A robot lower limb training method

Abstract

The present application belongs to the technical field of lower limb training robot, and specifically discloses a robot lower limb training method, comprising the following steps: step 1, fitting a hip joint rotation angle change curve, a center of gravity up and down position change curve and a center of gravity forward speed change curve according to gait information of a human body; step 2, using a gravity compensation PD control algorithm to control rotation of an exoskeleton motor according to the hip joint rotation angle change curve; using a PD control algorithm with weight reduction bias to control rotation of a lifting motor according to the center of gravity up and down position change curve; and using a PI control algorithm to control rotation of a bottom wheel according to the center of gravity forward speed change curve; the present application provides a robot and a lower limb training method capable of improving human-computer interaction compliance.

Description

Technical Field

[0001] This invention belongs to the field of lower limb training technology, and specifically relates to a robot lower limb training method. Background Technology

[0002] With the increasing aging of society, more and more elderly people are losing their ability to stand independently due to stroke, cardiovascular disease, and other reasons, causing great inconvenience to their daily lives and severely reducing their quality of life, thus requiring lower limb rehabilitation training. Traditional wheeled lower limb rehabilitation robots can only provide support for patients, who then drive the robot to move. The robot cannot provide auxiliary force to the patient's lower limb joints and muscles during rehabilitation training. For patients with poor hip flexion or joint abnormalities such as knee hyperextension or foot drop, the robot cannot help improve gait or provide effective training. When using a robot to passively guide a patient to walk, it is difficult to accurately control the human body's movement trajectory, resulting in low compliance between the human and the machine and low comfort. Moreover, traditional leg-type lower limb rehabilitation robots often require the use of an electric treadmill or smart cane. Leg-type lower limb rehabilitation robots with electric treadmills are bulky, and rehabilitation training is limited by space, which contradicts the trend of community-based and home-based rehabilitation training, and they are often expensive. Summary of the Invention

[0003] The purpose of this invention is to provide a robot capable of passively walking with its legs and a lower limb training method, which can improve the compliance of human-computer interaction.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A method for training the lower limbs of a robot, the robot including a wheeled mobile platform, the wheeled mobile platform including base wheels, the wheeled mobile platform connected to a lifting mechanism, the lifting mechanism including a lead screw, and a lifting motor connected to the lead screw; the lifting mechanism is connected to a pelvic mechanism, the pelvic mechanism is connected to an exoskeleton mechanical leg, the exoskeleton mechanical leg including an exoskeleton motor; including the following steps:

[0006] Step 1: Fit curves of hip joint rotation angle change, center of gravity vertical position change, and center of gravity forward speed change based on the human gait information.

[0007] Step 2: Based on the hip joint rotation angle change curve, use the gravity-compensated PD control algorithm to control the rotation of the exoskeleton motor; based on the center of gravity vertical position change curve, use the PD control algorithm with weight reduction bias to control the rotation of the lifting motor; based on the center of gravity forward speed change curve, use the PI control algorithm to control the rotation of the bottom wheel.

[0008] Furthermore, in step 2, the method for controlling the rotation of the bottom wheel is as follows: calculate the forward speed of the bottom wheel according to the following formula, then calculate the desired rotation speed of the bottom wheel based on the desired forward speed of the bottom wheel, and control the bottom wheel to rotate according to the desired rotation speed;

[0009] ω=K pDL ·e 底轮 +K iDL ·∫e 底轮 dt

[0010] In the formula, ω is the desired forward speed of the bottom wheel, and K pDL K is the proportionality coefficient. iDL e is the integral coefficient. 底轮 ω is the difference between the value of the curve of the forward velocity of the center of gravity at the previous moment and the actual forward velocity of the center of gravity at the previous moment. When calculating ω, first determine whether ω(t-1) has exceeded the limit value. If it has exceeded the limit value, only the negative deviation is accumulated. If it has not exceeded the limit, the positive deviation is accumulated.

[0011] Furthermore, in step 2, the method for controlling the rotation of the exoskeleton motor is as follows: calculate the torque of the exoskeleton motor according to the following formula, and control the exoskeleton motor to rotate according to the torque;

[0012]

[0013] In the formula, e 单关节 K represents the difference between the value of the hip joint rotation curve at the previous moment and the actual hip joint angle at the previous moment. pDGJ K is the proportionality coefficient. dDGJ These are the differential coefficients. This is an estimate of the gravitational torque.

[0014] Furthermore, in step 2, the method for controlling the rotation of the lifting motor is as follows: calculate the torque of the lifting motor according to the following formula, and control the lifting motor to rotate according to the torque;

[0015]

[0016] In the formula, L0 is the ball screw lead, F y To reduce gravity, m t Let g be the mass of the pelvic structure, and g be the acceleration due to gravity. The desired acceleration at the connection point between the pelvic structure and the hip joint. e represents the desired velocity at the connection point between the pelvic structure and the hip joint. 升降 The value of the curve showing the change in the center of gravity's vertical position at the previous moment is the difference between the actual vertical position of the center of gravity at the previous moment, μ is the coefficient of friction of the ball screw, and K is the coefficient of friction. pSJ K is the proportionality coefficient. dSJ is the differential coefficient.

[0017] Furthermore, in step 2, the method for controlling the rotation of the bottom wheel is as follows: adjust the parameters through the coordination verification algorithm, calculate the forward speed of the bottom wheel according to the following formula, then calculate the expected rotation speed of the bottom wheel based on the expected forward speed of the bottom wheel, and control the bottom wheel to rotate according to the expected rotation speed;

[0018]

[0019]

[0020] ω=K pDL ·e 底轮 +K iDL ·∫e 底轮 dt

[0021] In the formula, ω is the desired forward speed of the bottom wheel, and K pDL K is the proportionality coefficient. iDL ω is the integral coefficient. d The value of y is the curve showing the change in velocity of the center of gravity forward. d足 To determine the desired foot position, calculate the desired foot position y using the velocity curve along the direction of the center of gravity's forward movement. d足 y 足 For the actual foot position, K 协调 The adjustment coefficient is negative. When calculating ω, first determine whether ω(t-1) has exceeded the limit. If it has exceeded the limit, only the negative deviation is accumulated. If it has not exceeded the limit, the positive deviation is accumulated.

[0022] Furthermore, in step 2, the method for controlling the rotation of the lifting motor is as follows: First, calculate the target damping and target stiffness:

[0023]

[0024] In the formula, F ext For the human-computer interaction force in the previous moment, B d For target damping, K d Let B0 and K0 be the target stiffness, and v be the initial admittance parameters. hip k is the average angular velocity of the exoskeleton's mechanical leg. b With k f For the corresponding adjustment coefficient;

[0025] Then, using the difference between the previous moment's center of gravity vertical position change curve value and the actual position, and the difference between the previous moment's center of gravity vertical velocity change curve value and the actual velocity, the input values ​​of the PD control algorithm at the current moment are updated. and y e :

[0026] Finally, the torque of the lifting motor is calculated, and the lifting motor is controlled to rotate according to the torque.

[0027]

[0028] In the formula, τ y L0 is the torque of the lifting motor, L0 is the lead of the ball screw, and F is the torque of the lifting motor. y To reduce gravity, m t Let g be the mass of the pelvic structure, and g be the acceleration due to gravity. The desired acceleration at the connection point between the pelvic structure and the hip joint. Let μ be the desired velocity at the connection point between the pelvic mechanism and the hip joint, and K be the coefficient of friction of the ball screw. pSJ K is the proportionality coefficient. dSJ is the differential coefficient.

[0029] Furthermore, in step 2, the method for updating the input value of the PD control algorithm at the current moment is as follows: [The following equation is used to update the input value of the PD control algorithm at the current moment]. Integrate to obtain the input value of the PD control algorithm at the current moment. and y e ;

[0030]

[0031] In the formula, M d Let y be the target inertia matrix. e-1 This represents the difference between the curve value of the change in the vertical position of the center of gravity at the previous moment and the actual position. This is the difference between the value of the velocity change curve of the center of gravity at the previous moment and the actual velocity.

[0032] Furthermore, in step 1, the method for fitting the hip joint rotation change curve is as follows: by combining the hip joint rotation change data collected from gait cycle analysis, the hip joint rotation change curve is fitted using a sine function and a specific form.

[0033]

[0034] In the formula, f(T,h) is a function of gait period and height, where T is the gait period, h is the height, and a is the height. i b i and c i is a constant term, and K is the number of sine functions.

[0035] Furthermore, in step 1, the method for fitting the curve of the vertical position change of the center of gravity is as follows: by analyzing the collected data on the vertical displacement of the human body's center of gravity, the relationship between the collected amplitude of the body's center of gravity fluctuation and the gait cycle is fitted, and a sine function curve is used to establish the curve of the vertical position change of the center of gravity.

[0036]

[0037] In the formula, A is the amplitude of the sine function, T is the gait period, and t is the walking time. This is the initial phase.

[0038] Furthermore, in step 1, the method for fitting the curve of the change in the center of gravity velocity is as follows: by analyzing the collected data on the change in the forward velocity of the human body's center of gravity, a curve of the change in the forward velocity of the center of gravity is fitted.

[0039]

[0040] In the formula: B(T,h) and v avg (T,h) is a function of gait period T and height h, B(T,h) is the amplitude of the sine function, v avg (T,h) represents the curve v comx The mean.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] This invention collects human gait data, establishes standard human gait hip joint angles, vertical displacement of the center of gravity, and forward and backward displacement curves, and uses a trajectory tracking control algorithm combined with PID control method for control. It uses an exoskeleton mechanical leg to drive the patient's legs to walk, and adapts to the pelvic mechanism and human walking trajectory through a wheeled platform, so that the robot can actively drive the patient to train. The control method is simple, convenient, efficient and accurate.

[0043] By employing a coordination verification algorithm between an omnidirectional wheeled mobile platform and a single-joint exoskeleton mechanical leg, a freely adjustable weight reduction system (lifting mechanism) and a single-joint exoskeleton mechanical leg, and a variable parameter admittance control algorithm to address the issue of poor compliance in human-computer interaction due to the inaccurate tracking of the vertical trajectory of the gait center of gravity, the wheel-leg composite motion unit is fully integrated to ensure the patient's body stability and the compliance of human-computer interaction during walking training.

[0044] Since the controlled object is the thigh angle, the lower leg cannot be controlled in practice. However, the distance a person walks is closely related to the foot position. Using the built-in lidar sensor and gait data calculation algorithm, the device determines the patient's foot position by obtaining the real-time position of the lower leg near the ankle. Based on the relationship between the wheeled mobile platform and the actual leg movement, the input error value of the bottom wheel PI controller is modified. The input PI error is the error between the desired foot position and the actual foot position multiplied by an adjustment coefficient. The adjustment coefficient is negative. That is, when the actual displacement in the forward direction of the center of gravity is less than the predefined displacement, the actual error is transformed into a coordination error. In the tracking controller, this manifests as the actual position exceeding the predefined tracking trajectory position, which reduces the output speed to ensure tracking and thus coordination. Conversely, when the actual displacement in the forward direction of the center of gravity is greater than the predefined displacement, the actual error is transformed into a coordination error. In the tracking controller, this manifests as the actual position lagging behind the predefined tracking trajectory position, which increases the output speed to ensure tracking and thus ensure the coordinated movement of the human body system.

[0045] The variable parameter admittance algorithm adjusts the damping coefficient (Bd) and spring coefficient (Kd) to increase system responsiveness as leg speed increases and the frequency of the body's center of gravity displacement curve increases. Conversely, when the human-machine interaction force is excessive, the system stiffness parameter is increased to improve accuracy and prevent the trajectory error from escalating and causing incoordination between the weight-reduction system and other motion units. However, the variable parameter admittance control adjustment suffers from poor compliance during human-machine interaction due to the non-rigid connection between the pelvic passive support mechanism and the user, resulting in uncontrolled movement of the user's feet and lower legs. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the overall structure of Embodiment 1 of the present invention;

[0047] Figure 2 This is a flowchart of the PI controller with anti-saturation design in Embodiment 1 of the present invention;

[0048] Figure 3 This is a schematic diagram of a single-link model of the exoskeleton mechanical leg in Embodiment 1 of the present invention;

[0049] Figure 4 This is a flowchart of the method for controlling the exoskeleton motor in Embodiment 1 of the present invention;

[0050] Figure 5 This is a simplified model diagram of the weight reduction system according to Embodiment 1 of the present invention;

[0051] Figure 6 This is a flowchart of the method for controlling the bottom wheel motor in Embodiment 2 of the present invention;

[0052] Figure 7 This is a flowchart of the method for controlling the lifting motor in Embodiment 2 of the present invention;

[0053] Figure 8 This is a schematic diagram of the kinematic model of the wheeled mobile platform in Embodiment 3 of the present invention.

[0054] In the diagram: 1. Host computer tablet; 2. Lifting column; 3. Pelvic mechanism; 4. Exoskeleton mechanical leg; 5. Control cabinet; 6. Mobile platform. Detailed Implementation

[0055] Example 1

[0056] A method for training the lower limbs of a robot, such as Figure 1 As shown, the robot comprises three motion units: a wheeled mobile platform 6, which includes two longitudinally arranged support platforms, each connected to a base wheel. A horizontal control cabinet 5 is fixed to the front end of the two support platforms. A vertical lifting mechanism is connected to the middle of the control cabinet 5. The lifting mechanism includes a hollow lifting column 2 fixed to the control cabinet 5, with a ball screw inside the column 2 connected to a lifting motor. The ball screw is connected to a pelvic mechanism 3 via a screw nut. The pelvic mechanism 3 provides full-degree-of-freedom passive support and includes two longitudinal support arms, each connected to an exoskeleton mechanical leg 4. Each exoskeleton mechanical leg 4 includes an exoskeleton motor, and each exoskeleton motor is connected to a leg connector. The lifting mechanism only controls the movement of the body's center of gravity along the vertical axis; translational and torsional movements in the horizontal plane are passive degrees of freedom, achieved passively compliant by the parallel four-bar linkage and its internal springs in the pelvic mechanism 3. A host computer tablet 1 is mounted on the top of the lifting column 2. The host computer tablet 1 is used to set initial gait parameters and training parameters, including stride length, stride speed, weight reduction force, and training time. The robot includes sensors and a preprocessing module for collecting position and velocity information of the motion units.

[0057] The robot includes a predefined trajectory generation module, a basic trajectory tracking module, and a motion unit coordination and verification module. The predefined trajectory generation module generates a predefined trajectory based on a built-in gait trajectory algorithm and initial gait parameters. The basic trajectory tracking module calculates the position and velocity of the motion units based on the predefined trajectory, set parameters, and the physical model of the motion units, and then assigns these calculations to the respective drive motors. The motion unit coordination and verification module, based on the wheeled mobile platform 6 and the single-joint exoskeleton, uses gait cycle theory to correct the speed of the two motion units in conjunction with gait data while maintaining trajectory tracking control. It also corrects the speed of the two motion units in conjunction with gait data while maintaining the vertical height change of the center of gravity and the hip joint angle change in conjunction with the lifting mechanism and the single-joint exoskeleton. Furthermore, it employs variable parameter admittance control to adjust the compliance of human-machine interaction, avoiding poor compliance issues caused by the non-rigid connection between the passive pelvic support mechanism and the user, and by the lack of control over the user's feet and lower legs during human-machine interaction. This coordinates the relationship between the bottom wheels and the mechanical legs, and the relationship between the mechanical legs and the pelvis, thereby improving compliance.

[0058] During training, the patient stands between two support platforms, connects their thighs to the leg connectors, and uses a safety belt to connect their pelvis to the two support arms. The rotation of each motor is controlled according to set parameters, specifically including the following steps:

[0059] Step 1 involves collecting human gait information. This is primarily achieved by using the angle sensors and motor encoders of the wheel-leg hybrid lower limb rehabilitation robot to acquire hip joint motion information (angular velocity) during walking. Sensors are also used to obtain information on the forward and backward movement of the center of gravity, as well as its vertical movement. A mathematical model of human gait is established by dividing the gait cycle into phases, and then fitted with gait information from normal subjects to curves showing changes in hip joint angle, vertical position of the center of gravity, and forward velocity.

[0060] Step 101, the method for fitting the hip joint rotation change curve is as follows: By combining the hip joint rotation change data collected from gait cycle analysis, the hip joint rotation change curve is fitted using a sine function and a specific form.

[0061]

[0062]

[0063] In the formula, f(T,h) is a function of gait period and height, where T is the gait period, h is the height, and a is the height. i b i and c i V is a constant term, and K is the number of sine functions; c For average walking speed, L s The step size.

[0064] Multiple linear regression analysis was used to analyze the relationship between hip joint rotation amplitude and height and gait cycle, and a regression model for hip joint rotation amplitude was established.

[0065] f(T,h)=β0+β1T+β2h (3)

[0066] The influence factor β was estimated using the least squares method.

[0067]

[0068] Based on the specific gait information data, the final mathematical model of hip joint rotation is as follows:

[0069]

[0070] In the formula, coefficients a1 = 0.0879 and a2 = 0.0281.

[0071] Step 102, the method for fitting the curve of the vertical position change of the center of gravity is as follows: By analyzing the collected data on the vertical displacement of the human body's center of gravity, the relationship between the collected amplitude of the body's center of gravity fluctuation and the gait cycle is fitted, and a sine function curve is used to establish the curve of the vertical position change of the center of gravity.

[0072]

[0073] In the formula, A is the amplitude of the sine function, T is the gait period, and t is the walking time. This is the initial phase.

[0074] Step 103, the method for fitting the curve of the change in the center of gravity velocity is as follows: By analyzing the collected data on the change in the forward velocity of the human body's center of gravity, the curve of the change in the forward velocity of the center of gravity is fitted using the following formula:

[0075]

[0076] In the formula: B(T,h) and v avg (T,h) is a function of gait period T and height h, B(T,h) is the amplitude of the sine function, v avg (T,h) represents the curve v comx The mean.

[0077] By combining the collected data on the average forward velocity of the body's center of gravity of subjects of different heights walking at different speeds, and using the same analytical method as that used to calculate the amplitude of the hip joint rotation, the function fitting result was obtained, as shown in the following formula:

[0078] v avg(T,h)=-0.007585*T*h+0.02421*h+0.8926*T-2.57 (8)

[0079] In the formula, T is the gait period and h is the height.

[0080] Step 2: Based on the hip joint rotation angle change curve, use the gravity-compensated PD control algorithm to control the rotation of the exoskeleton motor; based on the center of gravity vertical position change curve, use the PD control algorithm with weight reduction bias to control the rotation of the lifting motor; based on the center of gravity forward speed change curve, use the anti-integral saturation PI control algorithm to control the rotation of the bottom wheel.

[0081] Step 201, the method for controlling the rotation of the bottom wheel is as follows: calculate the forward speed of the bottom wheel according to the following expression of the PI controller, then calculate the desired rotation speed of the bottom wheel according to the desired forward speed of the bottom wheel, and control the bottom wheel to rotate according to the desired rotation speed;

[0082] ω=K pDL e 底轮 +K iDL ·∫e 底轮 dt (9)

[0083] In the formula, ω is the desired forward speed of the bottom wheel, and K pDL K is the proportionality coefficient. iDL e is the integral coefficient. 底轮 It is the difference between the value of the curve showing the change in the center of gravity's forward velocity at the previous moment and the actual forward velocity of the center of gravity at the previous moment.

[0084] like Figure 2 As shown, the limit-based attenuation method is used to address the integral saturation of the controller. The controller includes an output limiting module. This module does not generate any new signals; it only attenuates the integral of the controller's output after it enters the saturation region (reaching the upper limit of the error). In the discrete system corresponding to the PLC, when calculating ω, it first determines whether ω at the previous moment has exceeded the limit value. If it has exceeded the limit value, only negative deviations are accumulated to reduce the deviation; if it has not exceeded the limit, positive deviations are accumulated.

[0085] Step 202, the method for controlling the rotation of the exoskeleton motor is as follows: Figure 3 As shown, a single-link model of the exoskeleton mechanical leg was constructed. The Lagrangian functional balance method was used to analyze the dynamics of the exoskeleton mechanical leg, and the relationship between hip joint torque and hip joint range of motion was obtained:

[0086]

[0087] In the formula, θ1 is the angle between the thigh link and the x-axis; m1 is the sum of the mass of the human lower limb thigh and the mass of the exoskeleton leg; d1 = l1 / 2, where l1 is the length of the exoskeleton leg, which is adjusted to match the length of the patient's thigh during training. Here, it is assumed that the mass distribution of the thigh link is uniform.

[0088] like Figure 4 As shown, the torque of the exoskeleton motor is calculated according to the following expression of the PD controller, and the exoskeleton motor is controlled to rotate according to the torque.

[0089]

[0090] In the formula, e 单关节 K represents the difference between the value of the hip joint rotation curve at the previous moment and the actual hip joint angle at the previous moment. pDGJ K is the proportionality coefficient. dDGJ These are the differential coefficients. To estimate the gravitational torque, we assume the human body moves completely along with the exoskeleton's mechanical legs, using an online estimation method, and assume the gravitational torque estimate is accurate.

[0091] Step 203, the method for controlling the rotation of the lifting motor is as follows: The weight reduction system includes a simplified model of the lifting mechanism and the fully free-degree-of-freedom passive pelvic support mechanism, and it is assumed that the motion of the human body's center of gravity on the vertical axis completely follows the robot's motion trajectory, resulting in the following dynamic equation:

[0092]

[0093] In the formula, τ y L0 is the joint torque, F is the ball screw lead, and F is the lead of the ball screw. y To reduce gravity, m t The total mass of the passive pelvic support mechanism that provides full degrees of freedom for the robot, where μ is the coefficient of friction of the ball screw.

[0094] like Figure 5 As shown, the weight reduction system only controls the movement of the body's center of gravity on the vertical axis. Translational and torsional movements in the horizontal plane are passive degrees of freedom, achieved through a parallel four-bar linkage and its internal springs within the fully free-degree-of-freedom pelvic passive support mechanism. The desired trajectory of the robot's end effector is tracked using a calculated torque method. A biased PD controller is employed, calculating the torque of the lifting motor according to the following expression for the PD controller, and controlling the lifting motor to rotate according to the torque.

[0095]

[0096] In the formula, L0 is the ball screw lead, F y To set the weight reduction force, m t Let g be the mass of the pelvic structure, and g be the acceleration due to gravity. The expected acceleration due to the vertical movement of the center of gravity. The desired speed of the vertical movement of the center of gravity. and e can be calculated from the curve of the change in the vertical position of the center of gravity. 升降 The value of the curve showing the change in the center of gravity's vertical position at the previous moment is the difference between the actual vertical position of the center of gravity at the previous moment, μ is the coefficient of friction of the ball screw, and K is the coefficient of friction. pSJ K is the proportionality coefficient. dSJ is the differential coefficient.

[0097] Example 2

[0098] Step 1 in this embodiment is the same as in embodiment 1. Step 202 in step 2 is the same as in embodiment 1. The other parts of step 2 differ from those in embodiment 2 as follows:

[0099] Step 201, as follows Figure 6 As shown, the method for controlling the rotation of the bottom wheel is as follows: the parameters are changed in real time through the coordination verification algorithm, the forward speed of the bottom wheel is calculated according to the following formula, the expected rotation speed of the bottom wheel is calculated based on the expected forward speed of the bottom wheel, and the bottom wheel is controlled to rotate according to the expected rotation speed.

[0100]

[0101]

[0102] ω=K pDL ·e 底轮 +K iDL ·∫e 底轮 dt (16)

[0103] In the formula, ω is the expected forward speed of the bottom wheel at the current moment, and K pDL K is the proportionality coefficient. iDL ω is the integral coefficient. d Let y be the value of the curve showing the change in the forward velocity of the center of gravity at the previous moment. d足 For the expected position of the previous moment, y 足 K represents the actual foot position at the previous moment. 协调 This is the adjustment coefficient. The distance traveled in the direction of the center of gravity's forward movement is calculated using the velocity curve of the center of gravity's forward movement, representing the distance the foot will travel to the desired position. The actual foot position is measured and calculated using sensors.

[0104] To address controller integral saturation, a limit-based attenuation method is employed. The controller includes an output limit module that does not generate any new signals; it only attenuates the controller's output after it enters the saturation region (reaching the upper limit of the error). In the discrete system corresponding to the PLC, when calculating ω, it first checks whether the ω at the previous moment has exceeded the limit value. If it has, only negative deviations are accumulated; if it has not exceeded the limit, positive deviations are accumulated.

[0105] Step 203, as follows Figure 7 As shown, the method for controlling the rotation of the lifting motor is as follows: Parameters are changed in real time through a coordination verification algorithm. First, the target damping and target stiffness are calculated.

[0106]

[0107] In the formula, F ext For the human-computer interaction force in the previous moment, B d For target damping, K d Let B0 and K0 be the target stiffness, and v be the initial admittance parameters. hip k is the average angular velocity of the exoskeleton's mechanical leg. b With k f For the corresponding adjustment coefficients, both the damping parameter B and the stiffness parameter K have fixed adjustment ranges, and all parameters are determined experimentally. The human-machine interaction force and the average angular velocity of the exoskeleton's mechanical leg are measured and calculated using sensors.

[0108] Then, in the following formula... Integrate and update the input value of the PD control algorithm at the current moment. and y e :

[0109]

[0110]

[0111] In the formula, M d Let y be the target inertia matrix. e-1 This represents the difference between the curve value of the change in the vertical position of the center of gravity at the previous moment and the actual position. This is the difference between the vertical velocity change curve value of the center of gravity at the previous moment and the actual velocity.

[0112] Finally, the torque of the lifting motor is calculated, and the lifting motor is controlled to rotate according to the torque.

[0113]

[0114] In the formula, τ y L0 is the torque of the lifting motor, L0 is the lead of the ball screw, and F is the torque of the lifting motor. y To reduce gravity, m tLet g be the mass of the pelvic structure, and g be the acceleration due to gravity. The desired acceleration at the connection point between the pelvic structure and the hip joint. Let μ be the desired velocity at the connection point between the pelvic mechanism and the hip joint, and K be the coefficient of friction of the ball screw. pSJ K is the proportionality coefficient. dSJ is the differential coefficient.

[0115] Example 3

[0116] The rest of this embodiment is the same as that of embodiment 2, except that when turning, the desired speed v of the human body's center of gravity in the sagittal plane is used. comx The desired velocities of the left and right drive wheels are calculated using the kinematics of a wheeled mobile platform. For example... Figure 8 As shown, the wheeled mobile platform model is established as a differential drive vehicle model:

[0117]

[0118] In the formula, v p The value of v is the curve representing the change in velocity of the center of gravity forward. L and v R These represent the speeds of the left and right drive wheels, respectively. The center point P is the midpoint of the line connecting the left and right drive wheels, L is the distance between the left and right drive wheels, and R is the turning radius.

Claims

1. A method for training a robot, the robot including a wheeled mobile platform, the wheeled mobile platform including bottom wheels, the wheeled mobile platform being connected to a lifting mechanism, the lifting mechanism including a lead screw, and further including a lifting motor connected to the lead screw; A lifting mechanism is connected to a pelvic mechanism, and the pelvic mechanism is connected to an exoskeleton mechanical leg, the exoskeleton mechanical leg including an exoskeleton motor; characterized by including the following steps: Step 1: Fit curves of hip joint rotation angle change, center of gravity vertical position change, and center of gravity forward speed change based on the human gait information. Step 2: Based on the hip joint rotation angle change curve, use the gravity-compensated PD control algorithm to control the rotation of the exoskeleton motor; based on the center of gravity vertical position change curve, use the PD control algorithm with weight reduction offset to control the rotation of the lifting motor; based on the center of gravity forward speed change curve, use the PI control algorithm to control the rotation of the bottom wheel. In step 2, the method for controlling the rotation of the lifting motor is as follows: calculate the torque of the lifting motor according to the following formula, and control the lifting motor to rotate according to the torque; In the formula, L0 is the ball screw lead, F y To reduce gravity, m t Let g be the mass of the pelvic structure, and g be the acceleration due to gravity. The desired acceleration at the connection point between the pelvic structure and the hip joint. e represents the desired velocity at the connection point between the pelvic structure and the hip joint. 升降 The value of the curve showing the change in the center of gravity's vertical position at the previous moment is the difference between the actual vertical position of the center of gravity at the previous moment, μ is the coefficient of friction of the ball screw, and K is the coefficient of friction. pSJ K is the proportionality coefficient. dSJ These are the differential coefficients; The method for controlling the rotation of the bottom wheel is as follows: calculate the forward speed of the bottom wheel according to the following formula, then calculate the desired rotation speed of the bottom wheel based on the desired forward speed of the bottom wheel, and control the bottom wheel to rotate at the desired rotation speed; ω=K pDL ·e 底轮 +K iDL ·∫e 底轮 dt In the formula, ω is the desired forward speed of the bottom wheel, and K pDL K is the proportionality coefficient. iDL e is the integral coefficient. 底轮 ω is the difference between the value of the curve of the forward velocity of the center of gravity at the previous moment and the actual forward velocity of the center of gravity at the previous moment. When calculating ω, first determine whether ω(t-1) has exceeded the limit value. If it has exceeded the limit value, only the negative deviation is accumulated. If it has not exceeded the limit, the positive deviation is accumulated.

2. The robot training method as described in claim 1, characterized in that, In step 2, the method for controlling the rotation of the exoskeleton motor is as follows: calculate the torque of the exoskeleton motor according to the following formula, and control the exoskeleton motor to rotate according to the torque; In the formula, e 单关节 K represents the difference between the value of the hip joint rotation curve at the previous moment and the actual hip joint angle at the previous moment. pDGJ K is the proportionality coefficient. dDGJ These are the differential coefficients. This is an estimate of the gravitational torque.

3. The robot training method as described in claim 1 or 2, characterized in that, In step 2, the method for controlling the rotation of the bottom wheel is as follows: calculate the forward speed of the bottom wheel according to the following formula, then calculate the desired rotation speed of the bottom wheel based on the desired forward speed of the bottom wheel, and control the bottom wheel to rotate according to the desired rotation speed; ω=K pDL ·e 底轮 +K iDL ·∫e 底轮 dt In the formula, ω is the desired forward speed of the bottom wheel, and K pDL K is the proportionality coefficient. iDL ω is the integral coefficient. d The value of y is the curve showing the change in velocity of the center of gravity forward. d足 For the desired foot position, y 足 For the actual foot position, K 协调 The adjustment coefficient is used. When calculating ω, it is first determined whether the ω at the previous moment has exceeded the limit value. If it has exceeded the limit value, only the negative deviation is accumulated. If it has not exceeded the limit value, the positive deviation is accumulated.

4. The robot training method as described in claim 1 or 2, characterized in that, In step 2, the method for controlling the rotation of the lifting motor is as follows: First, calculate the target damping and target stiffness: In the formula, F ext For the human-computer interaction force in the previous moment, B d For target damping, K d Let B0 and K0 be the target stiffness, and v be the initial admittance parameters. hip k is the average angular velocity of the exoskeleton's mechanical leg. b With k f For the corresponding adjustment coefficient; Then, using the difference between the previous moment's center of gravity vertical position change curve value and the actual position, and the difference between the previous moment's center of gravity vertical velocity change curve value and the actual velocity, the input values ​​of the PD control algorithm at the current moment are updated. and y e : Finally, the torque of the lifting motor is calculated, and the lifting motor is controlled to rotate according to the torque. In the formula, τ y L0 is the torque of the lifting motor, L0 is the lead of the ball screw, and F is the torque of the lifting motor. y To reduce gravity, m t Let g be the mass of the pelvic structure, and g be the acceleration due to gravity. The desired acceleration at the connection point between the pelvic structure and the hip joint. Let μ be the desired velocity at the connection point between the pelvic mechanism and the hip joint, and K be the coefficient of friction of the ball screw. pSJ K is the proportionality coefficient. dSJ is the differential coefficient.

5. The robot training method as described in claim 4, characterized in that, The method for updating the input value of the PD control algorithm at the current moment is as follows: For the following formula... Integrate to obtain the input value of the PD control algorithm at the current moment. and y e ; In the formula, M d Let y be the target inertia matrix. e-1 This represents the difference between the curve value of the change in the vertical position of the center of gravity at the previous moment and the actual position. This is the difference between the value of the velocity change curve of the center of gravity at the previous moment and the actual velocity.

6. The lower limb training method for a robot as described in claim 1, characterized in that, In step 1, the method for fitting the hip joint rotation change curve is as follows: by combining the hip joint rotation change data collected from gait cycle analysis, the hip joint rotation change curve is fitted using a sine function and a specific form. In the formula, f(T,h) is a function of gait period and height, where T is the gait period, h is the height, and a is the height. i b i and c i is a constant term, and K is the number of sine functions.

7. The robot training method as described in claim 1, characterized in that, In step 1, the method for fitting the curve of the vertical position change of the center of gravity is as follows: by analyzing the collected data on the vertical displacement of the human body's center of gravity, the relationship between the collected amplitude of the body's center of gravity fluctuation and the gait cycle is fitted, and a sine function curve is used to establish the curve of the vertical position change of the center of gravity. In the formula, A is the amplitude of the sine function, T is the gait period, and t is the walking time. This is the initial phase.

8. The robot training method as described in claim 1, characterized in that, In step 1, the method for fitting the curve of the change in the center of gravity velocity is as follows: by analyzing the collected data on the change in the forward velocity of the human body's center of gravity, a curve of the change in the forward velocity of the center of gravity is fitted. In the formula: B(T,h) and v avg (T,h) is a function of gait period T and height h, B(T,h) is the amplitude of the sine function, v avg (T,h) represents the curve v comx The mean.

Images