Asymmetric fin shape changes swimming dynamics of ancient marine reptiles' soft robophysical models.

Animals have evolved highly effective locomotion capabilities in terrestrial, aerial, and aquatic environments. Over life's history, mass extinctions have wiped out unique animal species with specialized adaptations, leaving paleontologists to reconstruct their locomotion through fossil analysis. Despite advancements, little is known about how extinct megafauna, such as the Ichthyosauria one of the most successful lineages of marine reptiles, utilized their varied morphologies for swimming. Traditional robotics struggle to mimic extinct locomotion effectively, but the emerging soft robotics field offers a promising alternative to overcome this challenge. This paper aims to bridge this gap by studying Mixosaurus locomotion with soft robotics, combining material modeling and biomechanics in physical experimental validation. Combining a soft body with soft pneumatic actuators, the soft robotic platform described in this study investigates the correlation between asymmetrical fins and buoyancy by recreating the pitch torque generated by extinct swimming animals. We performed a comparative analysis of thrust and torque generated by Carthorhyncus, Utatsusaurus, Mixosaurus, Guizhouichthyosaurus, and Ophthalmosaurus tail fins in a flow tank. Experimental results suggest that the pitch torque on the torso generated by hypocercal fin shapes such as found in model systems of Guizhouichthyosaurus, Mixosaurus and Utatsusaurus produce distinct ventral body pitch effects able to mitigate the animal's non-neutral buoyancy. This body pitch control effect is particularly pronounced in Guizhouichthyosaurus, which results suggest would have been able to generate high ventral pitch torque on the torso to compensate for its positive buoyancy. By contrast, homocercal fin shapes may not have been conducive for such buoyancy compensation, leaving torso pitch control to pectoral fins, for example. Across the range of the actuation frequencies of the caudal fins tested, resulted in oscillatory modes arising, which in turn can affect the for-aft thrust generated.

Viability leads to the emergence of gait transitions in learning agile quadrupedal locomotion on challenging terrains.

Quadruped animals are capable of seamless transitions between different gaits. While energy efficiency appears to be one of the reasons for changing gaits, other determinant factors likely play a role too, including terrain properties. In this article, we propose that viability, i.e., the avoidance of falls, represents an important criterion for gait transitions. We investigate the emergence of gait transitions through the interaction between supraspinal drive (brain), the central pattern generator in the spinal cord, the body, and exteroceptive sensing by leveraging deep reinforcement learning and robotics tools. Consistent with quadruped animal data, we show that the walk-trot gait transition for quadruped robots on flat terrain improves both viability and energy efficiency. Furthermore, we investigate the effects of discrete terrain (i.e., crossing successive gaps) on imposing gait transitions, and find the emergence of trot-pronk transitions to avoid non-viable states. Viability is the only improved factor after gait transitions on both flat and discrete gap terrains, suggesting that viability could be a primary and universal objective of gait transitions, while other criteria are secondary objectives and/or a consequence of viability. Moreover, our experiments demonstrate state-of-the-art quadruped robot agility in challenging scenarios.

Evaluation of controllers for augmentative hip exoskeletons and their effects on metabolic cost of walking: explicit versus implicit synchronization.

Background: Efficient gait assistance by augmentative exoskeletons depends on reliable control strategies. While numerous control methods and their effects on the metabolic cost of walking have been explored in the literature, the use of different exoskeletons and dissimilar protocols limit direct comparisons. In this article, we present and compare two controllers for hip exoskeletons with different synchronization paradigms. Methods: The implicit-synchronization-based approach, termed the Simple Reflex Controller (SRC), determines the assistance as a function of the relative loading of the feet, resulting in an emerging torque profile continuously assisting extension during stance and flexion during swing. On the other hand, the Hip-Phase-based Torque profile controller (HPT) uses explicit synchronization and estimates the gait cycle percentage based on the hip angle, applying a predefined torque profile consisting of two shorter bursts of assistance during stance and swing. We tested the controllers with 23 naïve healthy participants walking on a treadmill at 4 km ⋅ h-1, without any substantial familiarization. Results: Both controllers significantly reduced the metabolic rate compared to walking with the exoskeleton in passive mode, by 18.0% (SRC, p < 0.001) and 11.6% (HPT, p < 0.001). However, only the SRC led to a significant reduction compared to walking without the exoskeleton (8.8%, p = 0.004). The SRC also provided more mechanical power and led to bigger changes in the hip joint kinematics and walking cadence. Our analysis of mechanical powers based on a whole-body analysis suggested a reduce in ankle push-off under this controller. There was a strong correlation (Pearson's r = 0.778, p < 0.001) between the metabolic savings achieved by each participant with the two controllers. Conclusion: The extended assistance duration provided by the implicitly synchronized SRC enabled greater metabolic reductions compared to the more targeted assistance of the explicitly synchronized HPT. Despite the different assistance profiles and metabolic outcomes, the correlation between the metabolic reductions with the two controllers suggests a difference in individual responsiveness to assistance, prompting more investigations to explore the person-specific factors affecting assistance receptivity.

FARMS: Framework for Animal and Robot Modeling and Simulation.

The study of animal locomotion and neuromechanical control offers valuable insights for advancing research in neuroscience, biomechanics, and robotics. We have developed FARMS (Framework for Animal and Robot Modeling and Simulation), an open-source, interdisciplinary framework, designed to facilitate access to neuromechanical simulations for modeling, simulation, and analysis of animal locomotion and bio-inspired robotic systems. By providing an accessible and user-friendly platform, FARMS aims to lower the barriers for researchers to explore the complex interactions between the nervous system, musculoskeletal structures, and their environment. Integrating the MuJoCo physics engine in a modular manner, FARMS enables realistic simulations and fosters collaboration among neuroscientists, biologists, and roboticists. FARMS has already been extensively used to study locomotion in animals such as mice, drosophila, fish, salamanders, and centipedes, serving as a platform to investigate the role of central pattern generators and sensory feedback. This article provides an overview of the FARMS framework, discusses its interdisciplinary approach, showcases its versatility through specific case studies, and highlights its effectiveness in advancing our understanding of locomotion. In particular, we show how we used FARMS to study amphibious locomotion by presenting experimental demonstrations across morphologies and environments based on neural controllers with central pattern generators and sensory feedback circuits models. Overall, the goal of FARMS is to contribute to a deeper understanding of animal locomotion, the development of innovative bio-inspired robotic systems, and promote accessibility in neuromechanical research.

The spinal cord facilitates cerebellar upper limb motor learning and control; inputs from neuromusculoskeletal simulation.

Complex interactions between brain regions and the spinal cord (SC) govern body motion, which is ultimately driven by muscle activation. Motor planning or learning are mainly conducted at higher brain regions, whilst the SC acts as a brain-muscle gateway and as a motor control centre providing fast reflexes and muscle activity regulation. Thus, higher brain areas need to cope with the SC as an inherent and evolutionary older part of the body dynamics. Here, we address the question of how SC dynamics affects motor learning within the cerebellum; in particular, does the SC facilitate cerebellar motor learning or constitute a biological constraint? We provide an exploratory framework by integrating biologically plausible cerebellar and SC computational models in a musculoskeletal upper limb control loop. The cerebellar model, equipped with the main form of cerebellar plasticity, provides motor adaptation; whilst the SC model implements stretch reflex and reciprocal inhibition between antagonist muscles. The resulting spino-cerebellar model is tested performing a set of upper limb motor tasks, including external perturbation studies. A cerebellar model, lacking the implemented SC model and directly controlling the simulated muscles, was also tested in the same. The performances of the spino-cerebellar and cerebellar models were then compared, thus allowing directly addressing the SC influence on cerebellar motor adaptation and learning, and on handling external motor perturbations. Performance was assessed in both joint and muscle space, and compared with kinematic and EMG recordings from healthy participants. The differences in cerebellar synaptic adaptation between both models were also studied. We conclude that the SC facilitates cerebellar motor learning; when the SC circuits are in the loop, faster convergence in motor learning is achieved with simpler cerebellar synaptic weight distributions. The SC is also found to improve robustness against external perturbations, by better reproducing and modulating muscle cocontraction patterns.

Animal robots in the African wilderness: Lessons learned and outlook for field robotics.

In early 2016, we had the opportunity to test a pair of sprawling posture robots, one designed to mimic a crocodile and another designed to mimic a monitor lizard, along the banks of the Nile River in Uganda, Africa. These robots were developed uniquely for a documentary by the BBC called Spy in the Wild and fell at the intersection of our interests in developing robots to study animals and robots for disaster response and other missions in challenging environments. The documentary required that these robots not only walk and swim in the same harsh, natural environments as the animals that they were modeled on and film up close but also move and even look exactly like the real animals from an aesthetic perspective. This pushed us to take a fundamentally different approach to the design and building of biorobots compared with our typical laboratory-residing robots, in addition to collaborating with sculpting artists to enhance our robots' aesthetics. The robots needed to be designed on the basis of a systematic study of data on the model specimens, be fabricated rapidly, and be reliable and robust enough to handle what the wild would throw at them. Here, we share the research efforts of this collaboration, the design specifications of the robots' hardware and software, the lessons learned from testing these robots in the field first hand, and how the eye-opening experience shaped our subsequent work on disaster response robotics and biorobotics for challenging amphibious scenarios.

Gait Phase Estimation in Steady Walking: A Comparative Study of Methods Based on the Phase Portrait of the Hip Angle.

Accurate real-time estimation of the gait phase (GP) is crucial for many control methods in exoskeletons and prostheses. A class of approaches to GP estimation construct the phase portrait of a segment or joint angle, and use the normalized polar angle of this diagram to estimate the GP. Although several studies have investigated such methods, quantitative information regarding their performance is sparse. In this work, we assess the performance of 3 portrait-based methods in flat and inclined steady walking conditions, using quantitative metrics of accuracy, repeatability and linearity. Two methods use portraits of the hip angle versus angular velocity (AVP), and hip angle versus integral of the angle (IAP). In a novel third method, a linear transformation is applied to the portrait to improve its circularity (CSP). An independent heel-strike (HS) detection algorithm is employed in all algorithms, rather than assuming HSs to occur at a constant point on the portrait. The novel method shows improvements in all metrics, notably significant root-mean-square error reductions compared to IAP (-3%, p < 0.001) and AVP (-2.4%, p < 0.001) in slope, and AVP (-1.61%, p = 0.0015) in flat walking. A non-negligible inter-subject variability is observed between phase angles at HS (equivalent to up to 8.4% of error in the GP), highlighting the importance of explicit HS detection for portrait-based methods.

Novel Design and Implementation of a Neuromuscular Controller on a Hip Exoskeleton for Partial Gait Assistance.

Exoskeletons intended for partial assistance of walking should be able to follow the gait pattern of their users, via online adaptive control strategies rather than imposing predefined kinetic or kinematic profiles. NeuroMuscular Controllers (NMCs) are adaptive strategies inspired by the neuromuscular modeling methods that seek to mimic and replicate the behavior of the human nervous system and skeletal muscles during gait. This study presents a novel design of a NMC, applied for the first time to partial assistance hip exoskeletons. Rather than the two-phase (stance/swing) division used in previous designs for the modulation of reflexes, a 5-state finite state machines is designed for gait phase synchronisation. The common virtual muscle model is also modified by assuming a stiff tendon, allowing for a more analytical computation approach for the muscle state resolution. As a first validation, the performance of the controller was tested with 9 healthy subjects walking at different speeds and slopes on a treadmill. The generated torque profiles show similarity to biological torques and optimal assistance profiles in the literature. Power output profiles of the exoskeleton indicate good synchronization with the users' intended movements, reflected in predominantly positive work by the assistance. The results also highlight the adaptability of the controller to different users and walking conditions, without the need for extensive parameter tuning.

State-Based Versus Time-Based Estimation of the Gait Phase for Hip Exoskeletons in Steady and Transient Walking.

The growing demand for online gait phase (GP) estimation, driven by advancements in exoskeletons and prostheses, has prompted numerous approaches in the literature. Some approaches explicitly use time, while others rely on state variables to estimate the GP. In this article, we study two novel GP estimation methods: a State-based Method (SM) which employs the phase portrait of the hip angle (similar to previous methods), but uses a stretching transformation to reduce the nonlinearity of the estimated GP; and a Time-based Method (TM) that utilizes feature recognition on the hip angle signal to update the estimated cadence twice per gait cycle. The methods were tested across various speeds and slopes, encompassing steady and transient walking conditions. The results demonstrated the ability of both methods to estimate the GP in a range of conditions. The TM outperformed the SM, exhibiting a root-mean-squared error below 3% compared to 8.5% for the SM. However, the TM exhibited diminished performance during speed transitions, whereas the SM performed consistently in steady and transient conditions. The SM displayed a better performance in inclined walking and demonstrated higher linearity at faster speeds. Through the assessment of these methods in diverse conditions, this study lays the groundwork for further advancements in GP estimation methods and their application in assistive controllers.

A spinal cord neuroprosthesis for locomotor deficits due to Parkinson's disease.

People with late-stage Parkinson's disease (PD) often suffer from debilitating locomotor deficits that are resistant to currently available therapies. To alleviate these deficits, we developed a neuroprosthesis operating in closed loop that targets the dorsal root entry zones innervating lumbosacral segments to reproduce the natural spatiotemporal activation of the lumbosacral spinal cord during walking. We first developed this neuroprosthesis in a non-human primate model that replicates locomotor deficits due to PD. This neuroprosthesis not only alleviated locomotor deficits but also restored skilled walking in this model. We then implanted the neuroprosthesis in a 62-year-old male with a 30-year history of PD who presented with severe gait impairments and frequent falls that were medically refractory to currently available therapies. We found that the neuroprosthesis interacted synergistically with deep brain stimulation of the subthalamic nucleus and dopaminergic replacement therapies to alleviate asymmetry and promote longer steps, improve balance and reduce freezing of gait. This neuroprosthesis opens new perspectives to reduce the severity of locomotor deficits in people with PD.

Investigating the roles of reflexes and central pattern generators in the control and modulation of human locomotion using a physiologically plausible neuromechanical model.

Objective.Studying the neural components regulating movement in human locomotion is obstructed by the inability to perform invasive experimental recording in the human neural circuits. Neuromechanical simulations can provide insights by modeling the locomotor circuits. Past neuromechanical models proposed control of locomotion either driven by central pattern generators (CPGs) with simple sensory commands or by a purely reflex-based network regulated by state-machine mechanisms, which activate and deactivate reflexes depending on the detected gait cycle phases. However, the physiological interpretation of these state machines remains unclear. Here, we present a physiologically plausible model to investigate spinal control and modulation of human locomotion.Approach.We propose a bio-inspired controller composed of two coupled CPGs that produce the rhythm and pattern, and a reflex-based network simulating low-level reflex pathways and Renshaw cells. This reflex network is based on leaky-integration neurons, and the whole system does not rely on changing reflex gains according to the gait cycle state. The musculoskeletal model is composed of a skeletal structure and nine muscles per leg generating movement in sagittal plane.Main results.Optimizing the open parameters for effort minimization and stability, human kinematics and muscle activation naturally emerged. Furthermore, when CPGs were not activated, periodic motion could not be achieved through optimization, suggesting the necessity of this component to generate rhythmic behavior without a state machine mechanism regulating reflex activation. The controller could reproduce ranges of speeds from 0.3 to 1.9 m s-1. The results showed that the net influence of feedback on motoneurons (MNs) during perturbed locomotion is predominantly inhibitory and that the CPGs provide the timing of MNs' activation by exciting or inhibiting muscles in specific gait phases.Significance.The proposed bio-inspired controller could contribute to our understanding of locomotor circuits of the intact spinal cord and could be used to study neuromotor disorders.

Identification of the trade-off between speed and efficiency in undulatory swimming using a bio-inspired robot.

Anguilliform swimmers, like eels or lampreys, are highly efficient swimmers. Key to understanding their performances is the relationship between the body's kinematics and resulting swimming speed and efficiency. But, we cannot prescribe kinematics to living fish, and it is challenging to measure their power consumption. Here, we characterise the swimming speed and cost of transport of a free-swimming undulatory bio-inspired robot as we vary its kinematic parameters, including joint amplitude, body wavelength, and frequency. We identify a trade-off between speed and efficiency. Speed, in terms of stride length, increases for increasing maximum tail angle, described by the newly proposed specific tail amplitude and reaches a maximum value around the specific tail amplitude of unity. Efficiency, in terms of the cost of transport, is affected by the whole-body motion. Cost of transport decreases for increasing travelling wave-like kinematics, and lower specific tail amplitudes. Our results suggest that live eels tend to choose efficiency over speed and provide insights into the key characteristics affecting undulatory swimming performance.

Integration of feedforward and feedback control in the neuromechanics of vertebrate locomotion: a review of experimental, simulation and robotic studies.

Animal locomotion is the result of complex and multi-layered interactions between the nervous system, the musculo-skeletal system and the environment. Decoding the underlying mechanisms requires an integrative approach. Comparative experimental biology has allowed researchers to study the underlying components and some of their interactions across diverse animals. These studies have shown that locomotor neural circuits are distributed in the spinal cord, the midbrain and higher brain regions in vertebrates. The spinal cord plays a key role in locomotor control because it contains central pattern generators (CPGs) - systems of coupled neuronal oscillators that provide coordinated rhythmic control of muscle activation that can be viewed as feedforward controllers - and multiple reflex loops that provide feedback mechanisms. These circuits are activated and modulated by descending pathways from the brain. The relative contributions of CPGs, feedback loops and descending modulation, and how these vary between species and locomotor conditions, remain poorly understood. Robots and neuromechanical simulations can complement experimental approaches by testing specific hypotheses and performing what-if scenarios. This Review will give an overview of key knowledge gained from comparative vertebrate experiments, and insights obtained from neuromechanical simulations and robotic approaches. We suggest that the roles of CPGs, feedback loops and descending modulation vary among animals depending on body size, intrinsic mechanical stability, time required to reach locomotor maturity and speed effects. We also hypothesize that distal joints rely more on feedback control compared with proximal joints. Finally, we highlight important opportunities to address fundamental biological questions through continued collaboration between experimentalists and engineers.

The neuromechanics of animal locomotion: From biology to robotics and back.

Robotics and neuroscience are sister disciplines that both aim to understand how agile, efficient, and robust locomotion can be achieved in autonomous agents. Robotics has already benefitted from neuromechanical principles discovered by investigating animals. These include the use of high-level commands to control low-level central pattern generator-like controllers, which, in turn, are informed by sensory feedback. Reciprocally, neuroscience has benefited from tools and intuitions in robotics to reveal how embodiment, physical interactions with the environment, and sensory feedback help sculpt animal behavior. We illustrate and discuss exemplar studies of this dialog between robotics and neuroscience. We also reveal how the increasing biorealism of simulations and robots is driving these two disciplines together, forging an integrative science of autonomous behavioral control with many exciting future opportunities.

NeuroMechFly, a neuromechanical model of adult Drosophila melanogaster.

Animal behavior emerges from an interaction between neural network dynamics, musculoskeletal properties and the physical environment. Accessing and understanding the interplay between these elements requires the development of integrative and morphologically realistic neuromechanical simulations. Here we present NeuroMechFly, a data-driven model of the widely studied organism, Drosophila melanogaster. NeuroMechFly combines four independent computational modules: a physics-based simulation environment, a biomechanical exoskeleton, muscle models and neural network controllers. To enable use cases, we first define the minimum degrees of freedom of the leg from real three-dimensional kinematic measurements during walking and grooming. Then, we show how, by replaying these behaviors in the simulator, one can predict otherwise unmeasured torques and contact forces. Finally, we leverage NeuroMechFly's full neuromechanical capacity to discover neural networks and muscle parameters that drive locomotor gaits optimized for speed and stability. Thus, NeuroMechFly can increase our understanding of how behaviors emerge from interactions between complex neuromechanical systems and their physical surroundings.

Investigation of neural and biomechanical impairments leading to pathological toe and heel gaits using neuromusculoskeletal modelling.

This study investigates the pathological toe and heel gaits seen in human locomotion using neuromusculoskeletal modelling and simulation. In particular, it aims to investigate potential cause-effect relationships between biomechanical or neural impairments and pathological gaits. Toe and heel gaits are commonly present in spinal cord injury, stroke and cerebral palsy. Toe walking is mainly attributed to spasticity and contracture at plantar flexor muscles, whereas heel walking can be attributed to muscle weakness of biomechanical or neural origin. To investigate the effect of these impairments on gait, this study focuses on the soleus and gastrocnemius muscles as they contribute to ankle plantarflexion. We built a reflex circuit model based on previous work by Geyer and Herr with additional pathways affecting the plantar flexor muscles. The SCONE software, which provides optimisation tools for 2D neuromechanical simulation of human locomotion, is used to optimise the corresponding reflex parameters and simulate healthy gait. We then modelled various bilateral plantar flexor biomechanical and neural impairments, and individually introduced them in the healthy model. We characterised the resulting simulated gaits as pathological or not by comparing ankle kinematics and ankle moment with the healthy optimised gait based on metrics used in clinical studies. Our simulations suggest that toe walking can be generated by hyperreflexia, whereas muscle and neural weaknesses partially induce heel gait. Thus, this 'what if' approach is deemed of great interest as it allows investigation of the effect of various impairments on gait and suggests an important contribution of active reflex mechanisms to pathological toe gait. KEY POINTS: Pathological toe and heel gaits are commonly present in various conditions such as spinal cord injury, stroke and cerebral palsy. These conditions present various neural and biomechanical impairments, but the cause-effect relationships between these impairments and pathological gaits are difficult to establish clinically. Based on neuromechanical simulation, this study focuses on the plantar flexor muscles and builds a new reflex circuit controller to model and evaluate the potential effect of both neural and biomechanical impairments on gait. Our results suggest an important contribution of active reflex mechanisms to pathological toe gait. This 'what if' based on neuromechanical modelling is thus deemed of great interest to target potential causes of pathological gait.

Combining Reflexes and External Sensory Information in a Neuromusculoskeletal Model to Control a Quadruped Robot.

This article examines the importance of integrating locomotion and cognitive information for achieving dynamic locomotion from a viewpoint combining biology and ecological psychology. We present a mammalian neuromusculoskeletal model from external sensory information processing to muscle activation, which includes: 1) a visual-attention control mechanism for controlling attention to external inputs; 2) object recognition representing the primary motor cortex; 3) a motor control model that determines motor commands traveling down the corticospinal and reticulospinal tracts; 4) a central pattern generation model representing pattern generation in the spinal cord; and 5) a muscle reflex model representing the muscle model and its reflex mechanism. The proposed model is able to generate the locomotion of a quadruped robot in flat and natural terrain. The experiment also shows the importance of a postural reflex mechanism when experiencing a sudden obstacle. We show the reflex mechanism when a sudden obstacle is separately detected from both external (retina) and internal (touching afferent) sensory information. We present the biological rationale for supporting the proposed model. Finally, we discuss future contributions, trends, and the importance of the proposed research.

Spontaneous Gait Transitions of Sprawling Quadruped Locomotion by Sensory-Driven Body-Limb Coordination Mechanisms.

Deciphering how quadrupeds coordinate their legs and other body parts, such as the trunk, head, and tail (i.e., body-limb coordination), can provide informative insights to improve legged robot mobility. In this study, we focused on sprawling locomotion of the salamander and aimed to understand the body-limb coordination mechanisms through mathematical modeling and simulations. The salamander is an amphibian that moves on the ground by coordinating the four legs with lateral body bending. It uses standing and traveling waves of lateral bending that depend on the velocity and stepping gait. However, the body-limb coordination mechanisms responsible for this flexible gait transition remain elusive. This paper presents a central-pattern-generator-based model to reproduce spontaneous gait transitions, including changes in bending patterns. The proposed model implements four feedback rules (feedback from limb-to-limb, limb-to-body, body-to-limb, and body-to-body) without assuming any inter-oscillator coupling. The interplay of the feedback rules establishes a self-organized body-limb coordination that enables the reproduction of the speed-dependent gait transitions of salamanders, as well as various gait patterns observed in sprawling quadruped animals. This suggests that sensory feedback plays an essential role in flexible body-limb coordination during sprawling quadruped locomotion.

Emergence of robust self-organized undulatory swimming based on local hydrodynamic force sensing.

Undulatory swimming represents an ideal behavior to investigate locomotion control and the role of the underlying central and peripheral components in the spinal cord. Many vertebrate swimmers have central pattern generators and local pressure-sensitive receptors that provide information about the surrounding fluid. However, it remains difficult to study experimentally how these sensors influence motor commands in these animals. Here, using a specifically designed robot that captures the essential components of the animal neuromechanical system and using simulations, we tested the hypothesis that sensed hydrodynamic pressure forces can entrain body actuation through local feedback loops. We found evidence that this peripheral mechanism leads to self-organized undulatory swimming by providing intersegmental coordination and body oscillations. Swimming can be redundantly induced by central mechanisms, and we show that, therefore, a combination of both central and peripheral mechanisms offers a higher robustness against neural disruptions than any of them alone, which potentially explains how some vertebrates retain locomotor capabilities after spinal cord lesions. These results broaden our understanding of animal locomotion and expand our knowledge for the design of robust and modular robots that physically interact with the environment.