Humanoid lower limb exoskeleton robot motion control method and device and storage medium

Abstract

The application discloses a motion control method and device of a humanoid lower limb exoskeleton robot and a storage medium, and relates to the technical field of industrial robots. For a humanoid lower limb exoskeleton robot with the third, fourth and fifth joint rotation axes being parallel to each other, the method comprises the following steps: receiving a foot end desired pose to obtain a target pose matrix; constructing an algebraic equation to analytically solve a first joint angle; based on this, sequentially decoupling and calculating the angles of the space-related joints and the parallel-axis group joints; and finally converting the angles into cooperative driving instructions. Through a step-by-step decoupling analytical solving strategy, the application avoids problems such as matrix singularity and iteration non-convergence existing in traditional numerical iteration methods, can efficiently, accurately and stably obtain the rotation angles of each joint of the robot, and thus realizes precise and robust control of the exoskeleton robot with a specific configuration.

Description

Technical Field

[0001] This application relates to the field of industrial robot technology, and in particular to a motion control method, device and storage medium for a humanoid lower limb exoskeleton robot. Background Technology

[0002] Humanoid lower limb exoskeleton robots, as a typical multi-degree-of-freedom serial robot, show broad application prospects in fields such as medical rehabilitation and assisted walking. The core technological challenge lies in achieving precise, stable, and rapid motion control of the robot's lower limbs to ensure the safety and smoothness of human-robot interaction. The key to achieving these control objectives lies in the inverse kinematics algorithm, which is responsible for converting the desired foot pose (i.e., spatial position and attitude) planned by the host computer into the target rotation angles of each joint of the lower limb in real time and accurately. Therefore, the accuracy, speed, and stability of the inverse kinematics solution directly determine the overall control performance of the exoskeleton robot.

[0003] In existing technologies, the inverse kinematics solution path for six-joint serial robots is highly dependent on their specific configuration. For example, for ball-arm or UR-type robots that satisfy specific geometric constraints (such as the Pieper criterion), some joint axes intersect or are parallel, allowing the derivation of univariate analytical equations for joint angles using geometric methods, thus obtaining fast and accurate analytical solutions. However, for non-standard configurations that do not satisfy these classical geometric conveniences (such as the configuration where the third, fourth, and fifth joint axes are parallel as described in this application), it is difficult to obtain analytical solutions directly.

[0004] To address the challenges of solving non-standard configurations, existing technologies typically employ numerical iterative methods based on Jacobian matrices. However, these methods exhibit several inherent drawbacks in practical applications: matrix singularities during the solution process can cause the robot to lose its degrees of freedom in specific postures, leading to sudden motion changes and control instability; the accuracy of the numerical solution is limited by the iteration step size and convergence conditions, making it difficult to guarantee high-precision trajectory tracking and resulting in low control accuracy; and the instability of the iteration speed and the potential for iteration non-convergence can compromise the real-time performance of the control system, even leading to control task failure. This manifests as low system robustness, making the system prone to unexpected program errors in complex dynamic tasks, posing a potential threat to the wearer's safety. Summary of the Invention

[0005] This application provides a motion control method, device, and storage medium for a humanoid lower limb exoskeleton robot, to at least solve the above-mentioned technical problems existing in the prior art.

[0006] According to a first aspect of this application, a motion control method for a humanoid lower limb exoskeleton robot is provided. Each lower limb of the humanoid lower limb exoskeleton robot includes six joints, with the rotation axes of the third, fourth, and fifth joints being parallel to each other. The motion control method for the humanoid lower limb exoskeleton robot includes: S1: receiving the desired target pose of the foot relative to the body coordinate system from a host computer, and performing homogeneous coordinate transformation on it to obtain a target pose matrix, wherein the desired target pose includes a desired position and a desired attitude angle; S2: based on a preset set of link geometric parameters and the target pose matrix, performing... S3: Based on the first joint target angle, solve the non-parallel axis related elements in the target pose matrix to obtain the spatially associated joint angle set, which includes the second joint target angle and the sixth joint target angle; S4: Based on the constraints of the parallel axis related elements in the target pose matrix and combined with the first joint target angle and the spatially associated joint angle set, determine the parallel axis group joint angle set; S5: Convert the first joint target angle, the spatially associated joint angle set, and the parallel axis group joint angle set into corresponding pulse width modulation signals to obtain the six-axis cooperative drive command.

[0007] According to a second aspect of this application, a motion control device for a humanoid lower limb exoskeleton robot is provided. Each lower limb of the humanoid lower limb exoskeleton robot includes six joints, and the rotation axes of the third, fourth, and fifth joints are parallel to each other. The device includes: a target pose receiving unit, used to receive the desired target pose of the foot relative to the body coordinate system from a host computer, and perform homogeneous coordinate transformation on it to obtain a target pose matrix, wherein the desired target pose includes a desired position and a desired attitude angle; and a joint analytical solution unit, used to perform real-domain analytical solution on the first joint variables based on a preset set of link geometric parameters and the target pose matrix to obtain a first... The system includes: a joint target angle unit; a spatially associated joint calculation unit, used to calculate the non-parallel axis related elements in the target pose matrix based on the first joint target angle to obtain a set of spatially associated joint angles, which includes the second joint target angle and the sixth joint target angle; a parallel axis group joint determination unit, used to determine the set of parallel axis group joint angles based on the constraints of elements related to parallel axes in the target pose matrix and in combination with the first joint target angle and the set of spatially associated joint angles; and a drive signal generation unit, used to convert the first joint target angle, the set of spatially associated joint angles, and the set of parallel axis group joint angles into corresponding pulse width modulation signals to obtain a six-axis cooperative drive command.

[0008] According to a third aspect of this application, an electronic device is provided, comprising:

[0009] At least one processor;

[0010] A memory that is communicatively connected to the at least one processor;

[0011] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in this application.

[0012] According to a fourth aspect of this application, a non-transitory computer-readable storage medium is provided storing computer instructions for causing a computer to perform the methods described in this application.

[0013] Compared to existing technologies, this method employs a fully analytical solution process, rather than numerical iteration, thus fundamentally avoiding potential problems such as matrix singularities and iteration non-convergence, ensuring the algorithm's stability and robustness across the entire workspace. Secondly, the analytical solution yields the accurate theoretical truth value in one step, significantly improving the precision of robot motion control compared to the approximate solutions of numerical methods, enabling the feet to more accurately reproduce the desired trajectory. Furthermore, deterministic algebraic operations replace time-consuming iterative processes, resulting in a constant and extremely low computational cost per solution, fully meeting the stringent real-time requirements of humanoid lower limb exoskeleton robots for their control systems.

[0014] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0015] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of this application are illustrated in the drawings by way of example and not limitation, in which:

[0016] In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts.

[0017] Figure 1 A schematic diagram of the structure of the humanoid robot according to an embodiment of this application is shown;

[0018] Figure 2 A schematic diagram illustrating the implementation flow of the motion control method for the humanoid lower limb exoskeleton robot according to an embodiment of this application is shown.

[0019] Figure 3 This paper shows a schematic diagram of the composition structure of the motion control device for the humanoid lower limb exoskeleton robot according to an embodiment of this application;

[0020] Figure 4A schematic diagram of the composition structure of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0021] To further illustrate the technical means and effects adopted by this application in order to achieve the intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of this application is provided in conjunction with the accompanying drawings and preferred embodiments.

[0022] This application provides a motion control method for a humanoid lower limb exoskeleton robot, and a structural reference for the humanoid lower limb exoskeleton robot. Figure 1 As shown, each lower limb of the humanoid lower limb exoskeleton robot comprises six joints, with the rotation axes of the third, fourth, and fifth joints being parallel to each other. This specific configuration design is based on in-depth simulation of human lower limb biomechanics and optimization of the robot's kinematic solution path. During daily activities such as walking and squatting, the flexion and extension of the hip joint, the flexion and extension of the knee joint, and the plantar flexion / dorsiflexion of the ankle joint primarily occur in the sagittal plane, and their respective rotation axes remain approximately parallel in space. Therefore, the design of parallel axes for the third, fourth, and fifth joints allows the exoskeleton robot's movement pattern to better conform to the natural human gait, ensuring coordination and comfort in human-robot interaction. More importantly, this seemingly non-standard configuration, which increases the complexity of the solution, creates unique geometric constraints for abandoning traditional numerical iteration methods and realizing a completely new analytical solution path.

[0023] In one specific implementation of this application, the left and right leg structures of the lower limb exoskeleton robot are completely symmetrical along the sagittal plane, with consistent configuration and dimensional parameters. The six joints of a single lower limb typically correspond sequentially from top to bottom to the functions of the human hip, knee, and ankle joints. For example, the first and second joints realize hip abduction / adduction and internal / external rotation, providing degrees of freedom for lateral movement and steering; the third joint realizes hip flexion / extension; the fourth joint realizes knee flexion / extension; the fifth joint realizes ankle plantar flexion / dorsiflexion; and the sixth joint realizes ankle inversion / eversion. When establishing the robot's kinematic model, for example using the improved DiNavitt-Hartenberg method, the center of the robot's waist can be set as the base coordinate system, and corresponding joint coordinate systems can be established sequentially on the rotation axes of the six joints. In this model, the key feature that the rotation axes of the third, fourth, and fifth joints are parallel to each other means that the rotation axis vectors (e.g., the Z-axis) of each of these three joint coordinate systems have exactly the same direction in space. This geometric characteristic causes the movement of these three joints to physically constitute a planar linkage mechanism, with its main motion constrained within mutually parallel planes. This is the core of the lower limb's main movements, such as forward and backward swings. The motion control method of this application utilizes this parallel axis geometric constraint to achieve analytical solutions for inverse kinematics.

[0024] Specifically, such as Figure 2 As shown, the motion control method for the humanoid lower limb exoskeleton robot includes:

[0025] S1: Receives the desired target pose of the foot relative to the body coordinate system from the host computer and performs homogeneous coordinate transformation on it to obtain the target pose matrix. The desired target pose includes the desired position and desired attitude angles. It should be understood that in the control architecture of a humanoid lower limb exoskeleton robot, the host computer typically generates the desired motion trajectory of the foot relative to the body (usually the waist coordinate system) based on environmental perception and gait planning. This trajectory data exists in the form of Cartesian coordinate position and Euler angle attitude. However, subsequent solutions to the quartic equations based on the Ferrari method and the analytical calculations of each joint rely on specific elements in the homogeneous transformation matrix (such as normal vector components, position vector components, etc.) as algebraic coefficients. Therefore, homogeneous coordinate transformation must be performed to standardize the desired position and desired attitude angles into a target pose matrix containing rotational and position components, and further decompose it into the vector elements required by the algorithm.

[0026] In one specific implementation of this application, step S1 includes: First, the system receives discretized trajectory point data sent by a host computer and separates the position data and attitude data. Based on a preset rotation transformation order (e.g., ZYX order), the received desired attitude angle is converted into a three-dimensional orthogonal rotation matrix. It is assumed that the desired attitude angle includes a yaw angle. Pitch angle Roll angle The generated rotation matrix The calculation formula is as follows:

[0027] Formula (1)

[0028] in Represents the cosine function cos, Represents the sine function sin, ( , , ( ) represents the three-dimensional coordinates of the foot position relative to the waist of the fuselage along the X-axis. , , () represents the three-dimensional coordinates of the foot position relative to the waist of the fuselage along the Y-axis. , , () represents the three-dimensional coordinates of the foot position relative to the waist of the fuselage in the Z-axis direction.

[0029] Secondly, combined with the received expected location coordinates The target pose matrix is ​​obtained by assembling the block matrix. The specific operation is to... rotation matrix Expand to the top left corner block and set the desired position coordinates. Fill in the first three rows of the fourth column of the matrix, and fill the fourth row with constant values. The mathematical expression for this matrix is:

[0030] Formula (2)

[0031] in,( , , () represents the three-dimensional coordinates of the foot position relative to the waist of the machine body at the origin.

[0032] Finally, to match the requirements of subsequent analytical solution algorithms for specific matrix elements, the target pose matrix needs to be deconstructed into column vectors. The system extracts the first column of the matrix as the normal vector. Extract the second column as the orientation vector. Extract the third column as the proximity vector. Extract the fourth column as a position vector. These deconstructed vector components will serve as the core input stream, directly participating in the construction of the coefficients of the subsequent quartic equation.

[0033] In this embodiment, considering that the relative positions of the feet and waist of the fuselage are fixed, the target pose matrix can be a known matrix that can be directly obtained. It should be noted that... Figure 1 Each joint of the robot has a corresponding origin and X-axis, Y-axis, and Z-axis directions. Figure 1 The Z-axis of the third, fourth, and fifth joints are parallel to each other.

[0034] S2: Based on a preset set of link geometric parameters and the target pose matrix, the first joint variables are analytically solved in the real domain to obtain the target angle of the first joint. Specifically, for the non-standard configuration robot with parallel axes of the third, fourth, and fifth joints in this application, traditional geometric methods cannot directly separate the first joint variables, while numerical methods suffer from instability. Therefore, it is necessary to construct algebraic constraints containing link parameters and pose elements, and use trigonometric identities to eliminate intermediate variables (mainly the second joint variables), thereby establishing a mathematical model that only concerns the first joint variables. This process not only avoids matrix singularities but also lays a definite numerical foundation for the decoupling calculation of other joints, which is a key step in realizing fully analytical control logic.

[0035] The preset link geometry parameter set refers to a set of constants that are pre-measured and stored in the controller, describing the length of each link of the robot, such as the axial distance between joints. .

[0036] In one specific implementation of this application, step S2 includes: establishing a system of equations containing the first joint variable and the second joint variable based on the equivalence of the coordinate system transformation matrix. Specifically, it is assumed that the rotational positions of the six motor joint axes of a single leg are known:

[0037] Formula (3)

[0038] coordinate system The components are sequentially fixed to the waist body, hip abduction arm, hip rotation arm, thigh, calf, ankle pitch arm, and foot. It has six rotational axes. , , The three axes are parallel.

[0039] The relative rotation matrices of each joint are as follows:

[0040] Formula (4)

[0041] Formula (5)

[0042] Formula (6)

[0043] Formula (7)

[0044] Formula (8)

[0045] Formula (9)

[0046] in, ~ These are the relative rotation matrices between the first joint and the waist body (as the origin joint), the relative rotation matrix between the second joint and the first joint, the relative rotation matrix between the third joint and the second joint, the relative rotation matrix between the fourth joint and the third joint, the relative rotation matrix between the fifth joint and the fourth joint, and the relative rotation matrix between the sixth joint and the fifth joint. ~ These are the axial distances between the first joint and the waist section of the fuselage (as the origin joint), the axial distance between the second joint and the first joint, the axial distance between the third joint and the second joint, the axial distance between the fourth joint and the third joint, the axial distance between the fifth joint and the fourth joint, and the axial distance between the sixth joint and the fifth joint.

[0047] Then, construct the relative transformation matrix of the sixth joint relative to the second joint. There are two forms of expression. The first form is based on forward kinematics and is obtained by multiplying the transformation matrices of the third, fourth, fifth, and sixth joints. Since the axes of the third to fifth joints are parallel, some elements of this matrix exhibit a specific concise form, which can be obtained through linear algebra operations. The matrix of the first form is:

[0048] Formula (10)

[0049] in, , , This is the relative transformation matrix of the sixth joint relative to the second joint.

[0050] The second form is based on reverse derivation. That is, the relative transformation matrix of the sixth joint relative to the second joint can also be determined based on the pose matrix of the distal foot. Through linear algebra operations, the matrix of the second form can be obtained as follows:

[0051] Formula (11)

[0052] in, This is the relative transformation matrix of the sixth joint relative to the second joint. This is the relative rotation matrix between the first joint and the origin joint. Let be the relative rotation matrix between the first joint and the second joint. Let be the target pose matrix. In the above formula (11):

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061] Formula (12)

[0062] besides, The elements shown are not used in subsequent calculations, and are therefore not displayed in the matrix for the time being to make the formula more concise.

[0063] Then, by comparing specific elements in the two matrix forms, specifically comparing the 3rd and 4th elements of the first row (i.e., the first group of row elements) in both matrix forms, we can obtain information about... and Solve the system of equations:

[0064]

[0065]

[0066] Formula (13)

[0067] Transform it into something about and The system of linear equations, Let's assume it's a constant for now:

[0068]

[0069]

[0070] Formula (14)

[0071] It can be calculated.

[0072]

[0073]

[0074] Formula (15)

[0075] Then, using trigonometric identities Eliminate the second joint variable to obtain information about the first joint variable. The single-variable trigonometric equation is:

[0076]

[0077] Formula (16)

[0078] The key to constructing this trigonometric equation lies in determining the six intermediate coefficients. The system parses the position vector from the target pose matrix. and proximity vector And combined with a preset set of link geometry parameters (such as the axial distance between each joint) These coefficients are determined through algebraic operations. The specific calculation formula is as follows: [Formula for coefficients corresponding to the quadratic sine term]. The coefficient of the corresponding quadratic cosine term. ; the coefficients of the product of sine and cosine. ; the coefficient of the first-order sine term The coefficient of the corresponding first-order cosine term. ; and the coefficient of the constant term .

[0079] After determining the intermediate coefficients and constructing the trigonometric equations, the single-variable trigonometric function equations are transformed into a quartic equation in terms of the half-tangent value of the first joint variable using the half-angle tangent substitution. , , Substituting into the equation, eliminating the denominators and simplifying like terms, we obtain the equation regarding... The quartic equation in one variable is as follows:

[0080] Formula (17)

[0081] in, , , , , .

[0082] Finally, the quartic equation in one variable is solved in the real field. The system uses the Ferrari method to find four complex roots, discarding solutions with non-zero imaginary parts to obtain the real roots. Finally, inverse trigonometric functions are used... The real roots are restored to obtain the accurate target angle of the first joint.

[0083] S3: Based on the target angle of the first joint, solve for the non-parallel axis related elements in the target pose matrix to obtain the spatially associated joint angle set, which includes the target angles of the second and sixth joints. It should be understood that in the configuration of a humanoid lower limb exoskeleton robot, the motion states of the second joint (usually hip abduction / adduction) and the sixth joint (usually ankle inversion / eversion) are directly affected by the first joint (hip internal rotation / external rotation). Once the angle of the first joint is known, the solutions for these two joints can be obtained directly through simple algebraic mapping and geometric projection, without the need for complex simultaneous equations.

[0084] In one specific implementation of this application, step S3 includes: calculating the second joint target angle based on the equivalence constraint of the coordinate transformation matrix and combining the elements of the first joint target angle and the target pose matrix; calculating the sixth joint target angle using the projection geometric relationship between the components of the normal vector in the base coordinate system and the first joint target angle; and combining and packaging the second joint target angle and the sixth joint target angle to obtain a spatially associated joint angle set.

[0085] Based on the relative transformation matrix of the sixth joint relative to the second joint The equivalence between the transformation matrix derived from the end-position pose and the matrix equation is used to extract the relevant information. Strongly related elements. Constructing about The algebraic expressions for the sine and cosine terms are... and pose elements The function. Using the bivariate arctangent function. The uniquely determined target angle of the second joint is determined. The calculation formula is as follows:

[0086]

[0087] Formula (18)

[0088] Secondly, the system utilizes the projection geometric relationship between the components of the normal vector in the base coordinate system and the target angle of the first joint to calculate the target angle of the sixth joint. By comparing the first and second elements of the first row of the transformation matrix, we can obtain:

[0089]

[0090]

[0091] Formula (19)

[0092] Therefore, we can find that: = .

[0093] Then, using the components of the normal vector in the base coordinate system Angle with the first joint target From the geometric relationship, we can know that:

[0094]

[0095]

[0096] Formula (20)

[0097] Finally, the calculated and By combining and packaging, a set of spatially related joint angles is obtained.

[0098] S4: Based on the constraints of elements related to the parallel axes in the target pose matrix and combined with the target angle of the first joint and the spatially associated joint angle set, determine the joint angle set of the parallel axis group. Specifically, the robot's third, fourth, and fifth joints (usually corresponding to hip flexion / extension, knee flexion / extension, and ankle dorsiflexion / plantarflexion) constitute a planar linkage mechanism with parallel rotation axes. These three joints are mainly responsible for the robot's main motion functions such as walking and squatting in the sagittal plane. In the previous steps, the angles of the first joint and the spatially associated angles of the second and sixth joints have been determined analytically, which has actually determined the motion plane of the robot's lower limbs. Therefore, in this step, the six-degree-of-freedom spatial pose is transformed into this local plane using the known joint angles, thereby simplifying the remaining problem into the inverse kinematics problem of a planar three-bar linkage mechanism, so as to achieve accurate and fast calculation of the joint angle set of the parallel axis group.

[0099] In one specific implementation of this application, step S4 includes: transforming the base coordinate pose to an intermediate coordinate system based on the first joint target angle and the spatially associated joint angle set, and calculating the third joint target angle. That is, using known... and The pose in the base coordinate system is transformed to the second joint coordinate system, and the transformation matrix is ​​extracted. The corresponding positional component and Thus, we can obtain information about and Solve the system of equations:

[0100]

[0101]

[0102] Formula (21)

[0103] Combining equations yields information about a single variable. The equation is: .

[0104] Among them, the intermediate coefficient , , The calculation formula is:

[0105] ,

[0106] ,

[0107]

[0108] Formula (22)

[0109] Based on the universal substitution formula ,make Then the above can be converted into information about The polynomial equation was finally solved as follows:

[0110]

[0111] Formula (23)

[0112] Secondly, based on the target angle of the third joint and the closed-loop geometric relationship of the link, the target angle of the fourth joint is calculated. After the target angle of the third joint is obtained, a framework is constructed regarding... The expressions for sine and cosine are:

[0113]

[0114]

[0115] Formula (24)

[0116] By using the bivariate arctangent function to solve for the sum of angles, we can obtain:

[0117] .

[0118] Then, subtract We can obtain:

[0119] -

[0120] Formula (25)

[0121] Next, the pitch attitude components after the target pose matrix transformation are extracted using the parallel axis characteristics to determine the total pitch angle. The target angle of the fifth joint is then calculated by successively subtracting the target angles of the third and fourth joints from the total pitch angle. Based on the matrix transformation correspondence—that is, comparing the second and third elements of the second column of the two expressions—the total pitch angle can be obtained. Solve the system of equations:

[0122]

[0123]

[0124] Formula (26)

[0125] The total pitch angle can be obtained by using the bivariate arctangent function:

[0126] Formula (27)

[0127] Will , Substituting the value into formula (27), we can obtain

[0128]

[0129] Formula (28)

[0130] Finally, the target angles of the third joint, the fourth joint, and the fifth joint are combined and packaged to obtain the set of joint angles of the parallel axis group.

[0131] S5: Convert the target angle of the first joint, the spatially associated joint angle set, and the parallel axis group joint angle set into corresponding pulse width modulation signals to obtain six-axis cooperative drive commands. Since the exoskeleton robot acts directly on the human body, a simple mathematical inverse solution may lead to physical interference due to the lack of consideration for nonlinear coupling or mechanical constraints between joints, and the floating-point numbers output by the computer cannot directly drive the motors. Therefore, this step vectorizes and reassembles the discrete joint angle data, performs dynamic safety corrections, and linear mappings to ultimately generate six-axis cooperative drive commands with temporal synchronization, ensuring the safety, compliance, and accuracy of the robot's motion.

[0132] Specifically, step S5 includes: First, based on the physical identification order of the motors of the lower limb exoskeleton robot, the target angle of the first joint, the set of spatially associated joint angles, and the set of parallel axis group joint angles are vectorized and recombined to obtain the target joint angle vector. Specifically, the topological mapping problem between the algorithm's solution logic order and the hardware physical connection order is solved through vectorization and recombination. In the analytical solution process of this application, in order to overcome the mathematical coupling difficulties, a step-by-step decoupling calculation strategy is adopted, and the timing of its calculation output is as follows: first joint, spatially associated joints (second and sixth joints), and parallel axis group joints (third, fourth, and fifth joints). However, the underlying servo communication network (such as CAN bus or EtherCAT bus) of the lower limb exoskeleton robot is strictly addressed and data distributed according to the motor physical ID (usually identified as ID1 to ID6 from the hip to the foot). The discrete data output by the algorithm is out of order and independent in memory, and direct transmission can easily cause misalignment of the control channel. Through vectorization and recombination, the system forces the scattered calculation results to be aligned and encapsulated in the order of the motor physical IDs, establishing a unique mapping relationship from the abstract mathematical solution space to the physical control space, ensuring that control commands can be accurately delivered to the corresponding joint actuators.

[0133] In one specific implementation of this application, the vectorization and reorganization of the first joint target angle, the spatially associated joint angle set, and the parallel axis group joint angle set can be achieved by traversing a single axis and performing a fixed threshold hard truncation. In this specific implementation, the system first sequentially assembles the input discrete angle variables to form an initial target joint angle vector. Then iterate through each angle element in the vector. This is compared with a preset set of joint soft limit parameters (including the physical maximum angle). and minimum angle The comparison is performed. The execution logic is as follows: if... > Then force set ;like < Then force set = This outputs the final target joint angle vector.

[0134] While the aforementioned implementation methods are common in simple rigid body control, they reveal highly destructive limitations in real-world applications of high-degree-of-freedom anthropomorphic exoskeletons.

[0135] First, this specific implementation method severely lacks consideration of the coupling constraints in the configuration space. Exoskeleton systems, especially anthropomorphic configurations simulating the human lower limbs, do not have isotropic independent variables between their joints; rather, there is deep biomechanical coupling. For example, the hip flexion angle ( Changes in the knee joint will directly determine the knee joint's ( The maximum mechanical range of motion of the knee joint is often limited by the stroke of the hydraulic cylinder driving the two joints or the length of the linkage of the muscles of the two joints. If a static cutoff is mechanically set for the knee joint (e.g., fixed at 120 degrees), once the hip joint reaches its maximum extension state, the actual feasible region of the knee joint will inevitably be smaller than this static threshold, leading to serious accidents such as physical limit impact, damage from forced extension of the hydraulic cylinder, or even soft tissue strain of the wearer. Its essential flaw lies in the inability to recognize that the feasible region is a dynamic polytope that changes with the state, rather than a simple supercubic prism.

[0136] Secondly, the discontinuity in the trajectory caused by hard truncation logic poses a potential threat to controller stability. When the event is triggered, the original logic forcibly clamps the instruction to... This instantaneous step jump causes a mathematical abrupt change in the derivative (velocity) and second derivative (acceleration) of the position function. In a physical servo system, this infinitely large jump instantly translates into a surge in motor current, causing severe system jitter. This not only worsens the smoothness of human-machine interaction but also rapidly wears down transmission components and significantly reduces the lifespan of the controller.

[0137] In another specific implementation of this application, a multi-axis coupled dynamic flexible constraint method based on the potential energy field gradient is constructed. This method first performs a step of constructing multi-axis coupled dynamic constraint boundaries, thus breaking the assumption of single joint independence and establishing an associated constraint model that conforms to the physical entity.

[0138] In this specific implementation, a target joint angle vector is constructed based on the target angle of the first joint, the set of spatially associated joint angles, and the set of parallel axis group joint angles. This includes: vectorizing and recombining the target angle of the first joint, the set of spatially associated joint angles, and the set of parallel axis group joint angles based on the motor physical identification order of the lower limb exoskeleton robot to obtain an initial target joint angle vector; calculating the state dependency constraint relationship between adjacent joints based on a preset coupling constraint coefficient matrix and using an affine transformation algorithm to obtain a dynamic constraint boundary vector; determining the gradient correction angle vector based on the dynamic constraint boundary vector and the potential field gain parameter; and performing element-wise numerical minimization verification between the gradient correction angle vector and the dynamic constraint boundary vector to obtain the target joint angle vector.

[0139] Specifically, firstly, based on the physical identification order of the motors of the lower limb exoskeleton robot, the input discrete angle variables are sequentially assembled to form an initial target joint angle vector. Then it receives the initially generated target joint angle vector and combines it with the preset coupling constraint coefficient matrix, without calling the fixed... Instead of using affine transformations, this step utilizes real-time calculations of the safety boundaries that follow the current configuration. This allows for precise description of complex mechanical and physiological constraints, such as the smaller the permissible knee extension angle as the hip flexion angle increases, ensuring that control commands remain within a physically safe dynamic envelope. This step introduces an inter-joint mapping function to calculate the dynamic upper limit of the i-th joint. , means as follows:

[0140] Formula (29)

[0141] in, Let be the dynamic constraint boundary value of the i-th joint, representing the actual safety upper limit under the current configuration. This formula uses linear weighted superposition to quantify the impact of the state changes of other joints (j) on the feasible region of the current joint (i). This is the physical hard limit constant of the joint in an uncoupled state. For example, for the knee joint, this value can be set to the maximum extension angle allowed by mechanical limit (such as 0 radians or a slight flexion angle). These are the weighted elements in the coupling constraint coefficient matrix, quantifying the linear or nonlinear influence of joint j on the range of motion of joint i (for example, a coefficient of -0.5 indicates that for every 1 degree increase in joint j, the limit of joint i decreases by 0.5 degrees). This approach aims to upgrade the limit determination criterion from a static constant to a dynamic variable of state compliance, thereby eliminating configuration interference at its source.

[0142] Subsequently, based on the dynamic constraint boundary vector and the potential field gain parameter, the gradient correction angle vector is determined. To address the abrupt changes in velocity and acceleration caused by hard truncation, a soft constraint is implemented by introducing the concept of a virtual potential energy field. Specifically, a nonlinear potential energy function U(q) is defined, which is zero when the joint angle is at the center of the safe zone and increases as it approaches the dynamic boundary. With a pre-defined buffer, the potential energy increases exponentially. The system calculates the negative gradient of the potential energy field with respect to the joint angle and adds it as an obstacle avoidance correction to the original target command. This eliminates the discrete logic of if-else statements and utilizes a continuously differentiable potential energy function. When the joint is about to exceed the boundary but has not yet done so, a smoothly increasing reverse virtual thrust is applied, automatically and gently pushing the command back to the safe zone using gradient descent. This processing method calculates the corrected angle vector. The process is as follows:

[0143] Formula (30)

[0144] in, This represents the initial target joint angle vector. It is a gradient-corrected angle vector that has been smoothed out. The preset potential field buffer width defines the scope of the soft limiting mechanism intervention. The potential field gain parameter determines the strength of the corrected force field. In this process, the preset values ​​involved need to be determined based on the actual hardware characteristics: the preset potential field buffer width... The range of soft-limit intervention is defined, typically set to 0.087 radians (approximately 5 degrees), meaning that the correction mechanism is activated when the joint distance limit is less than 5 degrees; potential field gain parameter. The strength of the correction force field is determined, typically between 0.1 and 1.0. For example, setting it to 0.5 requires experimental tuning to balance the correction speed with system stability and avoid oscillations. The principle of this formula is gradient descent: the summation term in the formula represents the total potential energy. The function ensures that only when the joint enters the distance dynamic boundary... buffer Potential energy is only generated within a certain range; This represents taking the partial derivative with respect to the angle vector, i.e., calculating the direction in which the potential energy increases the most (the direction that exceeds the limit). This generates a reverse "virtual thrust," gently pushing the joint angle back to the safe zone. This ensures the out-of-bounds correction process is continuous, effectively eliminating inflection points in the position curve and thus avoiding mechanical jitter caused by sudden changes in commands.

[0145] Finally, the system performs a final verification and output based on kinematic consistency. Although the potential energy field can handle most operating conditions, considering occasional situations of extreme high dynamics or insufficient gain, a final physical safety barrier must be established. Specifically, the smoothed gradient correction angle vector is compared element-by-element with the dynamic constraint boundary vector, and the smaller value is selected. This calculation not only provides a logical safety net but also mathematically eliminates any residual out-of-bounds quantities, generating the final target joint angle vector. Thus, a complete control flow system, from dynamic boundary definition and smooth potential field correction to the final safety closed loop, is constructed.

[0146] This specific implementation aims to solve the challenges of safety and compliance control in complex humanoid exoskeleton systems under multi-joint coupled motion. By introducing a multi-axis coupled dynamic model, it successfully overcomes the bottleneck of traditional static limiting mechanisms in handling mechanical interference between joints, effectively preventing physical risks such as cylinder scoring, hard collisions, and soft tissue damage to the human body, and achieving precise and dynamic adaptation to the robot's feasible configuration space. Simultaneously, flexible correction using potential energy field gradients transforms the originally rigid wall-collision limiting into a gradual soft landing, mathematically guaranteeing the first and second-order continuity of the control signal at the boundaries. This not only eliminates motor overcurrent and body vibration caused by sudden command changes, significantly improving the wearer's human-machine interaction comfort, but also greatly reduces mechanical wear of the transmission mechanism, extending the overall service life of the system and achieving the comprehensive control goals of safety, compliance, and durability.

[0147] After completing the safety verification, the system performs a linear mapping from the physical angle to the PWM control signal. This is based on a preset set of motor calibration parameters (including the reduction ratio). Zero point offset and conversion gain ), using formula The radian values ​​of each joint are converted into servo pulse width data. Finally, the servo pulse width data is frame-encapsulated. The system creates a command data frame according to the underlying communication protocol, writes a synchronization control identifier in the frame header to ensure that the six motors move at the same time scale, fills the data segment with the PWM data of the six joints, and calculates the checksum (such as CRC16) and fills it in the frame tail, thereby generating the final six-axis coordinated drive command.

[0148] In summary, the motion control method for the humanoid lower limb exoskeleton robot described in the above embodiments employs a fully analytical solution process, rather than numerical iteration. This fundamentally avoids potential problems such as matrix singularities and iteration non-convergence during the solution process, ensuring the algorithm's stability and robustness across the entire workspace. Secondly, the analytical solution can obtain the accurate theoretical truth value in one step, significantly improving the accuracy of robot motion control compared to the approximate solutions of numerical methods, enabling the feet to more accurately reproduce the desired trajectory. Furthermore, deterministic algebraic operations replace the time-consuming iterative process, resulting in a constant and extremely low computational cost per solution, fully meeting the stringent real-time requirements of the humanoid lower limb exoskeleton robot's control system.

[0149] According to a second aspect of this application, a motion control device 300 for a humanoid lower limb exoskeleton robot is provided. Each lower limb of the humanoid lower limb exoskeleton robot includes six joints, and the rotation axes of the third, fourth, and fifth joints are parallel to each other. The motion control device 300 includes: a target pose receiving unit 310, used to receive the desired target pose of the foot relative to the body coordinate system from a host computer, and perform homogeneous coordinate transformation on it to obtain a target pose matrix, the desired target pose including a desired position and a desired attitude angle; and a joint analytical solution unit 320, used to perform real-domain analysis on the first joint variables based on a preset set of link geometric parameters and the target pose matrix. The system analyzes and solves the target angle of the first joint to obtain the target angle of the second joint. The spatially associated joint solving unit 330 is used to solve the non-parallel axis related elements in the target pose matrix based on the target angle of the first joint to obtain the set of spatially associated joint angles, which includes the target angles of the second and sixth joints. The parallel axis group joint determining unit 340 is used to determine the set of parallel axis group joint angles based on the constraints of elements related to the parallel axes in the target pose matrix and in combination with the target angle of the first joint and the set of spatially associated joint angles. The driving signal generating unit 350 is used to convert the target angle of the first joint, the set of spatially associated joint angles, and the set of parallel axis group joint angles into corresponding pulse width modulation signals to obtain the six-axis cooperative driving command.

[0150] It should be noted that the motion control device for the humanoid lower limb exoskeleton robot in this application embodiment is similar in principle to the aforementioned motion control method for the humanoid lower limb exoskeleton robot. Therefore, the implementation process, implementation principle, and beneficial effects of the motion control device for the humanoid lower limb exoskeleton robot can be found in the description of the implementation process, implementation principle, and beneficial effects of the aforementioned method, and will not be repeated.

[0151] According to embodiments of this application, this application also provides an electronic device and a readable storage medium.

[0152] Figure 4 A schematic block diagram of an example electronic device 400 that can be used to implement embodiments of this application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the application described and / or claimed herein.

[0153] like Figure 4 As shown, the electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. The RAM 403 may also store various programs and data required for the operation of the electronic device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0154] Multiple components in electronic device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0155] The computing unit 401 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as the motion control method for a humanoid lower limb exoskeleton robot. For example, in some embodiments, the motion control method for a humanoid lower limb exoskeleton robot can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of the motion control method for a humanoid lower limb exoskeleton robot described above can be performed. Alternatively, in other embodiments, computing unit 401 may be configured to perform motion control methods for a humanoid lower limb exoskeleton robot by any other suitable means (e.g., by means of firmware).

[0156] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transferring data and instructions to the storage system, the at least one input device, and the at least one output device.

[0157] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0158] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0159] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0160] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0161] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0162] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this application can be achieved, and this is not limited herein.

[0163] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0164] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A motion control method for a humanoid lower limb exoskeleton robot, wherein each lower limb of the humanoid lower limb exoskeleton robot comprises six joints, the six joints being, from top to bottom: a first joint serving as a hip abduction / adduction joint, a second joint serving as a hip internal rotation / external rotation joint, a third joint serving as a hip flexion / extension joint, a fourth joint serving as a knee flexion / extension joint, a fifth joint serving as an ankle plantar flexion / dorsiflexion joint, and a sixth joint serving as an ankle inversion / eversion joint, wherein the rotation axes of the third, fourth, and fifth joints are parallel to each other, characterized in that... include: S1: Receive the desired target pose of the foot relative to the body coordinate system from the host computer, and perform homogeneous coordinate transformation on it to obtain the target pose matrix. The desired target pose includes the desired position and the desired attitude angle. S2: Based on the preset set of link geometry parameters and target pose matrix, perform real-domain analytical solution on the first joint variables to obtain the target angle of the first joint; S3: Based on the first joint target angle, solve the non-parallel axis related elements in the target pose matrix to obtain the spatial associated joint angle set, which includes the second joint target angle and the sixth joint target angle. S4: Based on the constraints of elements related to the parallel axis in the target pose matrix and combined with the target angle of the first joint and the spatially associated joint angle set, determine the joint angle set of the parallel axis group; S5: Convert the target angle of the first joint, the set of spatially associated joint angles, and the set of joint angles of the parallel axis group into corresponding pulse width modulation signals to obtain the six-axis cooperative drive command.

2. The motion control method for a humanoid lower limb exoskeleton robot according to claim 1, characterized in that, Step S1 includes: Based on the preset rotation transformation order, the received desired attitude angle is converted into a three-dimensional orthogonal rotation matrix. The rotation matrix is ​​placed in the upper left corner of the fourth-order homogeneous transformation matrix. The desired position coordinates are filled into the first three rows of the fourth column of the matrix, and the fourth row of the matrix is ​​set to a constant value of homogeneous coordinates to assemble the target pose matrix. The target pose matrix is ​​deconstructed using column vectors. The first column vector of the target pose matrix is ​​extracted as the normal vector, the second column vector as the orientation vector, the third column vector as the proximity vector, and the fourth column vector as the position vector.

3. The motion control method for a humanoid lower limb exoskeleton robot according to claim 1, characterized in that, The preset set of link geometry parameters includes the lengths of each link. Step S2 includes: Based on the equivalence of the coordinate system transformation matrix, a system of equations containing the first joint variable and the second joint variable is established, and the second joint variable is eliminated by using trigonometric identities to obtain the single-variable trigonometric function equations with respect to the first joint variable. By using the half-angle tangent substitution, the single-variable trigonometric function equation is transformed into a quartic equation in terms of the half-angle tangent value of the first joint variable. Solve the quartic equation in the real field to obtain at least one real root, and use inverse trigonometric functions to restore the real root to obtain the target angle of the first joint.

4. The motion control method for the humanoid lower limb exoskeleton robot according to claim 3, characterized in that, A single-variable trigonometric equation includes a quadratic sine term, a quadratic cosine term, a product of sine and cosine terms, a linear sine term, a linear cosine term, and a constant term for the first joint variable. Before transforming the single-variable trigonometric function equation into a quartic equation in terms of the half-tangent value of the first joint variable using the half-tangent substitution, the process also includes: Extract the position vector and proximity vector from the target pose matrix; Based on the position vector, proximity vector, and link geometry parameter set, six intermediate coefficients corresponding to the quadratic sine term, quadratic cosine term, product of sine and cosine term, linear sine term, linear cosine term, and constant term are determined through algebraic operations to construct a single-variable trigonometric function equation.

5. The motion control method for a humanoid lower limb exoskeleton robot according to claim 2, characterized in that, Step S3 includes: Based on the equivalence constraint of the coordinate transformation matrix and combined with the elements of the first joint target angle and the target pose matrix, the second joint target angle is calculated. The target angle of the sixth joint is calculated by using the geometric relationship between the components of the normal vector in the base coordinate system and the projection of the target angle of the first joint. The target angles of the second joint and the target angles of the sixth joint are combined and packaged to obtain a spatially related set of joint angles.

6. The motion control method for a humanoid lower limb exoskeleton robot according to claim 1, characterized in that, Step S4 includes: Based on the target angle of the first joint and the spatially associated joint angle set, the base coordinate pose is transformed to the second joint coordinate system, and the target angle of the third joint is calculated. Based on the target angle of the third joint and the closed-loop geometric relationship of the link, the target angle of the fourth joint is calculated. The pitch attitude components after the target pose matrix transformation are extracted using the parallel axis characteristics to determine the total pitch angle. The target angle of the fifth joint is calculated by successively subtracting the target angles of the third and fourth joints from the total pitch angle. The target angles of the third joint, the fourth joint, and the fifth joint are combined and packaged to obtain the set of joint angles of the parallel axis group.

7. The motion control method for a humanoid lower limb exoskeleton robot according to claim 1, characterized in that, Step S5 includes: Based on the target angle of the first joint, the set of spatially related joint angles, and the set of joint angles of the parallel axis group, construct the target joint angle vector; Based on a preset set of motor calibration parameters, linear mapping calculations are performed on each element in the target joint angle vector to obtain servo pulse width data; The servo pulse width data is frame-encapsulated to obtain six-axis collaborative drive commands, which are then sent to the servo controllers of each joint of the lower limb exoskeleton robot.

8. The motion control method for a humanoid lower limb exoskeleton robot according to claim 7, characterized in that, Based on the target angle of the first joint, the spatially associated joint angle set, and the parallel axis group joint angle set, a target joint angle vector is constructed, including: Based on the physical identification sequence of the motors of the lower limb exoskeleton robot, the target angle of the first joint, the set of spatially associated joint angles, and the set of parallel axis group joint angles are vectorized and recombined to obtain the initial target joint angle vector. Based on a preset coupling constraint coefficient matrix, the state-dependent constraint relationship between adjacent joints is calculated using an affine transformation algorithm to obtain the dynamic constraint boundary vector. The gradient correction angle vector is determined based on the dynamic constraint boundary vector and the potential field gain parameter. The target joint angle vector is obtained by performing element-wise numerical minimization verification on the gradient correction angle vector and the dynamic constraint boundary vector.

9. A motion control device for a humanoid lower limb exoskeleton robot, characterized in that, The humanoid lower limb exoskeleton robot comprises six joints in each lower limb, which, from top to bottom, are: a first joint serving as a hip abduction / adduction joint, a second joint serving as a hip internal rotation / external rotation joint, a third joint serving as a hip flexion / extension joint, a fourth joint serving as a knee flexion / extension joint, a fifth joint serving as an ankle plantar flexion / dorsiflexion joint, and a sixth joint serving as an ankle inversion / eversion joint. The rotation axes of the third, fourth, and fifth joints are parallel to each other. The device includes: The target pose receiving unit is used to receive the desired target pose of the foot relative to the body coordinate system sent by the host computer, and perform homogeneous coordinate transformation on it to obtain the target pose matrix. The desired target pose includes the desired position and the desired attitude angle. The joint analytical solution unit is used to perform real-domain analytical solution of the first joint variables based on a preset set of link geometric parameters and target pose matrix to obtain the target angle of the first joint; The spatially associated joint solving unit is used to solve the non-parallel axis related elements in the target pose matrix based on the first joint target angle to obtain the spatially associated joint angle set, which includes the second joint target angle and the sixth joint target angle. The parallel axis group joint determination unit is used to determine the joint angle set of the parallel axis group based on the constraints of elements related to the parallel axis in the target pose matrix and in combination with the target angle of the first joint and the spatially associated joint angle set. The drive signal generation unit is used to convert the target angle of the first joint, the set of spatially associated joint angles, and the set of joint angles of the parallel axis group into corresponding pulse width modulation signals to obtain six-axis cooperative drive commands.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-8.

Images