coupling and neural mechanisms. Invertebrate nervous systems were found to be organized as decentralized systems with central pattern generators (CPGs) and sensory feedback affected by command neurons. These insights provided the knowledge base that was used to develop biologically inspired walking robots.

2.1.1 Central pattern generators, sensory feedback, and command neurons

Central pattern generators have been found to underlie many rhythmic behaviors in the nervous systems of both invertebrates and vertebrates (Delcomyn, 1980). CPGs are neuronal units capable of producing rhythmic outputs given non-rhythmic inputs from elsewhere in the nervous system. Despite being called “central” pattern generators (“central” refers to the central nervous system), there is not one central rhythmic signal underlying complex behaviors such as walking. Instead, walking arises from the interplay of multiple rhythmic units whose relative phasing are controlled by central and peripheral influences [for reviews, see (Buschmann et al., 2015; Bidaye, Bockemühl and Büschges, 2017)].

Experiments have demonstrated the presence of CPGs in various animals. Some of the earliest work was conducted in the flight wingbeats of locusts (Wilson, 1961; Wilson and Wyman, 1965). In this work, researchers elicited coordinated neural activity closely resembling normal flight motor patterns by stimulating the nerve cord of decapitated locusts. Concurrent work in the crayfish, another invertebrate, revealed similar functionality in the deafferented networks that control the swimmerets (Hughes and Wiersma, 1960). Subsequent work in other species revealed similar rhythmic behavior from deafferented networks (e.g., the heartbeat of the medicinal leech (Thompson and Stent, 1976a, 1976b)). However, the CPG network itself has only been identified and fully described in a few animals (e.g., the stomatogastric ganglion of the spiny lobster (Mulloney and Selverston, 1974b, 1974a; Perkel and Mulloney, 1974; Selverston et al., 1976)). In subsequent decades, experiments showed that rhythmic activity could be evoked from the thoracic networks that control insect leg movements [(Ryckebusch and Laurent, 1993; Büschges, Schmitz and Bässler, 1995), for a review see (Bidaye, Bockemühl and Büschges, 2017)], as well as from spinal cords of vertebrates (Cohen and Wallen, 1980; Grillner, 1985; Delvolvé et al., 1999). See (Marder and Bucher, 2001) and (Ijspeert, 2008) for further reviews. Such networks have inspired many hexapod robot control systems, which are discussed throughout this manuscript.

In addition to demonstrating the existence of CPGs, these and other studies showed how synaptic inputs to these networks can alter the phasing of their rhythms, a necessary element for the control of rhythmic motions such as walking. One example mechanism is von Holst’s “magnet effect”, in which one CPG synapses onto another, forcing the second CPG to oscillate with the same frequency as the first with a phase lag (Holst, 1939; Ayers and Selverston, 1979). It is called the “magnet effect” because the magnitude of the phase lag is stable when perturbed, exhibiting attractive dynamics. In legged systems, many such inputs come either directly or indirectly from sensory neurons, ensuring that CPGs only change phase when a particular leg state has been reached [e.g., (Hess and Büschges, 1999)]. Coupling CPGs in this manner allows for interjoint and interleg coordination between structures (e.g., joints, legs) operating at their own frequencies with their own CPG units (Pinsker and Ayers, 1983; Grillner, 1985). The distributed nature of such systems gives them the flexibility to produce different coordination patterns in different contexts.

Perhaps the most famous examples of insect-inspired coordination rules are the “Cruse Rules” for interleg coordination during walking (Cruse, 1990). These behavioral rules do not explicitly incorporate central pattern generators, but they have inspired inter-CPG connections for neuromechanical models of walking insects (Rubeo, Szczecinski and Quinn, 2017). The Cruse Rules describe how the points at which a leg lifts off at the end of a step (i.e., its posterior extreme position [PEP]) and at which it touches down at the beginning of a step (i.e., its anterior extreme position [AEP]) shift based on the motion of other, adjacent legs. For example, a leg’s PEP is moved posterior (prolonging stance) while the adjacent posterior leg is in swing, and a leg’s AEP is moved posterior (shortening swing) when the adjacent anterior leg begins swing. Speed-dependent interleg coordination patterns observed in insects and other arthropods emerged from these and other sensory-based rules (Cruse, 1990), demonstrating the adaptability that is possible when sensory feedback is utilized in control.

Peripheral sensory feedback has also been shown to entrain rhythm for motion in a variety of insects (Pearson, 1972; Wendler, 1974; Bässler and Büschges, 1998) as well as vertebrates [review in (Keir G Pearson, 1995)]. Such entrainment can mean altering the period of oscillations, as well as resetting cycle phasing to achieve a particular output. Even though studies of motor control often focus on rhythm generation, peripheral sensory feedback is absolutely crucial to the graceful behavior of intact animals. For example, in Wilson et al.’s (1965) early CPG study, input from the locust’s stretch receptors was necessary to attain a wingstroke frequency similar to an intact animal. Adaptive motion requires that CPG networks be flexible enough to adjust their phasing due to sensory input received during behavior.

Furthermore, evidence has been found for higher level inputs that influence CPG activity. So-called “command neurons”, or interneurons whose stimulation activates, modulates, or halts a CPG’s rhythmic activity, have been found in the thoracic ganglia (or spinal cord) and other central parts of the nervous system in invertebrates and vertebrates [for reviews see (Pinsker and Ayers, 1983; Pflüger, Duch and Heidel, 2004)]. Most CPGs are not spontaneously active, so these neurons must act as intermediaries between the higher-level command centers and the CPGs to activate particular behaviors. Various hormones and neuromodulators have also been found to produce entrainment and phasic reset in CPGs (Pinsker and Ayers, 1983; Pflüger, Duch and Heidel, 2004). The discovery of such influences contributes to a hierarchical model of motor control, in which local interactions between CPGs and sensory feedback produce rhythmic motions that can be toggled on and off by descending influences (e.g., synaptic inputs from command neurons and neuromodulators). However, in order to produce multiple behaviors in different contexts, the nervous system must have mechanisms by which to reconfigure this hierarchy.

2.1.2 Context-Dependent Reflex Reversal

In addition to the characterization of CPGs, experiments in both invertebrates and vertebrates provide insight into other pathways that might alter the motor control networks described above. As reviewed in (K. G. Pearson, 1995), responses to sensory afferent activity have been found to change between standing and walking in cats, crayfish, and stick insects. Similar aptly named “reflex reversals” have also been found in other insects, such as the locust (Burrows, 1980) and cockroach (Mu and Ritzmann, 2008). Further work on stick insects characterized this phenomenon as occurring in the low-level motor control centers comprised of the sensory organs, groupings of non-spiking interneurons, and the motor neurons (Driesang and Büschges, 1996; Sauer et al., 1996). In particular, (Sauer, Büschges and Stein, 1997) hypothesized that the nervous system may accomplish reflex reversals by primary afferent depolarization or some other means of weakening the effect of sensory input. Whether this inhibition stems from the CPGs or from some other descending inputs from the brain or other higher-level centers (e.g., the subesophageal/gnathal ganglion or the brain stem) is presently unknown, but the resulting change in behaviors highlight the great flexibility of the nervous system that has inspired robot control for decades.

2.2. Early Robotic Implementations

The experiments from the previous section provided a foundation for the application of neurobiological principles to the control of legged robots. The following sections provide a condensed timeline of the development of arthropod inspired, neurobiological robot controllers over the last several decades.

2.2.1 Functional/Behavior-Based Controllers

Early efforts to apply principles of a distributed arthropod-like nervous system to robotics were successful in demonstrating their potential power. Rodney Brooks’ Genghis was a hexapod robot whose posture and locomotion was controlled by a network of Augmented Finite State Machines (AFSM) assembled according to subsumption architecture (Brooks, 1989). In this architecture, higher-level behaviors (e.g., walking) could be built atop simpler behaviors (e.g., standing), mimicking the hierarchical organization of the nervous system. Transitions between states within the AFSM could be driven by sensory feedback or by the timing of a central clock, similar to a CPG network.

At about the same time, Beer, Chiel and Sterling developed a recurrent neural network (RNN) for controlling insect walking based upon current literature (Beer, Chiel and Sterling, 1989). This network of 37 neurons (6 for each leg and a command neuron) controlled a simulated insect such that it walked in the continuum of speed-dependent insect gaits described by Wilson (Wilson, 1966). Robot I was then developed by Espenschied, Quinn, Beer, and Chiel and controlled by this (RNN) such that it too mimicked the functional outputs of insect walking (Beer et al., 1992). The RNN controller for Robot I was more distributed than that of Genghis, with patterning arising from the interaction between leg-specific networks, not set by a central clock. Sensory feedback signaled the beginning and end of stance phase, but otherwise the coordination arose from central coupling.

The Cruse Rules (Cruse, 1990) were also implemented to control Robot I in a finite state machine, in place of the RNN (Beer and Chiel, 1989; Beer, Chiel and Sterling, 1989, 1990; Chiel and Beer, 1989), resulting in the same speed dependent continuum of insect gaits (Espenschied et al., 1993). The Cruse Rules implementation was resilient against “lesions” within the network, that is, when sensory information transmission was blocked between legs. Such resiliency does not alter robot coordination under nominal conditions, but prevents the robot from being incapacitated when sensors or computers are damaged, which improves the robot’s performance overall.

The application of biological principles also aided the tuning of robot controller parameters. A subsequent functional neural controller for Robot I was evolved using a genetic algorithm (in simulation (Beer and Gallagher, 1992); and implementation in Robot I in (Gallagher et al., 1996)). The evolved networks coordinated the legs of the simulated insect and the robot both when sensory feedback was provided and when it was disabled, demonstrating their robustness against robot damage. Such features arising from simulated evolution demonstrates the usefulness of biologically-inspired algorithms for tuning network parameters for functional robot behavior.

To increase the flexibility and adaptability of Robot I’s controller, the subsequent Robot II’s posture and locomotion were enhanced by the addition of several leg-local insect-inspired reflexes, for example, a reflex to step in response to large perturbations and a reflex to search for a foothold if none was found (Espenschied et al., 1995, 1996). When superimposed, these reflexes enabled the robot to walk over extremely cluttered terrain with no prior knowledge of the obstacles. The Cruse Rules were generalized for omnidirectional walking by setting the AEP in the direction of desired motion and the PEP behind that and varied according to the Cruse Rules. Thus, Robot II could walk in any direction including crab walking as well as exhibit a continuum of insect gaits. This biology-first approach of incorporating sensory-driven reflexes into Robot II’s controller improved its mobility and adaptability relative to Robot I’s.

Around the same time, the Technical University of Munich (TUM) Walking Machine was developed based on the stick insect. The TUM Walking Machine (Pfeiffer, Eltze and Weidemann, 1995; Steuer and Pfeiffer, 1997) featured a highly distributed controller that took close inspiration from the studies of Cruse et al. regarding interleg coordination (Cruse, 1976, 1990; Cruse et al., 1995). Like Robot I and Robot II described above, the controller utilized a distributed leg coordination module that passed the proprioceptive information of each leg to the adjacent legs and altered their anterior extreme positions (AEP) and posterior extreme positions (PEP). Each leg controller possessed reflexes that enabled the system to react to obstacles on a single-leg basis with no prior knowledge. Using its biology-first controller, the robot was subsequently able to stably walk along even terrains with obstacles and unstructured surfaces.

Also inspired by the stick insect, researchers at the Forschungszentrum Informatik (FZI) began developing the LAURON series of robots (Berns, Cordes and Ilg, 1994; Gaßmann, Scholl and Berns, 2001). Initially, LAURON’s controller was comprised of a two-layer neural network, with one layer consisting of modules for the control of each leg and the second layer managing leg coordination and path planning. The specific structures of the neural networks were generated using various reinforcement learning techniques (Ilg and Berns, 1995). For the control of subsequent iterations of the robot (e.g., LAURON III), traditional neural networks were removed, but the decentralized architecture remained. LAURON III’s controller used a series of local behaviors (e.g., cyclic gait generation, ground searching, collision reaction) in combination with higher level body position control to generate foot trajectories (Gaßmann, Scholl and Berns, 2001). Stepping pattern generation was created by a series of polynomial functions relating step parameters to leg phase and modulated by proprioceptive feedback. Leg coordination was handled using a series of coordination rules based on achieving pre-calculated goal leg phases and maintaining the robot’s static stability. Although this later iteration was structured in a more biologically plausible manner than the earlier neural network implementations, it did not feature reinforcement learning, meaning the walking parameters for this controller had to be manually tuned for each specific environment.

Each of these studies showed the potential for animal-like control systems for distributed control of multi-legged robots. However, the complexity of the control networks was limited by the biological data and computational power available. Advances in both areas fueled the next generation of neural-inspired hexapod control systems.

2.2.2 Functional Neural Simulation

The computational power available during the construction of these early robots greatly limited the ability to simulate neural dynamics. As a result, analog “very large-scale integration” (VLSI) circuits were developed as an alternative to simulation of neural dynamics on a von Neumann architecture (See section 3.3.1 for more discussion). These circuits enabled rapid computation of network dynamics in a highly parallelized fashion, making them theoretically capable of implementing very large neural controllers in real time with lower power consumption. Early efforts with analog VLSI circuits showed that oscillator models could be constructed in hardware to drive robots’ oscillatory leg movements (Still and Tilden, 1998; Brown, 2000).Later, more refined versions were produced that controlled the omnidirectional stepping of a legged robot via a biological control framework in which walking direction was determined via exteroception and communicated to the legs via descending commands (Ayers et al., 2010). However, due to the difficulty of tuning network parameters and the rise of more powerful desktop computers, analog VLSIs have not yet superseded digital simulation as the most common way to implement neuromorphic control systems on hexapod robots (Lewis et al., 2000).

2.3 Neurodynamical Robot Controllers

Over time, increased computational power enabled the simulation and implementation of control networks that more directly encapsulated the dynamics of neurons and synapses in the nervous system. In 2008, Ayers and Rulkov implemented the structure from an earlier finite state machine controller (Ayers, 2004) as a network of spiking neurons that controlled the directional walking of their robot, RoboLobster (Ayers and Rulkov, 2008). They built the controller using a computationally efficient, two-dimensional phenomenological model of a spiking neuron, in which each neuron was tuned to produce spiking, bursting, or quiescence (Rulkov, 2002). The adoption of phenomenological neural models that eschew the computational complexity of traditional models [e.g., the Hodgkin-Huxley model (Hodgkin and Huxley, 1952)] facilitated the development of this and subsequent biology-first robot controllers despite limited available computing power.

Around this time, AMOS-WD06 and AMOS-WD08 were developed with RNN controllers that generated omnidirectional locomotion of 6- and 8-legged robots (Manoonpong, Pasemann and Wörgötter, 2008; Steingrube et al., 2010). The controllers implemented dynamical CPG models to produce alternating stepping signals, with walking direction changing due to descending commands and carefully designed networks that altered the phasing of joint movements relative to the master CPG. These robots demonstrated how simple descending commands could drive complex behavioral changes, mimicking the function of the nervous system. These robots also demonstrated how design principles could be used to tune networks to conduct specific functionality without large-scale optimization. The resulting networks had more meaningful structure and overall transparency than if they had been developed using a generic machine learning method, e.g., a perceptron network.

The controller for the modular legged robot Octavio was also developed using an RNN to generate locomotion patterns (von Twickel et al., 2011). Researchers tested multiple controllers on Octavio. Some used evolutionary algorithms to develop and refine modular single joint controllers that were later coupled together, while others were neurobiologically derived to mimic the earlier Ekeberg model of insect leg coordination (Ekeberg, Blümel and Büschges, 2004; von Twickel, Büschges and Pasemann, 2011). The modularity of Octavio’s limbs allowed for rigorous experimentation with both single leg controllers and the interlimb coordination for the entire network with 4, 6, or 8 legs.

Also inspired by the stick insect, Tarry IIB was designed to be controlled by the modular neural network controller, Walknet (Schmitz et al., 2008; Schilling et al., 2013). Walknet was constructed from networks of perceptrons (i.e., neuronal activation functions without dynamics), whose dynamics are simpler than those of other controllers from that time. Despite this simplification, Walknet on board Tarry IIB demonstrated how local reflexes could simplify the body-level postural control of robots and animals. Many legged robots are kinematically redundant structures, meaning that for a given body position, there are infinitely many possible joint angles. Such problems are often solved by constructing a kinematic model of the entire body and then minimizing some metric to calculate sequences of joint angles (Lynch and Park, 2017). In contrast to this highly centralized approach, Walknet implemented positive velocity feedback at most joints in the legs, a reflex observed in some animal behaviors (Bässler, 1976, 1986; Schmitz et al., 1995). Such reflexes enabled the robot’s leg joints to settle into feasible configurations without resisting one another during motion, all without centralized planning. This solution both demonstrated that complex control problems could be solved in an entirely decentralized way and provided a concrete hypothesis for the postural control of insects. It should also be noted that in contrast to many other models of insect motor control, Walknet has consistently considered the magnitude of motor output, rather than focusing on rhythm generation or multi-leg coordination, alone.

Researchers at FZI continued expanding their LAURON platform with the development of LAURON V (Roennau et al., 2014). This updated platform utilized a behavior-based control system in which “behavior blocks” produced actuator commands given sensory inputs, a rating criterion, and a motivation signal (Kerscher et al., 2008). In LAURON’s controller, six separate local leg behavior groups composed of swing, stance, ground contact, and collision behaviors directed the stepping of each leg. These local behavior groups were coordinated in different walking patterns (i.e., tripod, tetrapod, pentapod, and free gait) by abstractions of the Cruse Rules. The higher-level posture control group, comprised of body height, inclination, and body position behaviors, then ensured the overall stability of the robot. Thus, given desired velocities and walking patterns as external inputs, the controller was able to generate stable, autonomous walking. Notably, LAURON V demonstrated ability to cope with the dynamic, unstructured terrain of a staged search and rescue site in the EU Taranis Field Exercise, highlighting the controller’s robustness and flexibility.

Eventually, software tools such as AnimatLab (Cofer et al., 2010), the AnimatLab Robotics Toolkit (Szczecinski et al., 2015), and Nengo (Bekolay et al., 2014) facilitated desktop-computer assembly and testing of dynamical neuromechanics models. In the case of Animatlab, these models could be ported directly for robot control. All these platforms demonstrated that it was practical to simulate dynamical neural networks with commercial off-the-shelf (COTS) computing and actuator hardware. Thus, it became more practical to mimic the function and more of the morphology of animal nervous systems in robot controllers.

2.4 Morphological Neural Control

Increasingly powerful off-the-shelf computing hardware, coupled with a growing body of research in the neuroscience of invertebrate behavior, has recently facilitated the development of robots with control systems that control motion through biologically detailed, morphological mechanisms. In short, technology has enabled a biology-first approach to locomotion and navigation, which has led to controllers that leverage more biological detail than ever before.

As mentioned at the beginning of the previous section, RoboLobster’s controller demonstrated that the anatomical structure of crustacean thoracic circuits was sufficient to produce directed underwater stepping (Ayers and Rulkov, 2008). In particular, complex calculations emerged from simple neural dynamics. Networks could even be implemented on low-power LEGO robot hardware, demonstrating that dynamical neural controllers could solve complex control problems while using very limited computer power (Blustein, Rosenthal and Ayers, 2013). This group’s biology-first approach enabled the development of controllers with capabilities greater than the sum of their parts, that is, for the cost of simulating simple neural dynamics, they obtained emergent complex neural computation for robot control.

MantisBot, a robot developed in 2015 based on the Chinese mantis, controlled leg stepping with a dynamical neural framework similar to RoboLobster’s CCCPG approach (Szczecinski et al., 2015). The stepping controller was decentralized, using sensory feedback from leg kinematics and leg strain to coordinate multiple pattern generator networks throughout the leg. Similar techniques were later applied to Drosophibot, the group’s subsequent robot inspired by the fruit fly, and its walking (in simulation) (Goldsmith, Szczecinski and Quinn, 2020). One goal of these projects was to understand how decentralized walking motor programs could be modified by simple descending commands to cause the robot to start stepping, stop stepping, and step in different directions. It was found that only two descending signals, one for start/stop and one for walking path curvature, were required to modify stepping (Szczecinski et al., 2017). The same signals were distributed to each leg pair’s “ganglion” control module, but each utilized them differently to execute different changes in movement (Szczecinski and Quinn, 2017). This mechanism mimicked recordings in the cockroach in which intraleg segment phasing was altered by stimulating regions of the brain associated with curved walking (Martin et al., 2015). This biology-first approach resulted in a robot controller that served as a concrete hypothesis for how descending signals direct the stepping of insect legs, despite the distributed nature of the control networks. Subsequent experiments with a robotic leg have extended this work by modeling specific afferent pathways in the insect nervous system and how they may regulate its activity to change the effect of descending signals (Goldsmith, Quinn and Szczecinski, 2021).

A recent iteration of Walknet, which controls a hexapod robot called Hector, has also been closely based on specific neural structures and sensory inputs (Dürr et al., 2019). Walknet has maintained the decentralized control present since its conception, leveraging an increased amount of data available from experiments with the stick insect to further increase the biomimicry of its various control modules. The controller utilizes distributed proprioception and load sensing to implement local leg reflexes, paying close attention to which sensory modalities are known to contribute to each. The most recent version, neuroWalknet, is constructed from dynamical simulated neurons, with each leg pair controlled by a separate ganglion (Schilling and Cruse, 2020). Notably, the controller does not utilize CPGs for rhythm generation, instead relies on a combination of local reflexes, sensory feedback, and local gating networks for each joint, in which premotor neurons inhibit one another (e.g., femur-tibia joint extension inhibits flexion and vice versa). NeuroWalknet has reproduced many of the adaptive behaviors observed in animals and CPG-based models, such as the continuum of velocity-dependent interleg coordination patterns, context-dependent intraleg reflexes, and even rhythmic motor output after the application of pilocarpine [for a preprint, see (Schilling and Cruse, 2022)], a muscarinic agonist that evokes rhythmic neural activity when applied to the thoracic ganglia of some insects (Ryckebusch and Laurent, 1993; Büschges, Schmitz and Bässler, 1995). NeuroWalknet is a strong demonstration of how the biology-first approach can produce biomimetic and highly functional robot controllers.

Studies have also begun to use the influx of new information regarding the insect central complex (CX) on robotic platforms. The CX integrates many sensory modalities and appears to play a critical role in how insects localize within and navigate through their environment [for a brief review, see (Heinze, 2017)]. In work by Stone et al., a neural model of path integration and steering in the bee CX was implemented as part of a robot controller, enabling the robot to biomimetically integrate its path as it explored an area, then head straight “home” when finished exploring (Stone et al., 2017). The connectivity, receptive fields, and functions in the model were directly derived from electrophysiology and electron microscopy measurements from the bee, with a one-to-one correspondence between anatomical neural types and the neurons in the robot controller. The results provide a functional interpretation for many architectural features of the CX and capture features not present in previously proposed models. Furthermore, this demonstration clearly shows that a biology-first approach to robotics can lead to the creation of robots that behave in adaptive, animal-like ways.

These and other studies show the promise of the biology-first approach toward robot control (for a review, see (Webb, 2020)). The implementation of increasingly detailed neural models for robot control has benefitted from the ever-increasing computational power available in COTS electronics. However, as transistors have been miniaturized to the atomic scale, an end to Moore’s law is in sight (Shalf, 2020), meaning that such advances in raw computational power are unlikely to continue indefinitely. Simultaneously, more biological data than ever before is available, including electron microscope scans of the fruit fly brain and ventral nerve cord and other optogenetic imaging and stimulation methods (Zheng et al., 2018; Scheffer et al., 2020). To continue to apply the computation of the insect nervous system to real-time robot control, new software and hardware architectures are needed.

3. The Future of Arthropod-Inspired Robot Control

We anticipate that two unfolding advances will lead to further breakthroughs in insect inspired robotic control: Neuromorphic computers and increasingly biomimetic robotic structures.

3.1 Neuromorphic Computing

A major hindrance in the development of neural-based robotic control is that biological neural systems are organized in a manner almost completely opposite to traditional computer architectures; arthropod nervous systems contain millions of neurons operating asynchronously and in parallel, while traditional von Neumann computers use one or a small number of dedicated central processors which operate synchronously at high speed (von Neumann, 1993). Traditional computers are more than capable of simulating neural dynamics on a smaller scale. However, simulating networks consisting of thousands to millions of dynamic neurons is out of reach of embedded systems suitable for use on-board a mobile robot. To solve this problem, specialized computer hardware is needed.

Designing computer hardware for simulating neural dynamics is not a new field, dating back to simulating the Hodgkin-Huxley neural models on analog computers (FitzHugh, 1961). This practice came to the forefront with the work of Carver Mead, who coined the term “neuromorphic engineering” for describing the design of “very large scale integrated” (VLSI) circuits that mimic the layout of the nervous system (Mead and Ismail, 1989). This approach of using specific analog circuits to mimic neural networks continued (for its use in robot controllers, see Section 2.2.2), but has not yet reached widespread adoption due to the difficulty of tuning and designing a unique VLSI circuit for each application.

In response to this analog circuit development, other researchers have been designing general-purpose digital neuromorphic computer hardware for solving computational problems that are difficult for the von Neumann model. One of the first of these modern neuromorphic solutions was SpiNNaker (Khan et al., 2008), which combined large numbers of small processing cores onto a single chip in a manner similar to the design of modern supercomputers. This was followed by chips from IBM (Akopyan et al., 2015) and Intel (Davies et al., 2018), which were dedicated to simulating arbitrary networks of leaky integrate-and-fire spiking neurons. These digital processors have proved to be easier to manufacture than their analog predecessors, and due to their improved availability have been used in a variety of robotic applications (Davies et al., 2021; Cohen, 2022) including CPGs for legged locomotion of robots [(Gutierrez-Galan et al., 2019); in simulation: (Polykretis, Tang and Michmizos, 2020; Angelidis et al., 2021)]. As the second generation of these chips, which are capable of simulating up to one million simple neurons per chip, start to become available(Orchard et al., 2021; Yan et al., 2022), we foresee practical neural control systems running on-board legged robots approaching the scale of those in arthropod nervous systems.

Looking further into the future, work is ongoing on designing neuromorphic systems that move away from digital processors (Christensen et al., 2022). A new hybrid digital-analog neuromorphic chip uses transistors operating in their subthreshold regime to simulate spiking neurons using less than 40 percent the energy used by the leading digital chip (Neckar et al., 2019). Recent work has also shown the ability to manufacture circuits using memristors that can emulate the behavior of spiking neurons without any transistors (Kumar, Williams and Wang, 2020), a discovery which may lead to extremely dense arrays of power-efficient neurons in silico. As the capacity of neurons and synaptic connections per chip increases, we foresee neural control solutions becoming more widely implemented in both legged robotics and robotics in general.

3.2 Biomimetic Robot Structures

Just as the mechanical structure of an organism is intricately tied to its nervous system (Chiel and Beer, 1997), robot hardware also affects what computations are necessary for control. Thus, the concept of “morphological computation” came into being: The structure of the body inherently performs computations needed for control (Pfeifer, Iida and Gómez, 2006). As such, we expect that more effective robot locomotion control will arise in tandem with more biologically-inspired hardware, as the potential mismatch between mechanical structure and nervous system controller is minimized. (Manoonpong et al., 2021) provides a recent review of several advancements made to the mechanical structures of insect inspired robots, and we will describe several of what we consider the most crucial advancements here. The increasing availability of backdrivable, high-torque, low-speed motors enables actuators to achieve strength-to-weight ratios closer than ever to those of muscles (Seok et al., 2012, 2013). These motors are able to function similar to muscles even if the underlying physics are completely different, thus minimizing any mismatch between the structure and a highly neuromorphic controller. Alternatively, recent work produced actuators 3D printed out of biological tissue (Webster et al., 2017, Won et al., 2020), harkening to future biohybrid robots controlled by biological neural circuits.

Another aspect of biological structures that could improve robotic locomotion control is compliance. Biological organisms have a great deal of compliance throughout their limbs, profoundly affecting the way their bodies manage external forces and energy and how their nervous systems control movement (Chiel and Beer, 1997; Pfeifer, Iida and Gómez, 2006; Laschi and Mazzolai, 2016). Several of the most recent robots we mentioned have attempted to account for this compliance in their robots’ structures (Dürr et al., 2019; Goldsmith, Szczecinski and Quinn, 2020), but further investigation could provide insight into precisely how such compliance affects nervous system control.

Feet are one area of compliance in biological organisms that is presently under-investigated in robotic systems, but could produce a variety of advancements for walking robots. Feet may be particularly important to insect-inspired robots, as insect tarsi are compliantly nested tarsal segments comprising nearly 30% of leg length (Manoonpong et al., 2021). Compliant feet may aid in conforming to a variety of substrates, with additional foot actuation possibly aiding in propulsion by “pushing off” at the end of stance and aiding in stability by gripping the substrate. Adding appropriate feet to robots may facilitate larger increases in efficiency than control software or hardware.

Biomimetic sensing could also provide insights into how the nervous system processes sensory feedback. Traditional robotic approaches prioritize a small number of central, high power, and high precision sensors for internal and external sensing. In contrast, the nervous system utilizes readings from a multitude of distributed, lower precision sensory organs. Several of these sensors are range-fractionated, with unique static and dynamic ranges that provide redundant and resilient information about the body’s state [for a review, see (Delcomyn, Nelson and Cocatre-Zilgien, 1996)]. Taking a similar approach in robotic systems may reduce computational complexity and increase the robustness and resilience of the robot’s sensory feedback. Additionally, place-coded sensory information may facilitate alternative computational approaches that sidestep the need for computationally expensive coordinate frame transformations [in print: (Guie and Szczecinski 2022)].

3.3 Final Thoughts

The past several decades have seen great progress in the application of biological principles to the control of insect-like legged locomotion and we foresee this trend expanding in the future. As principles are incorporated more directly, the robots serve as models of the animals under investigation. Furthermore, such a biology-first approach to robotics is a fruitful long-term investment strategy for understanding what makes animal motion so adaptable, robust, and resilient. Advances in neuroscience methods and data collection as well as computing power for robot control have supported the growth of bioinspired robotics. Current work in neuromorphic hardware and specialized actuators will encourage researchers to develop more animal-like robots in the coming years. We predict that further advances in computation due to software architectures, hardware architectures, morphology, and sensing will drive even more convergence between robotics, neuroscience, and biomechanics, emphasizing the importance of a biology-first approach.

The scope of this perspective article is limited to insect nervous system-inspired control of hexapedal robotic locomotion. The neuroscience of motor control, the biomechanics of locomotion, and robot control are all vast fields in their own rights. To learn more about these fields, the authors of this article suggest other review articles (Marder and Bucher, 2001; Mulloney and Smarandache, 2010; Ijspeert, 2014; Buschmann et al., 2015; Bidaye, Bockemühl and Büschges, 2017; Webb, 2020; Manoonpong et al., 2021).

References

Akopyan, F. et al. (2015) “TrueNorth: Design and Tool Flow of a 65 mW 1 Million Neuron Programmable Neurosynaptic Chip,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 34(10), pp. 1537–1557. Available at: https://doi.org/10.1109/TCAD.2015.2474396.

Angelidis, E. et al. (2021) “A Spiking Central Pattern Generator for the control of a simulated lamprey robot running on SpiNNaker and Loihi neuromorphic boards.” Available at: http://arxiv.org/abs/2101.07001.

Ayers, J. (2004) “Underwater walking,” Arthropod Structure and Development, 33(3), pp. 347–360. Available at: https://doi.org/10.1016/j.asd.2004.06.001.

Ayers, J. et al. (2010) “Controlling underwater robots with electronic nervous systems,” Applied Bionics and Biomechanics, 7(1), pp. 57–67. Available at: https://doi.org/10.1080/11762320903244843.

Ayers, J. and Rulkov, N. (2008) “Controlling Biomimetic Underwater Robots With Electronic Nervous Systems,” in Bio-mechanisms of Swimming and Flying. Tokyo: Springer Japan, pp. 295–306. Available at: https://doi.org/10.1007/978-4-431-73380-5_24.

Ayers, J.L. and Selverston, A.I. (1979) “Monosynaptic entrainment of an endogenous pacemaker network: A cellular mechanism for von Holst’s magnet effect,” Journal of Comparative Physiology A, 129(1), pp. 5–17. Available at: https://doi.org/10.1007/BF00679907.

Bässler, U. (1976) “Reversal of a reflex to a single motoneuron in the stick insect Çarausius morosus,” Biological Cybernetics, 24(1), pp. 47–49. Available at: https://doi.org/10.1007/bf00365594.

Bässler, U. (1986) “Afferent control of walking movements in the stick insect Cuniculina impigra II. Reflex reversal and the release of the swing phase in the restrained foreleg,” Journal of Comparative Physiology A, 158, pp. 351–362. Available at: https://doi.org/10.1007/BF00603619.

Bässler, U. and Büschges, A. (1998) “Pattern generation for stick insect walking movements—multisensory control of a locomotor program,” Brain Research Reviews, 27(1), pp. 65–88. Available at: https://doi.org/10.1016/S0165-0173(98)00006-X.

Beer, R. and Chiel, H. (1989) “Neural Implementation of Motivated Behavior: Feeding in an Artificial Insect,” in D. Touretzky (ed.) Advances in Neural Information Processing Systems. Morgan-Kaufmann. Available at: https://proceedings.neurips.cc/paper/1989/file/c52f1bd66cc19d05628bd8bf27af3ad6-Paper.pdf.

Beer, R.D. et al. (1992) “A Distributed Neural Network Architecture for Hexapod Robot Locomotion,” Neural Computation, 4, pp. 356–365.

Beer, R.D., Chiel, H.J. and Sterling, L. (1989) “Heterogeneous Neural Networks for Adaptive Behavior in Dynamic Environments,” Advances in Neural Information Processing Systems 1, pp. 577–585. Available at: http://papers.nips.cc/paper/115-heterogeneous-neural-networks-for-adaptive-behavior-in-dynamic-environments.pdf%5Cnfiles/4832/Beer et al. - 1989 - Heterogeneous Neural Networks for Adaptive Behavio.pdf%5Cnfiles/4833/115-heterogeneous-neural-networks-for-a.

Beer, R.D., Chiel, H.J. and Sterling, L.S. (1990) “A biological perspective on autonomous agent design,” Robotics and Autonomous Systems, 6(1–2), pp. 169–186. Available at: https://doi.org/10.1016/S0921-8890(05)80034-X.

Beer, R.D. and Gallagher, J.C. (1992) “Evolving Dynamical Neural Networks for Adaptive Behavior,” Adaptive Behavior, 1(1), pp. 91–122. Available at: https://doi.org/10.1177/105971239200100105.

Bekolay, T. et al. (2014) “Nengo: A Python tool for building large-scale functional brain models,” Frontiers in Neuroinformatics, 7(JAN). Available at: https://doi.org/10.3389/fninf.2013.00048.

Berns, K., Cordes, S. and Ilg, W. (1994) “Adaptive, neural control architecture for the walking machine LAURON,” in IEEE/RSJ/GI International Conference on Intelligent Robots and Systems. IEEE, pp. 1172–1177. Available at: https://doi.org/10.1109/iros.1994.407466.

Bidaye, S.S., Bockemühl, T. and Büschges, A. (2017) “Six-legged walking in insects: how CPGs, peripheral feedback, and descending signals generate coordinated and adaptive motor rhythms,” Journal of Neurophysiology [Preprint]. Available at: https://doi.org/10.1152/jn.00658.2017.

Blustein, D., Rosenthal, N. and Ayers, J. (2013) “Designing and Implementing Nervous System Simulations on LEGO Robots,” Journal of Visualized Experiments, (75), pp. 1–5. Available at: https://doi.org/10.3791/50519.

Brioist, P. (2020) “Leonardo da Vinci e il metodo biomimetico.”

Brooks, R.A. (1989) “A Robot that Walks; Emergent Behaviors from a Carefully Evolved Network,” Neural Computation, 1(2), pp. 253–262. Available at: https://doi.org/10.1162/neco.1989.1.2.253.

Brown, B.E. (2000) An analog VLSI implementation of a continuous-time recurrent neural network. Master of Science in Engineering. Case Western Reserve University.

Burrows, M. (1980) The control of sets of motoneurones by local interneurones in the locust, J. Physiol.

Büschges, A., Schmitz, J. and Bässler, U. (1995) “Rhythmic patterns in the thoracic nerve cord of the stick insect induced by pilocarpine,” Journal of Experimental Biology, 198(Pt 2), pp. 435–456. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9318078.

Buschmann, T. et al. (2015) “Controlling legs for locomotion - insights from robotics and neurobiology,” Bioinspiration and Biomimetics, 10(4), p. 41001. Available at: https://doi.org/10.1088/1748-3190/10/4/041001.

Chiel, H.J. and Beer, R.D. (1989) “A lesion study of a heterogeneous artificial neural network for hexapod locomotion,” in International Joint Conference on Neural Networks, pp. 407–414.

Chiel, H.J. and Beer, R.D. (1997) “The brain has a body: adaptive behavior emerges from interactions of nervous system, body and environment.,” Trends in neurosciences, 20(12), pp. 553–7. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9416664.

Christensen, D. v et al. (2022) “2022 roadmap on neuromorphic computing and engineering,” Neuromorphic Computing and Engineering, 2(2), p. 022501. Available at: https://doi.org/10.1088/2634-4386/ac4a83.

Cofer, D.W. et al. (2010) “AnimatLab: a 3D graphics environment for neuromechanical simulations.,” Journal of neuroscience methods, 187(2), pp. 280–8. Available at: https://doi.org/10.1016/j.jneumeth.2010.01.005.

Cohen, A.H. and Wallen, P. (1980) “The Neuronal Correlate of Locomotion in Fish,” Exp Brain Res, 41, pp. 11–18.

Cohen, G. (2022) “Gooaall!!!: Why we Built a Neuromorphic Robot to Play Foosball,” IEEE Spectrum, 59(3), pp. 44–50. Available at: https://doi.org/10.1109/MSPEC.2022.9729948.

Cruse, H. (1976) “The function of the legs in the free walking stick insect, Carausius morosus,” Journal of Comparative Physiology ■ A, 112(2), pp. 235–262. Available at: https://doi.org/10.1007/BF00606541.

Cruse, H. (1990) “What mechanisms coordinate leg movement in walking arthropods?,” Trends in neurosciences, 13(1), pp. 15–21. Available at: https://doi.org/10.1016/0166-2236(90)90057-H.

Cruse, H. et al. (1995) “Walking: A Complex Behavior Controlled by Simple Networks,” Adaptive Behavior, 3(4), pp. 385–418. Available at: https://doi.org/10.1177/105971239500300403.

Davies, M. et al. (2018) “Loihi: A Neuromorphic Manycore Processor with On-Chip Learning,” IEEE Micro, 38(1), pp. 82–99. Available at: https://doi.org/10.1109/MM.2018.112130359.

Davies, M. et al. (2021) “Advancing Neuromorphic Computing with Loihi: A Survey of Results and Outlook,” Proceedings of the IEEE, 109(5), pp. 911–934. Available at: https://doi.org/10.1109/JPROC.2021.3067593.

Delcomyn, F. (1980) “Neural basis of rhythmic behavior in animals.,” Science, 210(4469), pp. 492–498. Available at: https://doi.org/10.1126/science.7423199.

Delcomyn, F., Nelson, M.E. and Cocatre-Zilgien, J.H. (1996) “Sense organs of insect legs and the selection of sensors for agile walking robots,” International Journal of Robotics Research, 15(2), pp. 113–127. Available at: https://doi.org/10.1177/027836499601500201.

Delvolvé, I. et al. (1999) “Fictive Rhythmic Motor Patterns Induced by NMDA in an In Vitro Brain Stem–Spinal Cord Preparation From an Adult Urodele,” Journal of Neurophysiology, 82(2), pp. 1074–1077. Available at: https://doi.org/10.1152/jn.1999.82.2.1074.

Driesang, R.B. and Büschges, A. (1996) “Physiological changes in central neuronal pathways contributing to the generation of a reflex reversal,” Journal of Comparative Physiology - A Sensory, Neural, and Behavioral Physiology, 179(1), pp. 45–57. Available at: https://doi.org/10.1007/BF00193433.

Dürr, V. et al. (2019) “Integrative Biomimetics of Autonomous Hexapedal Locomotion,” Frontiers in Neurorobotics, 13(October), pp. 1–32. Available at: https://doi.org/10.3389/fnbot.2019.00088.

Ekeberg, Ö., Blümel, M. and Büschges, A. (2004) “Dynamic simulation of insect walking,” Arthropod Structure and Development, 33(3), pp. 287–300. Available at: https://doi.org/10.1016/j.asd.2004.05.002.

Espenschied, K. et al. (1995) “Biologically-Inspired Hexapod Robot Project: Second Robot,” in Proceedings of 1995 IEEE International Conference on Robotics and Automation. IEEE. Available at: https://doi.org/10.1109/ROBOT.1995.525750.

Espenschied, K.S. et al. (1993) “Leg coordination mechanisms in the stick insect applied to hexapod robot locomotion,” Adaptive Behavior, 1(4), pp. 455–468.

Espenschied, K.S. et al. (1996) “Biologically based distributed control and local reflexes improve rough terrain locomotion in a hexapod robot,” Robotics and Autonomous Systems, 18(1–2), pp. 59–64. Available at: https://doi.org/10.1016/0921-8890(96)00003-6.

FitzHugh, R. (1961) “Impulses and Physiological States in Theoretical Models of Nerve Membrane,” Biophysical Journal, 1(6), pp. 445–466. Available at: https://doi.org/10.1016/S0006-3495(61)86902-6.

Gallagher, J.C. et al. (1996) “Application of evolved locomotion controllers to a hexapod robot,” Robotics and Autonomous Systems, 19, pp. 95–103.

Gaßmann, B., Scholl, K.U. and Berns, K. (2001) “Locomotion of LAURON III in rough terrain,” IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM, 2(July), pp. 959–964. Available at: https://doi.org/10.1109/aim.2001.936810.

Goldsmith, C., Szczecinski, N.S. and Quinn, R.D. (2020) “Neurodynamic modeling of the Fruit Fly Drosophila melanogaster,” Bioinspiration & Biomimetics [Preprint]. Available at: https://doi.org/10.1088/1748-3190/ab9e52.

Goldsmith, C.A., Quinn, R.D. and Szczecinski, N.S. (2021) “Investigating the role of low level reinforcement reflex loops in insect locomotion,” Bioinspiration & Biomimetics, 16(6), p. 065008. Available at: https://doi.org/10.1088/1748-3190/ac28ea.

Grillner, S. (1985) “Neural Control of Vertebrate Locomotion - Central Mechanisms and Reflex Interaction with Special Reference to the Cat,” in Feedback and Motor Control in Invertebrates and Vertebrates. Dordrecht: Springer Netherlands, pp. 35–56. Available at: https://doi.org/10.1007/978-94-011-7084-0_3.

Gutierrez-Galan, D. et al. (2019) “NeuroPod: a real-time neuromorphic spiking CPG applied to robotics.” Available at: https://doi.org/10.1016/j.neucom.2019.11.007.

Heinze, S. (2017) “Neural Coding: Bumps on the Move,” Current Biology, 27(11), pp. R409–R412. Available at: https://doi.org/10.1016/j.cub.2017.03.072.

Hess, D. and Büschges, A. (1999) “Role of Proprioceptive Signals From an Insect Femur-Tibia Joint in Patterning Motoneuronal Activity of an Adjacent Leg Joint,” Journal of Neurophysiology, 81(4), pp. 1856–1865.

Hodgkin, A.L. and Huxley, A.F. (1952) “A quantitative description of membrane current and its application to conduction and excitation in nerve,” The Journal of physiology, 117(4), p. 500. Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1392413/ (Accessed: October 6, 2011).

Holst, E. (1939) “Die relative Koordination als Phänomen und als Methode zentral-nervöser Funktionsanalyse,” Ergebnisse der Physiologie Biologischen Chemie und Experimentellen Pharmakologie, 42(1), pp. 228–306. Available at: https://doi.org/10.1007/BF02322567.

Hughes, G.M. and Wiersma, C.A.G. (1960) “The Co-ordination of Swimmeret Movements in the Crayfish, Procambarus Clarkii (Girard),” Journal of Experimental Biology, 37(4), pp. 657–670. Available at: https://doi.org/10.1242/jeb.37.4.657.

Ijspeert, A.J. (2008) “Central pattern generators for locomotion control in animals and robots: a review.,” Neural networks : the official journal of the International Neural Network Society, 21(4), pp. 642–53. Available at: https://doi.org/10.1016/j.neunet.2008.03.014.

Ijspeert, A.J. (2014) “Biorobotics: using robots to emulate and investigate agile locomotion.,” Science (New York, N.Y.), 346(6206), pp. 196–203. Available at: https://doi.org/10.1126/science.1254486.

Ilg, W. and Berns, K. (1995) “A learning architecture based on reinforcement learning for adaptive control of the walking machine LAURON,” Robotics and Autonomous Systems, 15(4), pp. 321–334. Available at: https://doi.org/10.1016/0921-8890(95)00009-5.

Kerscher, T. et al. (2008) Behaviour-based control of the six-legged walking machine LAURON IVe. Available at: www.jzi.de.

Khan, M.M. et al. (2008) “SpiNNaker: Mapping neural networks onto a massively-parallel chip multiprocessor,” in 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence). Hong Kong, China: IEEE, pp. 2849–2856. Available at: https://doi.org/10.1109/IJCNN.2008.4634199.

Kumar, S., Williams, R.S. and Wang, Z. (2020) “Third-order nanocircuit elements for neuromorphic engineering,” Nature, 585(7826), pp. 518–523. Available at: https://doi.org/10.1038/s41586-020-2735-5.

Laschi, C. and Mazzolai, B. (2016) “Lessons from animals and plants: The symbiosis of morphological computation and soft robotics,” IEEE Robotics and Automation Magazine, 23(3), pp. 107–114. Available at: https://doi.org/10.1109/MRA.2016.2582726.

Lewis, M.A. et al. (2000) “Toward biomorphic control using custom a VLSI CPG chips,” Proceedings - IEEE International Conference on Robotics and Automation, 1, pp. 494–500. Available at: https://doi.org/10.1109/robot.2000.844103.

Lynch, K.M. and Park, F. (2017) Modern Robotics -Mechanics, Planning, and Control, Cambridge University Press. Available at: http://hades.mech.northwestern.edu/images/7/7f/MR.pdf.

Manoonpong, P. et al. (2021) “Insect-inspired robots: Bridging biological and artificial systems,” Sensors, 21(22), pp. 1–44. Available at: https://doi.org/10.3390/s21227609.

Manoonpong, P., Pasemann, F. and Wörgötter, F. (2008) “Sensor-driven neural control for omnidirectional locomotion and versatile reactive behaviors of walking machines,” Robotics and Autonomous Systems, 56(3), pp. 265–288. Available at: https://doi.org/10.1016/j.robot.2007.07.004.

Marder, E. and Bucher, D. (2001) Central pattern generators and the control of rhythmic movements, Current Biology.

Martin, J.P. et al. (2015) “Central-complex control of movement in the freely walking cockroach,” Current Biology, 25(21), pp. 2795–2803. Available at: https://doi.org/10.1016/j.cub.2015.09.044.

Mayr, C., Hoeppner, S. and Furber, S. (2019) “SpiNNaker 2: A 10 Million Core Processor System for Brain Simulation and Machine Learning.”

Mead, C. and Ismail, M. (1989) Analog VLSI implementation of neural systems. Springer Science & Business Media.

Mu, L. and Ritzmann, R.E. (2008) “Interaction between descending input and thoracic reflexes for joint coordination in cockroach: I. Descending influence on thoracic sensory reflexes,” Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology, 194(3), pp. 283–298. Available at: https://doi.org/10.1007/s00359-007-0307-x.

Mulloney, B. and Selverston, A.I. (1974a) “Organization of the stomatogastric ganglion of the spiny lobster - I. Neurons Driving the Lateral Teeth,” Journal of Comparative Physiology, 91(1), pp. 1–32. Available at: https://doi.org/10.1007/BF00696154.

Mulloney, B. and Selverston, A.I. (1974b) “Organization of the stomatogastric ganglion of the spiny lobster - III. Coordination of the two subsets of the gastric system,” Journal of Comparative Physiology, 91(1), pp. 53–78. Available at: https://doi.org/10.1007/BF00696154.

Mulloney, B. and Smarandache, C. (2010) “Fifty years of CPGs: Two neuroethological papers that shaped the course of neuroscience,” Frontiers in Behavioral Neuroscience. Available at: https://doi.org/10.3389/fnbeh.2010.00045.

Neckar, A. et al. (2019) “Braindrop: A Mixed-Signal Neuromorphic Architecture with a Dynamical Systems-Based Programming Model,” Proceedings of the IEEE, 107(1), pp. 144–164. Available at: https://doi.org/10.1109/JPROC.2018.2881432.

von Neumann, J. (1993) “First draft of a report on the EDVAC,” IEEE Annals of the History of Computing, 15(4), pp. 27–75. Available at: https://doi.org/10.1109/85.238389.

Orchard, G. et al. (2021) “Efficient Neuromorphic Signal Processing with Loihi 2,” in IEEE Workshop on Signal Processing Systems, SiPS: Design and Implementation. Institute of Electrical and Electronics Engineers Inc., pp. 254–259. Available at: https://doi.org/10.1109/SiPS52927.2021.00053.

Pearson, K.G. (1972) “Central programming and reflex control of walking in the cockroach,” Journal of Experimental Biology, 56, pp. 173–193.

Pearson, Keir G (1995) “Proprioceptive regulation of locomotion,” Current Opinion in Neurobiology, 5(6), pp. 786–791. Available at: https://doi.org/10.1016/0959-4388(95)80107-3.

Pearson, K. G. (1995) “Reflex Reversal in the Walking Systems of Mammals and Arthropods,” in Neural Control of Movement. Boston, MA: Springer US, pp. 135–141. Available at: https://doi.org/10.1007/978-1-4615-1985-0_18.

Perkel, D.H. and Mulloney, B. (1974) “Motor pattern production in reciprocally inhibitory neurons exhibiting postinhibitory rebound,” Science, 185, pp. 181–182.

Pfeifer, R., Iida, F. and Gómez, G. (2006) “Morphological computation for adaptive behavior and cognition,” International Congress Series, 1291, pp. 22–29. Available at: https://doi.org/10.1016/j.ics.2005.12.080.

Pfeiffer, F., Eltze, J. and Weidemann, H.-J. (1995) “The tum-walking machine,” Intelligent Automation and Soft Computing, 1(3), pp. 307–323.

Pflüger, H.J., Duch, C. and Heidel, E. (2004) “Neuromodulatory octopaminergic neurons and their functions during insect motor behaviour,” Acta Biologica Hungarica, 55(1–4), pp. 3–12. Available at: https://doi.org/10.1556/ABiol.55.2004.1-4.2.

Pinsker, H.M. and Ayers, J. (1983) “Neuronal Oscillators,” in The Clinical Neurosciences, pp. 203–266. Available at: https://doi.org/10.1093/acprof:oso/9780195371833.003.0007.

Polykretis, I., Tang, G. and Michmizos, K.P. (2020) “An Astrocyte-Modulated Neuromorphic Central Pattern Generator for Hexapod Robot Locomotion on Intel’s Loihi,” in International Conference on Neuromorphic Systems 2020. New York, NY, USA: ACM, pp. 1–9. Available at: https://doi.org/10.1145/3407197.3407205.

Roennau, A. et al. (2014) “LAURON V: A versatile six-legged walking robot with advanced maneuverability,” IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM, pp. 82–87. Available at: https://doi.org/10.1109/AIM.2014.6878051.

Rubeo, S., Szczecinski, N. and Quinn, R. (2017) “A Synthetic Nervous System Controls a Simulated Cockroach,” Applied Sciences, 8(1), p. 6. Available at: https://doi.org/10.3390/app8010006.

Rulkov, N.F. (2002) “Modeling of spiking-bursting neural behavior using two-dimensional map,” Physical Review E - Statistical, Nonlinear, and Soft Matter Physics, 65(4), pp. 1–9. Available at: https://doi.org/10.1103/PhysRevE.65.041922.

Ryckebusch, S. and Laurent, G. (1993) “Rhythmic patterns evoked in locust leg motor neurons by the muscarinic agonist pilocarpine.,” Journal of neurophysiology, 69(5), pp. 1583–95. Available at: http://www.ncbi.nlm.nih.gov/pubmed/8389831.

Sauer, A.E. et al. (1996) “Distributed processing on the basis of parallel and antagonistic pathways simulation of the femur-tibia control system in the stick insect,” Journal of Computational Neuroscience, 3(3), pp. 179–198. Available at: https://doi.org/10.1007/BF00161131.

Sauer, A.E., Büschges, A. and Stein, W. (1997) “Role of presynaptic inputs to proprioceptive afferents in tuning sensorimotor pathways of an insect joint control network,” Journal of Neurobiology, 32(4), pp. 359–376. Available at: https://doi.org/10.1002/(SICI)1097-4695(199704)32:4<359::AID-NEU1>3.0.CO;2-5.

Scheffer, L.K. et al. (2020) “A connectome and analysis of the adult drosophila central brain,” eLife, 9, pp. 1–74. Available at: https://doi.org/10.7554/ELIFE.57443.

Schilling, M. et al. (2013) “Walknet, a bio-inspired controller for hexapod walking.,” Biological cybernetics, 107(4), pp. 397–419. Available at: https://doi.org/10.1007/s00422-013-0563-5.

Schilling, M. and Cruse, H. (2020) “Decentralized control of insect walking: A simple neural network explains a wide range of behavioral and neurophysiological results,” PLoS Computational Biology, 16(4), pp. 1–48. Available at: https://doi.org/10.1371/journal.pcbi.1007804.

Schilling, M. and Cruse, H. (2022) “neuroWalknet, a controller for hexapod walking allowing for context dependent behavior,” bioRxiv preprint [Preprint]. Available at: https://doi.org/10.1101/2022.04.27.489633.

Schmitz, J. et al. (1995) “Adaptive properties of hard-wired neuronal systems,” Verh. dt. zool., 88.2, pp. 95–105. Available at: http://pub.uni-bielefeld.de/publication/1681412.

Schmitz, J. et al. (2008) “No need for a body model: Positive velocity feedback for the control of an 18-DOF robot walker,” Applied Bionics and Biomechanics, 5(3), pp. 135–147. Available at: https://doi.org/10.1080/11762320802221074.

Selverston, A.I. et al. (1976) “The stomatogastric nervous system: Structure and function of a small neural network,” Progress in Neurobiology, 7, pp. 215–289. Available at: https://doi.org/10.1016/0301-0082(76)90008-3.

Seok, S. et al. (2012) “Actuator design for high force proprioceptive control in fast legged locomotion,” in IEEE International Conference on Intelligent Robots and Systems, pp. 1970–1975. Available at: https://doi.org/10.1109/IROS.2012.6386252.

Seok, S. et al. (2013) “Design principles for highly efficient quadrupeds and implementation on the MIT Cheetah robot,” in Proceedings - IEEE International Conference on Robotics and Automation, pp. 3307–3312. Available at: https://doi.org/10.1109/ICRA.2013.6631038.

Shalf, J. (2020) “The future of computing beyond Moore’s Law,” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences. Royal Society Publishing. Available at: https://doi.org/10.1098/rsta.2019.0061.

Steingrube, S. et al. (2010) “Self-organized adaptation of a simple neural circuit enables complex robot behaviour,” Nature Physics, 6(3), pp. 224–230. Available at: https://doi.org/10.1038/nphys1508.

Steuer, J. and Pfeiffer, F. (1997) CONTROL OF A SIX-LEGGED-WALKING-MACHINE WORKING IN UNEVEN TERRAIN, Copyright (eJ IFAC Robot Control.

Still, S. and Tilden, M.W. (1998) “Controller for a Four-Legged Walking Machine,” in Neuromorphic Systems, pp. 138–148. Available at: https://doi.org/10.1142/9789812816535_0012.

Stone, T. et al. (2017) “An Anatomically Constrained Model for Path Integration in the Bee Brain,” Current Biology, 27(20), pp. 3069-3085.e11. Available at: https://doi.org/10.1016/j.cub.2017.08.052.

Szczecinski, N.S. et al. (2015) “Introducing MantisBot: Hexapod robot controlled by a high-fidelity, real-time neural simulation,” in IEEE International Conference on Intelligent Robots and Systems. Hamburg, DE, pp. 3875–3881. Available at: https://doi.org/10.1109/IROS.2015.7353922.

Szczecinski, N.S. et al. (2017) “MantisBot is a Robotic Model of Visually Guided Motion in the Praying Mantis,” Arthropod Structure and Development [Preprint]. Available at: https://doi.org/10.1016/j.asd.2017.03.001.

Szczecinski, N.S. and Quinn, R.D. (2017) “Template for the neural control of directed stepping generalized to all legs of MantisBot,” Bioinspiration & Biomimetics, 12(4), p. 045001. Available at: https://doi.org/10.1088/1748-3190/aa6dd9.

Thompson, W.J. and Stent, G.S. (1976a) “Neuronal Control of Heartbeat in the Medicinal Leech I. Generation of the Vascular Constriction Rhythm by Heart Motor Neurons,” J. comp. Physiol. III, 261, p. 279.

Thompson, W.J. and Stent, G.S. (1976b) Neuronal Control of Heartbeat in the Medicinal Leech II. Intersegmental Coordination of Heart Motor Neuron Activity by Heart Interneurons, J. comp. Physiol.

von Twickel, A. et al. (2011) “Neural control of a modular multi-legged walking machine: Simulation and hardware,” Robotics and Autonomous Systems, 60(2), pp. 227–241. Available at: https://doi.org/10.1016/j.robot.2011.10.006.

von Twickel, A., Büschges, A. and Pasemann, F. (2011) “Deriving neural network controllers from neuro-biological data: Implementation of a single-leg stick insect controller,” Biological Cybernetics, 104(1–2), pp. 95–119. Available at: https://doi.org/10.1007/s00422-011-0422-1.

Webb, B. (2020) “Robots with insect brains,” Science, 368(6488), pp. 244–245. Available at: https://doi.org/10.1126/science.aaz6869.

Webster, V.A. et al. (2017) “3D-Printed Biohybrid Robots Powered by Neuromuscular Tissue Circuits from Aplysia californica,” in M. Mangan et al. (eds). Cham: Springer International Publishing (Lecture Notes in Computer Science), pp. 475–486. Available at: https://doi.org/10.1007/978-3-319-63537-8_40.

Wendler, G. (1974) “The influence of proprioceptive feedback on Locust flight co-ordination,” Journal of Comparative Physiology, 88(2), pp. 173–200. Available at: https://doi.org/10.1007/BF00695406.

Wilson, D.M. (1961) “The central nervous control of flight in a locust,” Journal of Experimental Biology, 38(2), pp. 471–490.

Wilson, D.M. (1966) “Insect walking,” Annual review of entomology, 11(1), pp. 103–122.

Wilson, D.M. and Wyman, R.J. (1965) “Motor Output Patterns during Random and Rhythmic Stimulation of Locust Thoracic Ganglia,” Biophysical Journal, 5(2), pp. 121–143. Available at: https://doi.org/10.1016/S0006-3495(65)86706-6.

Yan, X. et al. (2022) “Progress and Challenges for Memtransistors in Neuromorphic Circuits and Systems,” ADVANCED MATERIALS, p. 2108025.

Zheng, Z. et al. (2018) “A Complete Electron Microscopy Volume of the Brain of Adult Drosophila melanogaster,” Cell, 174(3), pp. 730-743.e22. Available at: https://doi.org/10.1016/j.cell.2018.06.019.