Central Pattern Generator Network Research Articles

Structural Design and Control Research of Multi-Segmented Biomimetic Millipede Robot

Due to their advantages of good stability, adaptability, and flexibility, multi-legged robots are increasingly important in fields such as rescue, military, and healthcare. This study focuses on the millipede, a multi-segmented organism, and designs a novel multi-segment biomimetic robot based on an in-depth investigation of the millipede’s biological characteristics and locomotion mechanisms. Key leg joints of millipede locomotion are targeted, and a mathematical model of the biomimetic robot’s leg joint structure is established for kinematic analysis. Furthermore, a central pattern generator (CPG) control strategy is studied for multi-jointed biomimetic millipede robots. Inspired by the millipede’s neural system, a simplified single-loop CPG network model is constructed, reducing the number of oscillators from 48 to 16. Experimental trials are conducted using a prototype to test walking in a wave-like gait, walking with a leg removed, and walking on complex terrain. The results demonstrate that under CPG waveform input conditions, the robot can walk stably, and the impact of a leg failure on overall locomotion is acceptable, with minimal speed loss observed when walking on complex terrain. The research on the structure and motion control algorithms of multi-jointed biomimetic robots lays a technical foundation, expanding their potential applications in exploring unknown environments, rescue missions, agriculture, and other fields.

Bionic fish position control with the desired heading angle of the target position

This paper proposes an underdriven bionic fish position control system with the desired heading angle at the target position. The developed position control system controls the bionic fish to reach the desired target position while maintaining the heading angle at the target position. The system comprises a virtual waypoint-based line-of-sight (VW-LOS) guidance system, an inner-loop control system, and a linear extended state observer (LESO). First, the VW-LOS system generates a virtual waypoint behind the target position, transforming the bionic fish's position control problem into a straight-line path tracking problem. The heading and speed guidance laws are used to generate the desired heading angle and speed required during the tracking process. An inner-loop controller is subsequently used to regulate the controlled reference, including a heading tracking controller, a speed tracking controller, and a fuzzy inference module that converts forces and moments into inputs to the bionic fish's central pattern generator (CPG) network. Moreover, the LESO estimates the unmeasurable speed and sideslip angle and feeds back to the VW-LOS system and the inner-loop controller. Finally, simulation and experimental results verify the effectiveness of the proposed control system.

The synaptic drive of central pattern-generating networks to leg motor neurons of a walking insect is motor neuron pool specific.

Rhythmic locomotor activity, such as flying, swimming, or walking, results from an interplay between higher-order centers in the central nervous system, which initiate, maintain, and modify task-specific motor activity, downstream central pattern-generating neural circuits (CPGs) that can generate a default rhythmic motor output, and, finally, feedback from sense organs that modify basic motor activity toward functionality.1,2,3 In this context, CPGs provide phasic synaptic drive to motor neurons (MNs) and thereby support the generation of rhythmic activity for locomotion. We analyzed the synaptic drive that the leg MNs supplying the three main leg joints receive from CPGs in pharmacologically activated and deafferented preparations of the stick insect (Carausius morosus). We show that premotor CPGs pattern the tonic activity of five of the six leg MN pools by phasic inhibitory synaptic drive. These are the antagonistic MN pools supplying the thoraco-coxal joint and the femur-tibial joint4,5 and the levator MN pool supplying the coxa-trochanteral (CTr) joint. In contrast, rhythmic activity of the depressor MN pool supplying the CTr joint was found to be primarily based on a phasic excitatory drive. This difference is likely related to the pivotal role of the depressor muscle in generating leg stance during any walking situation. Thus, our results provide evidence for qualitatively differing mechanisms to generate rhythmic activity between MN pools in the same locomotor system.

Immature functional development of lumbar locomotor networks in adult Irf8-/- mice.

To date, research on the role of the brainstem and spinal cord in motor behavior has relied on in vitro preparations of the neonatal rodent spinal cord, with or without the brainstem; their spatial and temporal scope are subject to technical limitations imposed by low oxygen tension in deep tissues. Therefore, we created an arterially perfused in situ preparation that allowed us to investigate functional interactions in the CNS from the neonatal to adult period. Decerebrated rodents were kept alive via total artificial cardiopulmonary bypass for extracorporeal circulation; the plasma oxygen and ion components needed for survival were supplied through the blood vessels. Interferon regulatory factor 8 (IRF8) is a transcription factor that promotes myeloid cell development and stimulates innate immune responses. In the brain, IRF8 is expressed only in microglia and directs the expression of many genes that serve microglial functions. Recent evidence indicates that IRF8 affects behavior and modulates Alzheimer's disease progression in a mouse model. However, whether this immune deficiency arising from the absence of IRF8 influences the development of the neuronal network in the spinal cord is unknown. We applied the above methodology to mice of all ages and electrophysiologically explored whether the absence of IRF8 influences the development of lumbar central pattern generator (CPG) networks. In mice of all ages, bilateral neuronal discharges by the normal CPG networks activated by the modulated sympathetic tone via descending pathways at high flow rates became organized into discharge episodes punctuated by periods of quiescence. Similar discharge episodes were generated by the adult CPG networks (≥P14 days) activated by drug application. However, discharge episodes elicited by activating the neonatal-juvenile CPG networks (<P14 days) occurred alternately on the left and right sides. Interestingly, discharge episodes elicited by the CPG networks in adult IRF8 knockout mice (P11-12 weeks) consisted of those elicited by the CPG networks of both periods. Thus, it was suggested that growing up with immunodeficiency due to loss of IRF8 might interfere with the normal development of functions exerted by the lumbar CPG network because IRF8 plays a role in the normal development of the lumbar CPG network.

Periodic self-triggered strategy for highly non-linear hybrid stochastic delayed networks based on impulsive control

This article investigates the stabilization issue of highly non-linear hybrid stochastic delayed networks (HSDNs) via periodic self-triggered control under impulse (PS-TCI). Firstly, the existence of a unique global solution for highly non-linear HSDNs under PS-TCI is studied. Then, a stabilization criterion for highly non-linear HSDNs is established, by combining a graph-theoretic approach with a novel Lyapunov-based analysis, based on a ‘genuine’ Lyapunov function defined by introducing an auxiliary timer. Therein, the less conservative polynomial growth condition and local Lipschitz condition for the drift and diffusion coefficients are used than the linear growth condition and global Lipschitz condition. Meanwhile, the design idea of PS-TCI is based on the evolution of an upper bound of the mathematical expectation for Lyapunov function (not directly Lyapunov function or system state), which implies that the triggered instant of PS-TCI is not a random variable. Finally the theoretical results are employed to study the stability of a class of FitzHugh–Nagumo circuits networks and the central pattern generators networks of a hexapod robot, and correlative numerical simulations are provided for demonstration.

Adaptive walking control for quadruped robot by using oscillation patterns.

To improve the adaptability of quadruped robot in multiple scenarios, an adaptive locomotive system based on the double-layered central pattern generator (CPG) is proposed. The novel CPG network consists of double master units and subsets of slave units based on gyroscope signals including yaw and pitch angle. The response of master units provides the ability to control the 1st joins of quadruped robot, while slave units can generate the symmetry signals to control the 2nd and 3rd joints. The CPG network enables the seamless switching of locomotion gaits to stops and starts by using an ultrasonic sensor. Through adjusting the mutually dependent parameters, joints can generate the joint angles to achieve steering behavior. For adaptive movement on an irregular surface, stable ranges of the robot body yaw and pitch angles are proposed by using gyroscope signals. The experimental results verify that the quadruped robot with the proposed double-layered CPG network can perform stable trot pattern in a complex environment.

Vortex phase matching of a self-propelled model of fish with autonomous fin motion

It has been a long-standing problem how schooling fish optimize their motion by exploiting the vortices shed by the others. A recent experimental study showed that a pair of fish reduce energy consumption by matching the phases of their tailbeat according to their distance. In order to elucidate the dynamical mechanism by which fish control the motion of caudal fins via vortex-mediated hydrodynamic interactions, we introduce a new model of a self-propelled swimmer with an active flapping plate. The model incorporates the role of the central pattern generator network that generates rhythmic but noisy activity of the caudal muscle, in addition to hydrodynamic and elastic torques on the fin. For a solitary fish, the model reproduces a linear relation between the swimming speed and tailbeat frequency, as well as the distributions of the speed, tailbeat amplitude, and frequency. For a pair of fish, both the distribution function and energy dissipation rate exhibit periodic patterns as functions of the front-back distance and phase difference of the flapping motion. We show that a pair of fish spontaneously adjust their distance and phase difference via hydrodynamic interaction to reduce energy consumption.

Gap junctions desynchronize a neural circuit to stabilize insect flight

Insect asynchronous flight is one of the most prevalent forms of animal locomotion used by more than 600,000 species. Despite profound insights into the motor patterns1, biomechanics2,3 and aerodynamics underlying asynchronous flight4,5, the architecture and function of the central-pattern-generating (CPG) neural network remain unclear. Here, on the basis of an experiment–theory approach including electrophysiology, optophysiology, Drosophila genetics and mathematical modelling, we identify a miniaturized circuit solution with unexpected properties. The CPG network consists of motoneurons interconnected by electrical synapses that, in contrast to doctrine, produce network activity splayed out in time instead of synchronized across neurons. Experimental and mathematical evidence support a generic mechanism for network desynchronization that relies on weak electrical synapses and specific excitability dynamics of the coupled neurons. In small networks, electrical synapses can synchronize or desynchronize network activity, depending on the neuron-intrinsic dynamics and ion channel composition. In the asynchronous flight CPG, this mechanism translates unpatterned premotor input into stereotyped neuronal firing with fixed sequences of cell activation that ensure stable wingbeat power and, as we show, is conserved across multiple species. Our findings prove a wider functional versatility of electrical synapses in the dynamic control of neural circuits and highlight the relevance of detecting electrical synapses in connectomics.

Parameter search of a CPG network using a genetic algorithm for a snake robot with tactile sensors moving on a soft floor.

When a snake robot explores a collapsed house as a rescue robot, it needs to move through various obstacles, some of which may be made of soft materials, such as mattresses. In this study, we call mattress-like environment as a soft floor, which deforms when some force is added to it. We focused on the central pattern generator (CPG) network as a control for the snake robot to propel itself on the soft floor and constructed a CPG network that feeds back contact information between the robot and the floor. A genetic algorithm was used to determine the parameters of the CPG network suitable for the soft floor. To verify the obtained parameters, comparative simulations were conducted using the parameters obtained for the soft and hard floor, and the parameters were confirmed to be appropriate for each environment. By observing the difference in snake robot's propulsion depending on the presence or absence of the tactile sensor feedback signal, we confirmed the effectiveness of the tactile sensor considered in the parameter search.

A Novel Double-Layered Central Pattern Generator-Based Motion Controller for the Hexapod Robot

To implement the various movement control of the hexapod robot, a motion controller based on the double-layered central pattern generator (CPG) is proposed in this paper. The novel CPG network is composed of a rhythm layer and a pattern layer. The CPG neurons are constructed based on Kuramoto nonlinear oscillator. The parameters including the frequency, coupling strength, and phase difference matrix of the CPG network for four typical gaits are planned. The mapping relationship between the signals of the CPG network and the joint trajectories of the hexapod robot is designed. The co-simulations and experiments have been conducted to verify the feasibility of the proposed CPG-based controller. The actual average velocities of the wave gait, the tetrapod gait, the tripod gait, and the self-turning gait are 10.8 mm/s, 25.5 mm/s, 37.8 mm/s and 26°/s, respectively. The results verify that the hexapod robot with the proposed double-layered CPG-based controller can perform stable and various movements.

Central pattern generator network model for the alternating hind limb gait of rats based on the modified Van der Pol equation.

Herein, we employed a central pattern generator (CPG), a spinal cord neural network that regulates lower-limb gait during intra-spinal micro-stimulation (ISMS). Particularly, ISMS was used to determine the spatial distribution pattern of CPG sites in the spinal cord and the signal regulation pattern that induced the CPG network to produce coordinated actions. Based on the oscillation phenomenon of the single CPG neurons of Van der Pol (VDP) oscillators, a double-cell CPG neural network model was constructed to realise double lower limbs, six-joint modelling, the simulation of 12 neural circuits, the CPG loci characterising stimuli-inducing alternating movements and changes in polarity stimulation signals in rat hindlimbs, and leg-state change movements. The feasibility and effectiveness of the CPG neural network were verified by recording the electromyographic burst-release mode of the flexor and extensor muscles of the knee joints during CPG electrical stimulation. The results revealed that the output pattern of the CPG presented stable rhythm and coordination characteristics. The 12-neuron CPG model based on the improved VDP equation realised single-point control while significantly reducing the number of stimulation electrodes in the gait training of spinal cord injury patients. We believe that this study advances motor function recovery in rehabilitation medicine.

NA-CPG: A robust and stable rhythm generator for robot motion control

Central pattern generators (CPGs) have been widely applied in robot motion control for the spontaneous output of coherent periodic rhythms. However, the underlying CPG network exhibits good convergence performance only within a certain range of parameter spaces, and the coupling of oscillators affects the network output accuracy in complex topological relationships. Moreover, CPGs may diverge when parameters change drastically, and the divergence is irreversible, which is catastrophic for the control of robot motion. Therefore, normalized asymmetric CPGs (NA-CPGs) that normalize the amplitude parameters of Hopf-based CPGs and add a constraint function and a frequency regulation mechanism are proposed. NA-CPGs can realize parameter decoupling, precise amplitude output, and stable and rapid convergence, as well as asymmetric output waveforms. Thus, it can effectively cope with large parameter changes to avoid network oscillations and divergence. To optimize the parameters of the NA-CPG model, a reinforcement-learning-based online optimization method is further proposed. Meanwhile, a biomimetic robotic fish is illustrated to realize the whole optimization process. Simulations demonstrated that the designed NA-CPGs exhibit stable, secure, and accurate network outputs, and the proposed optimization method effectively improves the swimming speed and reduces the lateral swing of the multijoint robotic fish by 6.7% and 41.7%, respectively. The proposed approach provides a significant improvement in CPG research and can be widely employed in the field of robot motion control.

Neurogenic mechanisms for locomotor-respiratory coordination in mammals.

Central motor rhythm-generating networks controlling different functions are generally considered to operate mostly independently from one another, each controlling the specific behavioral task to which it is assigned. However, under certain physiological circumstances, central pattern generators (CPGs) can exhibit strong uni- or bidirectional interactions that render them closely inter-dependent. One of the best illustrations of such an inter-CPG interaction is the functional relationship that may occur between rhythmic locomotor and respiratory functions. It is well known that in vertebrates, lung ventilatory rates accelerate at the onset of physical exercise in order to satisfy the accompanying rapid increase in metabolism. Part of this acceleration is sustained by a coupling between locomotion and ventilation, which most often results in a periodic drive of the respiratory cycle by the locomotor rhythm. In terrestrial vertebrates, the likely physiological significance of this coordination is that it serves to reduce the mechanical interference between the two motor systems, thereby producing an energetic benefit and ultimately, enabling sustained aerobic activity. Several decades of studies have shown that locomotor-respiratory coupling is present in most species, independent of the mode of locomotion employed. The present article aims to review and discuss mechanisms engaged in shaping locomotor-respiratory coupling (LRC), with an emphasis on the role of sensory feedback inputs, the direct influences between CPG networks themselves, and finally on spinal cellular candidates that are potentially involved in the coupling of these two vital motor functions.

Embedded Real-Time Implementation of Bio-Inspired Central Pattern Generator with Self-Repairing Function

Both robustness and self-repairing of the rhythmic behaviors generated by central pattern generators (CPGs) play significant roles in locomotion control. Although current CPG models have been established to mimic rhythmic outputs, the mechanisms by which the self-repairing capacities of CPG systems are formed are largely unknown. In this paper, a novel bio-inspired self-repairing CPG model (BiSRP-CPG) is proposed based on the tripartite synapse, which reveals the critical role of astrocytes in the dynamic coordination of CPGs. BiSRP-CPG is implemented on the parallel FPGA platform to simulate CPG systems on real physiological scale, in which a hardware implementation method without multiplier is utilized to break the limitation of FPGA hardware resources. The experimental results verified both the robustness and self-repairing capabilities of rhythm of BiSRP-CPG in the presence of stochastic synaptic inputs and “faulty” synapse. Under the synaptic failure rate of 20%, BiSRP-CPG suffered only 10.53% performance degradation, which was much lower than the 36.84% spike loss rate of CPG networks without astrocytes. This paper provides an insight into one of the possible self-repair mechanisms of brain rhythms which can be utilized to develop autonomously fault-tolerant electronic systems.

Similarity Evaluation Rule and Motion Posture Optimization for a Manta Ray Robot

The current development of manta ray robots is usually based on functional bionics, and there is a lack of bionic research to enhance the similarity of motion posture. To better exploit the characteristics of bionic, a similarity evaluation rule is constructed herein by a Dynamic Time Warping (DTW) algorithm to guide the optimization of the control parameters of a manta ray robot. The Central Pattern Generator (CPG) network with time and space asymmetry oscillation characteristics is improved to generate coordinated motion control signals for the robot. To optimize similarity, the CPG network is optimized with the genetic algorithm and particle swarm optimization (GAPSO) to solve the problems of multiple parameters, high non-linearity, and uncertain parameter coupling in the CPG network. The experimental results indicate that the similarity between the forward motion pose of the optimized manta ray robot and the manta ray is improved to 88.53%.

The effects of temperature changes on the mammalian spinal locomotor CPG network and the role of TRP channels in functional compensation

Preliminary data suggests that spinal cord neural networks that contain intact sensory pathways, which include dorsal root ganglia (DRG) and thermal transient receptor potential (TRP) channels are more resilient to gradual temperature changes than isolated preparations. Moreover, how acute temperature changes affect mammalian CPG‐driven motor activity remains uncertain. We studied how acute and gradual temperature changes impact the stability of motor networks with or without sensory pathways. More specifically, we aim to investigate: 1. the difference between how acute temperature changes versus gradual temperature changes can have an impact on the stability and robustness of motor networks; 2. how sensory inputs and TRP channels affect the adaptation of spinal central pattern generator (CPG) networks to changes in temperature. We compared changes in motor neuron bursting activity between ventral L2 and L5 roots of neonatal mouse spinal cord preparations under acute (from 20℃ directly to 29℃) and gradual (gradually from 20℃, to 23℃, to 26℃, and to 29℃) temperature change conditions. Spinal cord preparations both with and without DRGs were compared. We found that the locomotor activities were increased as temperature increase under both chronic and acute changes; however, spinal cord preparations without intact DRG showed activities with less stable phasing and more overall motor bursting variability compared to those preparations with intact DRGs (on L3, L4, and L5 spinal levels). Our results suggest that the lumbar spinal locomotor CPG network activity shows more susceptibility and less adaptation to temperature changes in the absence of sensory feedback provided by DRG sensory afferents. Further studies will be conducted in the presence of TRP channel blockers to preparations with DRGs to confirm the potential compensatory role that sensory feedback afferents convey to spinal CPG networks.

Locomotor Pattern and Force Generation Modulated by Ionic Channels: A Computational Study of Spinal Networks Underlying Locomotion.

Locomotion is a fundamental movement in vertebrates produced by spinal networks known as central pattern generators (CPG). During fictive locomotion cat lumbar motoneurons (MNs) exhibit changes in membrane properties, including hyperpolarization of voltage threshold, reduction of afterhyperpolarization and input resistance, and amplification of nonlinear membrane properties. Both modeling and electrophysiological studies suggest that these changes can be produced by upregulating voltage-gated sodium channel (VGSC), persistent sodium (NaP), or L-type calcium channel (LTCC) or downregulating delayed-rectifier potassium (K(DR)) or calcium-dependent potassium channel (KCa) in spinal MNs. Further studies implicate that these channel modulations increase motor output and facilitate MN recruitment. However, it remains unknown how the channel modulation of CPG networks or MN pools affects the rhythmic generation of locomotion and force production of skeletal muscle during locomotion. In order to investigate this issue, we built a two-level CPG model composed of excitatory interneuron pools (Exc-INs), coupled reciprocally with inhibitory interneuron pools (Inh-INs), and projected to the flexor-extensor MN pools innervating skeletal muscles. Each pool consisted of 100 neurons with membrane properties based on cat spinal neurons. VGSC, K(DR), NaP, KCa, LTCC, and H-current channels were included in the model. Simulation results showed that (1) upregulating VGSC, NaP, or LTCC or downregulating KCa in MNs increased discharge rate and recruitment of MNs, thus facilitating locomotor pattern formation, increased amplitude of electroneurogram (ENG) bursting, and enhanced force generation of skeletal muscles. (2) The same channel modulation in Exc-INs increased the firing frequency of the Exc-INs, facilitated rhythmic generation, and increased flexor-extensor durations of step cycles. (3) Contrarily, downregulation of NaP or LTCC in MNs or Exc-INs or both CPG (Exc-INs and Inh-INs) and MNs disrupted locomotor pattern and reduced or even blocked the ENG bursting of MNs and force generation of skeletal muscles. (4) Pharmacological experiments showed that bath application of 25 μM nimodipine or 2 μM riluzole completely blocked fictive locomotion in isolated rat spinal cord, consistent with simulation results. We concluded that upregulation of VGSC, NaP, or LTCC or downregulation of KCa facilitated rhythmic generation and force production during walking, with NaP and LTCC playing an essential role.

Localization of muscarinic acetylcholine receptor-dependent rhythm-generating modules in the Drosophila larval locomotor network.

Mechanisms of rhythm generation have been extensively studied in motor systems that control locomotion over terrain in limbed animals; however, much less is known about rhythm generation in soft-bodied terrestrial animals. Here we explored how muscarinic acetylcholine receptor (mAChR)-modulated rhythm-generating networks are distributed in the central nervous system (CNS) of soft-bodied Drosophila larvae. We measured fictive motor patterns in isolated CNS preparations, using a combination of Ca2+ imaging and electrophysiology while manipulating mAChR signaling pharmacologically. Bath application of the mAChR agonist oxotremorine potentiated bilaterally asymmetric activity in anterior thoracic regions and promoted bursting in posterior abdominal regions. Application of the mAChR antagonist scopolamine suppressed rhythm generation in these regions and blocked the effects of oxotremorine. Oxotremorine triggered fictive forward crawling in preparations without brain lobes. Oxotremorine also potentiated rhythmic activity in isolated posterior abdominal CNS segments as well as isolated anterior brain and thoracic regions, but it did not induce rhythmic activity in isolated anterior abdominal segments. Bath application of scopolamine to reduced preparations lowered baseline Ca2+ levels and abolished rhythmic activity. Overall, these results suggest that mAChR signaling plays a role in enabling rhythm generation at multiple sites in the larval CNS. This work furthers our understanding of motor control in soft-bodied locomotion and provides a foundation for study of rhythm-generating networks in an emerging genetically tractable locomotor system.NEW & NOTEWORTHY Using a combination of pharmacology, electrophysiology, and Ca2+ imaging, we find that signaling through mACh receptors plays a critical role in rhythmogenesis in different regions of the Drosophila larval CNS. mAChR-dependent rhythm generators reside in distal regions of the larval CNS and provide functional substrates for central pattern-generating networks (CPGs) underlying headsweep behavior and forward locomotion. This provides new insights into locomotor CPG operation in soft-bodied animals that navigate over terrain.

Hop Mice Display Synchronous Hindlimb Locomotion and a Ventrally Fused Lumbar Spinal Cord Caused by a Point Mutation in Ttc26.

Identifying the spinal circuits controlling locomotion is critical for unravelling the mechanisms controlling the production of gaits. Development of the circuits governing left-right coordination relies on axon guidance molecules such as ephrins and netrins. To date, no other class of proteins have been shown to play a role during this process. Here, we have analyzed hop mice, which walk with a characteristic hopping gait using their hindlimbs in synchrony. Fictive locomotion experiments suggest that a local defect in the ventral spinal cord contributes to the aberrant locomotor phenotype. Hop mutant spinal cords had severe morphologic defects, including the absence of the ventral midline and a poorly defined border between white and gray matter. The hop mice represent the first model where, exclusively found in the lumbar domain, the left and right components of the central pattern generators (CPGs) are fused with a synchronous hindlimb gait as a functional consequence. These defects were associated with abnormal developmental processes, including a misplaced notochord and reduced induction of ventral progenitor domains. Whereas the underlying mutation in hop mice has been suggested to lie within the Ttc26 gene, other genes in close vicinity have been associated with gait defects. Mouse embryos carrying a CRISPR replicated point mutation within Ttc26 displayed an identical morphologic phenotype. Thus, our data suggest that the assembly of the lumbar CPG network is dependent on fully functional TTC26 protein.

A CPG-based gait planning and motion performance analysis for quadruped robot

PurposeTo achieve stable gait planning and enhance the motion performance of quadruped robot, this paper aims to propose a motion control strategy based on central pattern generator (CPG) and back-propagation neural network (BPNN).Design/methodology/approachFirst, the Kuramoto phase oscillator is used to construct the CPG network model, and a piecewise continuous phase difference matrix is designed to optimize the duty cycle of walk gait, so as to realize the gait planning and smooth switching. Second, the mapper between CPG output and joint drive is established based on BP neural network, so that the quadruped robot based on CPG control has better foot trajectory to enhance the motion performance. Finally, to obtain better mapping effect, an evaluation function is resigned to evaluate the proximity between the actual foot trajectory and the ideal foot trajectory. Genetic algorithm and particle swarm optimization are used to optimize the initial weights and thresholds of BPNN to obtain more accurate foot trajectory.FindingsThe method provides a solution for the smooth gait switching and foot trajectory of the robot. The quintic polynomial trajectory is selected to testify the validity and practicability of the method through simulation and prototype experiment.Originality/valueThe paper solved the incorrect duty cycle under the walk gait of CPG network constructed by Kuramoto phase oscillator, and made the robot have a better foot trajectory by mapper to enhance its motion performance.