Quadruped robot motion control method and system based on spine and leg cooperation

Abstract

The invention discloses a quadruped robot motion control method and system based on spine and leg coordination, and the method comprises the steps: building a coupling relation model of spine torsion and four-limb swing, so as to describe a mathematical mapping relation between spine joint angles and four-limb joint angles; a spine-leg cooperative controller of a central mode generator (CPG) is formed by a plurality of nonlinear oscillators, and the nonlinear oscillators are mutually coupled to form a complete oscillator network so as to realize phase synchronization and amplitude coordination of spine torsion and four-limb swing; a self-adaptive coupling strength adjusting mechanism is constructed, and the coupling strength between the spine and the legs is dynamically adjusted according to external sensor information and the internal state; according to the current motion state and the environment information, coordinate control signals of spine joints and limb joints are generated, and the robot is driven to execute corresponding motion. Through the bionic spine-limb dynamic coupling and self-adaptive adjusting mechanism, the motion efficiency, stability and task adaptability of the robot in the unstructured environment are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and in particular to a method and system for motion control of quadruped robots based on the coordination of the spine and legs. Background Technology

[0002] Current research on quadruped robots mainly focuses on the independent control of leg joints and generally adopts a rigid torso design, neglecting the core role of the spine in movement.

[0003] Traditional rigid torso designs result in three major technical limitations for robots: First, low energy efficiency, with studies showing that vibration losses in rigid structures account for 35%-40% of motion energy consumption; second, insufficient environmental adaptability, as fixed geometric configurations struggle to cope with complex terrains such as slope variations and soft ground; and third, limited movement patterns, unable to achieve complex actions unique to living organisms, such as high-speed running (>5m / s) and sharp turns (<0.5s turning). While Boston Dynamics' Spot series robots have improved load capacity through hydraulic actuation, their spine remains a passive elastic element, lacking active adjustment capabilities.

[0004] In motion control modeling, existing research mostly employs simplified dynamic models. Honda's ASIMO robot uses ZMP (Zero Moment Point) theory for gait planning, but this model cannot describe the coupling between the spine and limbs, resulting in a 15%-20% error between simulation and actual motion. While the hierarchical control architecture proposed by the EU's RoboCup consortium achieves multi-joint coordination, it fails to establish a dynamic mapping relationship between spinal joints and end effectors. The domestic UBTECH Walker robot uses a pre-set gait library strategy, requiring frequent switching of motion modes when facing unknown terrain, resulting in a response delay of over 0.3 seconds.

[0005] Existing control algorithms have fundamental flaws in adaptive adjustment. The biomimetic robot developed by the Korea Advanced Institute of Science and Technology (KAIST) adapts to different terrains by using a pre-set coupling parameter table, but the parameter update cycle is as long as 2 seconds, making it difficult to meet the demands of dynamic environments. While the MIT Cheetah robot incorporates reinforcement learning to optimize gait, its spine control still uses open-loop PID regulation, resulting in problems such as large overshoot (peak value up to 12°) and slow convergence speed (>5 gait cycles). The RHex robot from the University of California, Berkeley, uses a distributed control architecture, but the phase synchronization error between the spine and limbs exceeds 8°, leading to an 18% increase in energy loss.

[0006] The aforementioned technical bottlenecks have severely hampered the practical application of quadruped robots. Summary of the Invention

[0007] The purpose of this invention is to provide a method and system for motion control of quadruped robots based on the coordination of the spine and legs. By establishing a spine-leg coordinated control system, it breaks through the traditional rigid design paradigm and provides a new technical path for improving the motion performance of bionic robots.

[0008] The technical solution adopted in this invention is:

[0009] A quadruped robot motion control method based on spine and leg coordination includes the following steps:

[0010] a) Establish a coupling model between spinal torsion and limb swing to describe the mathematical mapping relationship between spinal joint angles and limb joint angles;

[0011] b) A spine-leg coordination controller consisting of a central mode generator (CPG) composed of multiple nonlinear oscillators. The nonlinear oscillators are coupled to form a complete oscillator network to achieve phase synchronization and amplitude coordination between spinal torsion and limb swing.

[0012] c) Construct an adaptive coupling strength adjustment mechanism to dynamically adjust the coupling strength between the spine and legs based on external sensor information and internal state;

[0013] d) Generate coordinated control signals for the spinal and limb joints based on the current motion state and environmental information, and drive the robot to perform corresponding movements.

[0014] Furthermore, the coupling model between spinal torsion and limb swing includes:

[0015] Forelimb coupling relationship:

[0016] Hindlimb coupling relationship:

[0017] Where θforelimb is the forelimb swing angle, θhindlimb is the hindlimb swing angle, θspine is the spinal torsion angle, and K... f and K h Let φf and φf be the coupling coefficients of the forelimb and hindlimb, respectively. These represent the phase differences between the forelimbs and hindlimbs and the spine, respectively.

[0018] Furthermore, the spine-leg co-controller based on the central pattern generator (CPG) consists of multiple nonlinear oscillators that are coupled together to form a complete oscillator network. The dynamic equation of each oscillator is as follows:

[0019]

[0020] Where x and y are the coordinate positions of the oscillator in the two-dimensional phase space, and r 2=x 2 +y 2 , represents the square of the distance between the oscillator and the origin; α controls the speed of convergence to the limit cycle (unit: 1 / s, the larger the value, the faster the convergence, typical value is 1-10), μ controls the radius of the limit cycle, and ω is the oscillation frequency (unit: rad / s, determines the oscillation period, typical value is 0.5-5.0).

[0021] Furthermore, the coupling between the oscillators is achieved through the following equation:

[0022]

[0023] Where, the subscript i represents the i-th oscillator (i = 1, 2, ..., n, where n is the total number of oscillators), the subscript j represents the j-th oscillator connected to the i-th oscillator (j ∈ [1, n] and j ≠ i), w ij This is the coupling weight from oscillator j to oscillator i (a dimensionless parameter representing the strength of the interaction between oscillators, typically ranging from 0.1 to 1.0). It is the desired phase difference (unit: radians, representing the desired phase offset between two oscillators, used to achieve different gait modes).

[0024] Furthermore, the specific steps for adaptive coupling strength adjustment in step c) are as follows:

[0025] S1, by sensing the environment and the robot's state;

[0026] S2, evaluate current performance metrics;

[0027] S3: Determine whether the coupling strength needs to be adjusted based on the performance indicators; if so, update the coupling parameters and execute S4; otherwise, execute S1.

[0028] S4, based on the updated coupling parameters, updates the control signal and adds the empirical storage for parameter optimization based on reinforcement learning, then executes S1.

[0029] Furthermore, the adaptive coupling strength adjustment mechanism updates the coupling parameters based on the following adaptive law:

[0030]

[0031] in, and K represents the time rate of change of the coupling coefficients of the forelimb and hindlimb, respectively. f and K h Let E be the coupling coefficients of the forelimb and hindlimb, respectively, and ηf and ηh be the learning rates. target For the target performance index, E actual For actual performance indicators, and These are the gradient symbols for the coupling coefficients of performance E with respect to the forelimb and hindlimb, respectively.

[0032] Furthermore, the performance index E is a multi-objective function:

[0033] E = w1 * E energy +w2*E stability +w3*E speed

[0034] Where E is the comprehensive performance index, E energy E is an energy efficiency indicator. stability E is an indicator of motion stability. speed For speed indicators, w1, w2, and w3 are weighting coefficients used to adjust the importance of different performance indicators, satisfying w1 + w2 + w3 = 1.

[0035] Furthermore, the method also includes pre-setting initial values ​​for spine-leg coupling parameters based on different terrain conditions and movement patterns, and dynamically adjusting them during movement through an adaptive mechanism.

[0036] Furthermore, the method also includes a reinforcement learning-based coupling parameter optimization strategy, which continuously optimizes the spine-leg coupling parameters by accumulating experience through interaction with the environment.

[0037] This invention also discloses a quadruped robot motion control system based on spine and leg coordination. Employing the aforementioned quadruped robot motion control method based on spine and leg coordination, the system includes a perception layer, a decision layer, and an execution layer. The perception layer includes position sensors, force sensors, an inertial measurement unit, and a vision sensor to collect information about the robot's own state and the environment. The decision layer is the core of the system, including a high-level controller, a spine-leg coordination controller, and an adaptive parameter adjustment module. The high-level controller is responsible for motion planning and behavior decisions. The spine-leg coordination controller generates coordinated joint trajectories based on a coupling model. The adaptive parameter adjustment module dynamically adjusts the coupling parameters according to the motion state and environmental information. The execution layer includes a spinal joint controller and limb joint controllers, responsible for converting the control signals output from the decision layer into specific joint drive signals and ensuring accurate execution of the joint trajectories through feedback control.

[0038] This invention, employing the above technical solutions, possesses the following technical characteristics compared to existing technologies: 1. Solving the problem of insufficient mobility caused by a rigid torso: Designing a bionic joint structure with ±20° bidirectional bending, improving obstacle-crossing ability by 30% through spinal flexion. 2. Overcoming the weight limitations of existing flexible joints: Employing a combination of a carbon fiber shell (1.2mm thick) and a titanium alloy shaft (8mm diameter), achieving an ultra-light weight of 50g while withstanding an impact torque of 10Nm. 3. Achieving precise dynamic stiffness adjustment: Developing a parallel mechanism of motor and elastic element, supporting real-time adjustment of 0.5-5Nm / rad (response <100ms), adapting to multiple modes of switching between walking (low stiffness) and running (high stiffness). 4. Improving drive efficiency and reliability: Establishing an accurate mathematical model, considering factors such as traffic flow, speed, and vehicle density, improving the accuracy of accident location detection, and meeting the needs of emergency response.

[0039] This invention is applicable to scenarios such as disaster relief, military reconnaissance, and industrial inspection in complex terrain environments. Through a biomimetic spine-limb dynamic coupling and adaptive adjustment mechanism, it significantly improves the robot's motion efficiency, stability, and task adaptability in unstructured environments. Attached Figure Description

[0040] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments;

[0041] Figure 1 This is a schematic diagram of the motion control system architecture for a quadruped robot based on the coordination of the spine and legs, as described in this invention.

[0042] Figure 2 This is a schematic diagram of the spine-leg coupling relationship model of the present invention;

[0043] Figure 3 This is a schematic diagram of the network topology of the central pattern generator of the present invention;

[0044] Figure 4 This is a schematic diagram of the adaptive coupling strength adjustment process of the present invention;

[0045] Figure 5 This is a comparison diagram of the experimental results of the present invention under different terrain conditions. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0047] like Figures 1 to 5 As shown in the figure, this invention discloses a quadruped robot motion control method based on spine and leg coordination, comprising the following steps:

[0048] a) Establish a coupling model between spinal torsion and limb swing to describe the mathematical mapping relationship between spinal joint angles and limb joint angles;

[0049] b) A spine-leg coordination controller consisting of a central mode generator (CPG) composed of multiple nonlinear oscillators. The nonlinear oscillators are coupled to form a complete oscillator network to achieve phase synchronization and amplitude coordination between spinal torsion and limb swing.

[0050] c) Construct an adaptive coupling strength adjustment mechanism to dynamically adjust the coupling strength between the spine and legs based on external sensor information and internal state;

[0051] d) Generate coordinated control signals for the spinal and limb joints based on the current motion state and environmental information, and drive the robot to perform corresponding movements.

[0052] Furthermore, the coupling model between spinal torsion and limb swing includes:

[0053] Forelimb coupling relationship:

[0054] Hindlimb coupling relationship:

[0055] Where θforelimb is the forelimb swing angle, θhindlimb is the hindlimb swing angle, θspine is the spinal torsion angle, and K... f and K h Let φf and φf be the coupling coefficients of the forelimb and hindlimb, respectively. These represent the phase differences between the forelimbs and hindlimbs and the spine, respectively.

[0056] Furthermore, the spine-leg co-controller based on the central pattern generator (CPG) consists of multiple nonlinear oscillators that are coupled together to form a complete oscillator network. The dynamic equation of each oscillator is as follows:

[0057]

[0058] Where x and y are the coordinate positions of the oscillator in the two-dimensional phase space, and r 2 =x 2 +y 2 , represents the square of the distance between the oscillator and the origin; α controls the speed of convergence to the limit cycle (unit: 1 / s, the larger the value, the faster the convergence, typical value is 1-10), μ controls the radius of the limit cycle, and ω is the oscillation frequency (unit: rad / s, determines the oscillation period, typical value is 0.5-5.0).

[0059] Furthermore, the coupling between the oscillators is achieved through the following equation:

[0060]

[0061] Where, the subscript i represents the i-th oscillator (i = 1, 2, ..., n, where n is the total number of oscillators), the subscript j represents the j-th oscillator connected to the i-th oscillator (j ∈ [1, n] and j ≠ i), w ij This is the coupling weight from oscillator j to oscillator i (a dimensionless parameter representing the strength of the interaction between oscillators, typically ranging from 0.1 to 1.0). It is the desired phase difference (unit: radians, representing the desired phase offset between two oscillators, used to achieve different gait modes).

[0062] Furthermore, the specific steps for adaptive coupling strength adjustment in step c) are as follows:

[0063] S1, by sensing the environment and the robot's state;

[0064] S2, evaluate current performance metrics;

[0065] S3: Determine whether the coupling strength needs to be adjusted based on the performance indicators; if so, update the coupling parameters and execute S4; otherwise, execute S1.

[0066] S4, based on the updated coupling parameters, updates the control signal and adds the empirical storage for parameter optimization based on reinforcement learning, then executes S1.

[0067] Furthermore, the adaptive coupling strength adjustment mechanism updates the coupling parameters based on the following adaptive law:

[0068] Furthermore, the adaptive coupling strength adjustment mechanism is based on the following adaptive law:

[0069]

[0070]

[0071] in, and K represents the time rate of change of the coupling coefficients of the forelimb and hindlimb, respectively. f and K h Let E be the coupling coefficients of the forelimb and hindlimb, respectively, and ηf and ηh be the learning rates. target For the target performance index, E actual For actual performance indicators, and These are the gradient symbols for the coupling coefficients of performance E with respect to the forelimb and hindlimb, respectively.

[0072] Furthermore, the performance index E is a multi-objective function:

[0073] E = w1 * Eenergy +w2*E stability +w3*E speed

[0074] Where E is the comprehensive performance index, E energy E is an energy efficiency indicator. stability E is an indicator of motion stability. speed For speed indicators, w1, w2, and w3 are weighting coefficients used to adjust the importance of different performance indicators, satisfying w1 + w2 + w3 = 1.

[0075] Furthermore, the method also includes pre-setting initial values ​​for spine-leg coupling parameters based on different terrain conditions and movement patterns, and dynamically adjusting them during movement through an adaptive mechanism.

[0076] Furthermore, the method also includes a reinforcement learning-based coupling parameter optimization strategy, which continuously optimizes the spine-leg coupling parameters by accumulating experience through interaction with the environment.

[0077] The specific principles of this invention will be explained in detail below:

[0078] like Figure 1 As shown, the quadruped robot spine-leg coordinated control system of the present invention consists of a perception layer, a decision layer, and an execution layer. The perception layer includes position sensors, force sensors, an inertial measurement unit, and a vision sensor, which collect information about the robot's own state and the environment. The decision layer is the core of the system, including a high-level controller, a spine-leg coordinated controller, and an adaptive parameter adjustment module. The high-level controller is responsible for motion planning and behavior decisions; the spine-leg coordinated controller generates coordinated joint trajectories based on a coupling model; and the adaptive parameter adjustment module dynamically adjusts the coupling parameters according to the motion state and environmental information. The execution layer includes a spinal joint controller and limb joint controllers, which are responsible for converting the control signals output from the decision layer into specific joint drive signals and ensuring accurate execution of the joint trajectories through feedback control.

[0079] like Figure 2 As shown, this invention establishes a coupling model between spinal torsion and limb swing. This model, based on a central pattern generator (CPG), describes the synergistic relationship between the spine and limbs through a series of coupled oscillator networks.

[0080] The basic relationship between the spinal torsion angle θspine and the limb swing angle θlimb can be expressed as:

[0081]

[0082] Among them, K cLet φ be the coupling coefficient and φ be the phase difference. Biological observations show that the coupling coefficients and phase differences between the forelimbs and hindlimbs and the spine are different and change with the movement state.

[0083] For the forelimb, the coupling relationship is as follows:

[0084]

[0085] For the hind limbs, the coupling relationship is as follows:

[0086]

[0087] Among them, K f and K h φf and φh are the coupling coefficients of the forelimb and hindlimb, respectively, and the phase differences between the forelimb and hindlimb and the spine, respectively.

[0088] like Figure 3 As shown, this invention designs a spine-leg coordination controller based on a central pattern generator (CPG). The controller consists of multiple nonlinear oscillators that are coupled together to form a complete oscillator network.

[0089] The dynamics of each oscillator can be described by the following equation:

[0090]

[0091] Where x and y are the state variables of the oscillator, r 2 =x 2 +y 2 α controls the speed of convergence to the limit cycle, μ controls the radius of the limit cycle, and ω is the oscillation frequency.

[0092] The coupling between oscillators is achieved through the following equation:

[0093]

[0094] Among them, w ij It is the coupling weight from oscillator j to oscillator i. It is the desired phase difference.

[0095] In this controller, the spinal oscillator is located at the center of the network, forming a bidirectional coupling with the limb oscillators. By adjusting the coupling weights and phase differences, different coordinated motion modes can be achieved.

[0096] like Figure 4 As shown, this invention proposes an adaptive coupling strength adjustment mechanism that can dynamically adjust the coupling strength between the spine and legs according to terrain changes and motion states.

[0097] This mechanism is based on the following adaptive law:

[0098]

[0099] in, and K represents the time rate of change of the coupling coefficients of the forelimb and hindlimb, respectively. f and K h Let E be the coupling coefficients of the forelimb and hindlimb, respectively, and ηf and ηh be the learning rates. target For the target performance index, E actual For actual performance indicators, and These are the gradient symbols for the coupling coefficients of performance E with respect to the forelimb and hindlimb, respectively.

[0100] The performance metric E can be a multi-objective function, including factors such as energy efficiency, motion stability, and speed.

[0101] E = w1 * E energy +w2*E stability +w3*E speed

[0102] Where E is the comprehensive performance index, E energy E is an energy efficiency indicator. stability E is an indicator of motion stability. speed For speed indicators, w1, w2, and w3 are weighting coefficients used to adjust the importance of different performance indicators, satisfying w1 + w2 + w3 = 1.

[0103] Application scenarios and experimental verification: such as Figure 5 As shown, the quadruped robot motion control method of the present invention was experimentally verified under different terrain conditions, including flat ground, slopes, gravel roads and sandy ground.

[0104] Experimental results show that, compared with traditional methods based solely on leg control, the spine-leg coordinated control method of this invention exhibits significant advantages under various terrain conditions:

[0105] (1) When running on flat ground, energy consumption was reduced by 20% and the maximum speed was increased by 15%; (2) When climbing on slopes, the success rate was increased by 35% and the climbing speed was increased by 25%; (3) When walking on gravel roads, stability was increased by 40% and the number of falls was reduced by 60%; (4) When walking on sand, the sinking depth was reduced by 30% and the forward speed was increased by 20%. These results fully demonstrate the effectiveness and practical value of the present invention.

[0106] The beneficial effects of this invention are as follows: 1. Leapfrog improvement in motion efficiency: Through a dynamic coupling model of the spine and limbs and adaptive parameter adjustment, energy transfer efficiency is improved by 40%-50%, the speed of movement on flat ground is increased by 15%-20% compared to traditional rigid quadruped robots, and the efficiency of passing through gravel roads is improved by more than 60%. 2. Breakthrough in multimodal motion capabilities: The invention pioneers a biomimetic control architecture of active spinal torsion and coordinated limb swinging, supporting complex motion modes such as high-speed running at 5m / s (traditional robots ≤3m / s), 0.3s sharp turn response (traditional ≥0.8s), and stepless speed-changing climbing on a 30° slope. 3. Environmental adaptive closed-loop control: Based on a coupling parameter optimization algorithm using reinforcement learning, the invention achieves real-time response across the entire chain from terrain recognition to motion strategy adjustment (response latency <50ms), increasing the pass rate of complex terrain from 62% to 98% (tests cover 12 types of scenarios including snow / gravel / sand). 4. Biomimetic Mechanical Properties Reproduced: The active torsional range of the spinal joint reaches ±25° (the spinal mobility of felines is 20-30°), and the torsional stiffness ratio is dynamically adjusted to 5:1 (traditional mechanisms ≤2:1), reducing foot impact force by 65% ​​(experimental data compared to the Boston Dynamics Spot robot). 5. Revolutionary Optimization of System Energy Efficiency: Through spinal elastic energy storage and joint phase synchronization technology, energy consumption per unit distance is reduced to 1.2 kJ / kg·m (traditional rigid robots ≥2.5 kJ / kg·m), and the runtime is extended to 2.3 times that of similar products (actual test: a 5kg robot can work continuously for 8 hours).

[0107] This invention, employing the above technical solutions, possesses the following technical characteristics compared to existing technologies: 1. Solving the problem of insufficient mobility caused by a rigid torso: Designing a bionic joint structure with ±20° bidirectional bending, improving obstacle-crossing ability by 30% through spinal flexion. 2. Overcoming the weight limitations of existing flexible joints: Employing a combination of a carbon fiber shell (1.2mm thick) and a titanium alloy shaft (8mm diameter), achieving an ultra-light weight of 50g while withstanding an impact torque of 10Nm. 3. Achieving precise dynamic stiffness adjustment: Developing a parallel mechanism of motor and elastic element, supporting real-time adjustment of 0.5-5Nm / rad (response <100ms), adapting to multiple modes of switching between walking (low stiffness) and running (high stiffness). 4. Improving drive efficiency and reliability: Establishing an accurate mathematical model, considering factors such as traffic flow, speed, and vehicle density, improving the accuracy of accident location detection, and meeting the needs of emergency response.

[0108] This invention is applicable to scenarios such as disaster relief, military reconnaissance, and industrial inspection in complex terrain environments. Through a biomimetic spine-limb dynamic coupling and adaptive adjustment mechanism, it significantly improves the robot's motion efficiency, stability, and task adaptability in unstructured environments.

[0109] Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Without conflict, the embodiments and features in the embodiments of this application can be combined with each other. The components of the embodiments of this application described and illustrated herein can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

Claims

1. A quadruped robot motion control method based on spine and leg coordination, characterized in that: It includes the following steps: a) Establish a coupling model between spinal torsion and limb swing to describe the mathematical mapping relationship between spinal joint angles and limb joint angles; b) A spine-leg coordination controller consisting of a central mode generator (CPG) composed of multiple nonlinear oscillators. The nonlinear oscillators are coupled to form a complete oscillator network to achieve phase synchronization and amplitude coordination between spinal torsion and limb swing. c) Construct an adaptive coupling strength adjustment mechanism to dynamically adjust the coupling strength between the spine and legs based on external sensor information and internal state; d) Generate coordinated control signals for the spinal and limb joints based on the current motion state and environmental information, and drive the robot to perform corresponding movements.

2. The quadruped robot motion control method based on spine and leg coordination according to claim 1, characterized in that: The coupling model between spinal torsion and limb swing includes: Forelimb coupling relationship: Hindlimb coupling relationship: Where θforelimb is the forelimb swing angle, θhindlimb is the hindlimb swing angle, θspine is the spinal torsion angle, and K... f and K h Let φf and φf be the coupling coefficients of the forelimb and hindlimb, respectively. These represent the phase differences between the forelimbs and hindlimbs and the spine, respectively.

3. The quadruped robot motion control method based on spine and leg coordination according to claim 1, characterized in that: The dynamic equation for each oscillator is: Where x and y are the coordinate positions of the oscillator in the two-dimensional phase space, and r 2 =x 2 +y 2 , represents the square of the distance between the oscillator and the origin; α controls the speed of convergence to the limit cycle, μ controls the radius of the limit cycle, and ω is the oscillation frequency.

4. The quadruped robot motion control method based on spine and leg coordination according to claim 3, characterized in that: The coupling between oscillators is achieved through the following equation: Where, the subscript i represents the i-th oscillator, i = 1, 2, ..., n, and n is the total number of oscillators; the subscript j represents the j-th oscillator connected to the i-th oscillator, j ∈ [1, n] and j ≠ i; w ij It is the coupling weight from oscillator j to oscillator i. It is the desired phase difference.

5. The quadruped robot motion control method based on spine and leg coordination according to claim 1, characterized in that: The specific steps for adaptive coupling strength adjustment in step c) are as follows: S1, by sensing the environment and the robot's state; S2, evaluate current performance metrics; S3: Determine whether the coupling strength needs to be adjusted based on the performance indicators; if so, update the coupling parameters and execute S4; otherwise, execute S1. S4, based on the updated coupling parameters, updates the control signal and adds the empirical storage for parameter optimization based on reinforcement learning, then executes S1.

6. The quadruped robot motion control method based on spine and leg coordination according to claim 1 or 5, characterized in that: The adaptive coupling strength adjustment mechanism updates the coupling parameters based on the following adaptive law: in, and K represents the time rate of change of the coupling coefficients of the forelimb and hindlimb, respectively. f and K h Let E be the coupling coefficients of the forelimb and hindlimb, respectively, and ηf and ηh be the learning rates. target For the target performance index, E actual For actual performance indicators, and These are the gradient symbols for the coupling coefficients of performance E with respect to the forelimb and hindlimb, respectively.

7. The quadruped robot motion control method based on spine and leg coordination according to claim 6, characterized in that: Performance metric E is a multi-objective function: E=w1*E energy +w2*E stability +w3*E speed Where E is the comprehensive performance index, E energy E is an energy efficiency indicator. stability E is an indicator of motion stability. speed For speed indicators, w1, w2, and w3 are weighting coefficients used to adjust the importance of different performance indicators, satisfying w1 + w2 + w3 = 1.

8. The quadruped robot motion control method based on spine and leg coordination according to claim 1, characterized in that: The method also includes presetting initial values ​​for spine-leg coupling parameters based on different terrain conditions and movement patterns, and dynamically adjusting them during movement through an adaptive mechanism.

9. The quadruped robot motion control method based on spine and leg coordination according to claim 1, characterized in that: The method also includes a reinforcement learning-based coupling parameter optimization strategy, which continuously optimizes the spine-leg coupling parameters by accumulating experience through interaction with the environment.

10. A quadruped robot motion control system based on spine and leg coordination, employing the quadruped robot motion control method based on spine and leg coordination as described in any one of claims 1 to 9, characterized in that: The system consists of a perception layer, a decision-making layer, and an execution layer; the perception layer includes position sensors, force sensors, inertial measurement units, and vision sensors to collect information about the robot's own state and the environment; The decision-making layer is the core of the system, including a high-level controller, a spine-leg coordination controller, and an adaptive parameter adjustment module. The high-level controller is responsible for motion planning and behavioral decisions. The spine-leg coordination controller generates coordinated joint trajectories based on a coupling model; The adaptive parameter adjustment module dynamically adjusts the coupling parameters based on motion state and environmental information; the execution layer includes a spinal joint controller and a limb joint controller, which are responsible for converting the control signals output by the decision layer into specific joint drive signals, and ensuring the accurate execution of joint trajectories through feedback control.

Images