A waist exoskeleton robot adaptive assistance control method

Abstract

This invention provides an adaptive assistive control method for a lumbar exoskeleton robot. A hierarchical control system is designed, where an upper-level parameter generator generates parameters for the fully assisted state mode in a middle-level finite state machine based on the lumbar flexion angle and plantar pressure values ​​at the lowest point of bending for different wearers. The middle-level finite state machine sets four state modes based on the flexion-extension angle, angular velocity, and plantar pressure of the human lumbar spine. Effective transitions between these modes are possible. Each of the four state modes sends different control parameters to a lower-level force-position hybrid controller, which can achieve single or mixed force-position control of the exoskeleton actuator. This control method fully considers the characteristics of human-machine compatible lumbar assistive exoskeleton operation, is simple to implement, and allows for personalized adjustments of assistive parameters based on the actual working conditions of different individuals, greatly improving the control state diversity and human-machine coordination of the lumbar assistive exoskeleton.

Description

Technical Field

[0001] This invention belongs to the field of exoskeleton control technology and is an adaptive assist control method for a waist exoskeleton robot. Background Technology

[0002] In the industrial field, waist-assisted exoskeleton robots have become a hot research topic in recent years and have developed rapidly both domestically and internationally. Research areas include biomimetic structural design based on human kinematics, human motion intention recognition based on multiple sensors, and human-in-the-loop control strategies.

[0003] To achieve effective assistance, lumbar-assisted exoskeleton robots need to accurately recognize handling movements and then apply different control effects to each recognized movement. Human-machine compatible lumbar-assisted exoskeletons are structurally adaptable to human kinematics, but due to their structural flexibility, the control technology requirements are more complex than those of traditional rigid exoskeletons. Therefore, designing a control method that is sufficiently adaptive to changes in the human body and handling operations, and exhibits good coordination, is one of the most challenging tasks to be solved in the field of lumbar-assisted exoskeleton robot research.

[0004] Based on the underlying driving mechanism of exoskeleton actuators, commonly used lumbar assistive exoskeleton control methods can be divided into position control methods and force control methods. Position-based control methods connect actuators such as motors and hydraulic rods to specific mechanical structures or mechanisms to form complete execution units, allowing the lumbar assistive exoskeleton to accurately output different positions according to different human movement states. This method is mostly used for rigid lumbar assistive exoskeletons. For flexible lumbar assistive exoskeletons, the inherent elasticity and flexibility make it difficult for position control methods to accurately output positions. Furthermore, position control methods struggle to meet human-machine interaction coordination requirements, and are more likely to cause discomfort for the wearer.

[0005] Force-based control methods enable actuators to accurately output force or torque, which is then transmitted to the human body through specific structures or mechanisms, allowing the body to fully absorb the assistive effect. This method, by adding appropriate force compensation values ​​to the force trajectory, can eliminate the inherent elasticity and flexibility of the flexible structure itself, making it suitable for both rigid and flexible lumbar assistive exoskeletons. However, current pure force control methods are primarily used for fixed-track assistance during lumbar extension movements. In lumbar flexion movements, where assistance is not required, pure force control struggles to achieve the effect of actuator follow-up, leading to poor human-machine coordination during flexion. Furthermore, current exoskeleton control strategies for handling movements generally employ predefined trajectories or fixed-parameter control, lacking effective adaptive control that adapts to changes in the human body and the handling operation.

[0006] Chinese Patent 2022104159986 discloses a lumbar exoskeleton robot system and its assistive control method. This patent employs a finite state machine with priority judgment for assistive control of walking and lumbar flexion / extension movements during human handling. Specifically, it can assist in controlling the wearer's walking and lumbar extension movements as determined by the finite state machine. In the lumbar extension assistive control stage, an adaptive assist parameter determination method is designed based on the lowest point of the human body's flexion. However, this finite state machine solution lacks judgment of the lumbar flexion movement state and follow-up control of joint drives. The adaptive assist parameter determination method does not consider changes in the task, leading to problems of uneven and uncomfortable use for the wearer. Summary of the Invention

[0007] To address the issues of limited state modes and poor coordination in existing human-machine compatible lumbar assistive exoskeleton force control methods, this invention provides an adaptive assistive control method for lumbar exoskeleton robots. Based on the multi-state mode of finite state machines and the force-position hybrid control method driven by motor current loops, a three-layer control system is designed, which can effectively improve the diversity of control states and human-machine coordination of the lumbar exoskeleton, and reduce the complexity of the control strategy.

[0008] The technical solution to achieve the purpose of this invention is as follows:

[0009] An adaptive assist control method for a lumbar exoskeleton robot, specifically including:

[0010] Step 1: Installation of data acquisition equipment and execution unit, and information collection;

[0011] Step 2: Upper-level control system—Constructing the assist parameters for the fully assisted state mode;

[0012] Step 3: Mid-level control system—Constructing a finite state machine;

[0013] Step 4: Underlying Control System - Constructing a force-position hybrid controller driven by the motor current loop.

[0014] Furthermore, step 1 specifically includes:

[0015] Step 1.1: Installation design of inertial measurement unit, pressure sensor, actuator, and microcontroller;

[0016] Step 1.2: Collect individual motion data and work weight data;

[0017] The inertial measurement unit in the lower back is used to detect the Euler angles and angular velocities of the human body's lumbar flexion and extension movements. The inertial measurement unit in the thigh is used to detect the Euler angles and angular velocities of the human body's legs. The plantar pressure sensor detects the total weight. The motor output shaft needs to be fixedly connected to the end of the human-machine compatible exoskeleton actuator. The other end of the actuator is fixedly connected to the outer garment above the human body's back.

[0018] Furthermore, step 2 specifically includes:

[0019] Step 2.1: Define the fully assisted state:

[0020] The process from when the wearer bends over to lift the object to be moved into a ready state to when they return to an upright position is defined as the fully assisted process of the finite state machine of the middle-level control system. The assistance parameters of this process depend on the parameters determined by the upper-level parameter generator.

[0021] Step 2.2: Construct adaptive human motion assistance parameters;

[0022] The formula for determining the adaptive human motion assistance parameters is as follows:

[0023]

[0024] In the formula, K pLowest To adapt to human motion assistance parameters, K ref To help establish a baseline value for the parameter, θ lowest The position of the lowest point of bending over during the wearer's carrying motion, recorded by the microcontroller;

[0025] Step 2.3: Construct adaptive task weight assist parameters;

[0026] The formula for the adaptive work weight assist parameter is as follows:

[0027]

[0028] In the formula, K pMax For adaptive operation weight assist parameters, F max F represents the maximum plantar pressure value recorded by the microcontroller after the wearer grasps an object and leaves the ground. ref The maximum plantar pressure value recorded by the microcontroller before the wearer grasps an object;

[0029] Step 2.4: Construct the final assist parameters;

[0030] The final assist parameters determined by the upper-level parameter generator are determined by the product of the adaptive human motion assist parameters and the adaptive work weight assist parameters, as shown in the following formula:

[0031] K pFinal =K pLowest K pMax

[0032] T1 = K pFinal (θ set -θ now )

[0033] In the formula, K pFinal This is the final assist parameter, which plays a role in the fully assisted mode. T1 is the assist torque value when the target is a normal upright position, and θ is the final assist parameter. set The target position is the normal upright position, θ now This is the current location.

[0034] Furthermore, in step 3, the middle-level control system uses a finite state machine to divide the human body into four state modes based on the angle and angular velocity of the lumbar and back flexion and extension and the plantar pressure: normal standing, zero-force tracking, assist preparation and full assistance. Each state mode sends different parameters to the lower-level control system, and the transition between state modes is based on the lumbar and back angle, angular velocity and plantar pressure.

[0035] Furthermore, the state mode, the threshold for transitioning to the next state, and the instructions sent to the underlying layer are as follows:

[0036] When standing normally, θ t >0°, the motor remains stationary;

[0037] During zero-force tracking, F t -F t-1 >f, the motor tracks the rotation angle of the lower back;

[0038] Help with preparation, ω t >ω, the motor outputs a smaller torque;

[0039] Full power, θ t ≤0°, the adjusted torque output of the motor.

[0040] Of the four state modes above, θ t For the wearer's real-time flexion angle of the lower back, F t and F t-1 These represent the wearer's current plantar pressure value and the previous plantar pressure value, respectively, ω t The real-time angular velocity of the wearer's lower back extension is given by f and ω, which are threshold values ​​that need to be set manually.

[0041] Furthermore, the specific steps for constructing a finite state machine are as follows:

[0042] Step 3.1: During initialization, the wearer needs to stand naturally and then power on the control system. The inertial measurement unit, plantar pressure sensor and brushless DC motor are initialized and put into working state.

[0043] Step 3.2: The finite state machine enters the normal standing state mode, in which the underlying control system keeps the motor stationary;

[0044] Step 3.3: Determine whether to enter the zero-force tracking mode based on whether the flexion angle of the human lower back is greater than 0°. If yes, proceed to step 3.4. Otherwise, maintain the normal standing mode and continue to judge.

[0045] Step 3.4: The finite state machine enters the zero-force tracking state mode. In this state mode, the underlying control system causes the motor actuator to perform lumbar and back flexion follow-up, that is, the motor actuator tracks the rotation angle of the human body's lumbar and back.

[0046] Step 3.5: Determine whether to enter the assisted preparation state mode based on whether the change value of plantar pressure is greater than a fixed threshold. If yes, proceed to step 3.6; otherwise, maintain the zero-force tracking state mode and continue to judge.

[0047] Step 3.6: The finite state machine enters the assist preparation state mode. In this state mode, the underlying control system causes the motor actuator to output a small initial torque value, giving the wearer sufficient adaptability and eliminating the slack caused by the flexible structure.

[0048] Step 3.7: Determine whether to enter the fully assisted state mode based on whether the angular velocity of the human lower back extension is greater than a fixed threshold. If yes, proceed to step 3.8; otherwise, maintain the assisted preparation state mode and continue to judge.

[0049] Step 3.8: The finite state machine enters the fully assisted state mode. In this state mode, the underlying control system causes the motor actuator to output the torque command that has been adjusted according to the upper controller, so as to provide the wearer with sufficient assistance.

[0050] Step 3.9: Determine whether to return to the normal standing position mode based on the extension angle of the human waist and back. If yes, proceed to step 3.2. Otherwise, maintain the fully assisted state mode and continue to judge.

[0051] Furthermore, in step 4, the underlying control system is a force-position hybrid controller based on motor current loop drive. Its direct target is a brushless DC motor driven by FOC current loop, achieving pure force control, pure position control, and simultaneous force-position control. Essentially, it enables the motor to accurately output the required torque. The formula for calculating the torque directly output by the brushless DC motor driven by FOC current loop is:

[0052] T output =K p (P set -P now )+K d (V set -V now )+Tset

[0053] In the formula, T output P represents the torque directly output by a brushless DC motor driven by the FOC current loop. set For the set target location, P now V represents the current position detected by the encoder. set For the set target speed, V now K is the current speed detected by the encoder. p K is the stiffness coefficient. d T is the damping coefficient. set The set feedforward torque.

[0054] Furthermore, in step 4, the underlying force-position hybrid controller receives different instruction parameters.

[0055] When standing normally, the motor remains stationary, K p =0,K d =0,T set =0;

[0056] During zero-force tracking, the motor tracks the rotation angle of the lower back, K p =k,P set =θ t ,K d =0,T set =0;

[0057] During the preparation for power assistance, the motor outputs a smaller torque, K p =0,K d =0,T set =τ;

[0058] When fully assisted, the motor outputs the adjusted torque, K. p =K pFinal ,P set =0,K d =0,T set =τ.

[0059] Furthermore, in step 4, the specific steps for constructing the control strategy of the force-position hybrid controller driven by the motor current loop are as follows:

[0060] Step 4.1: When the finite state machine is in the normal standing state mode, the force-position hybrid controller does not control the motor, so that the motor is in a passive stationary state;

[0061] Step 4.2: When the finite state machine is in zero-force tracking mode, the force-position hybrid controller performs pure position control on the motor and sends the position command as the human lumbar flexion angle detected by the inertial measurement unit.

[0062] Step 4.3: When the finite state machine is in the assist preparation state mode, the force-position hybrid controller performs pure force control on the motor and sends a force command with a small torque value.

[0063] Step 4.4: When the finite state machine is in the fully assisted state mode, the force-position hybrid controller performs simultaneous force and position control on the motor. The force command sent is the torque compensation value, which is used to eliminate the slack generated by the flexible structure of the exoskeleton. The position command sent is the initial angle, i.e., 0°. At this time, the stiffness coefficient is the final assist parameter K determined in step 2.4. pFinal .

[0064] Furthermore, in steps 4.2 and 4.4, if no overshoot oscillation occurs when the motor rotates a specific angle using the position control loop, then it is not necessary to set the damping coefficient and target speed. If overshoot oscillation occurs, then it is necessary to add a damping coefficient and set the target speed to 0.

[0065] K d =ξ,V set =0.

[0066] Furthermore, if the waist-assisted exoskeleton robot does not affect the human lower limbs during use, then the leg inertial measurement unit is not used, and there is no need to determine the walking state of the lower limbs. If the waist-assisted exoskeleton robot interacts with the human lower limbs during use, then the leg inertial measurement unit needs to be used and the walking state recognition function needs to be considered.

[0067] The significant advantages of this invention compared to existing technologies are:

[0068] This invention provides an adaptive assistive control method for a lumbar exoskeleton robot. A hierarchical control system is designed, where an upper-level parameter generator generates parameters for the fully assisted state mode in a middle-level finite state machine based on the lumbar flexion angle and plantar pressure values ​​at the lowest point of bending for different wearers. The middle-level finite state machine sets four state modes based on the flexion-extension angle, angular velocity, and plantar pressure of the human lumbar spine. Effective transitions between these modes are possible. Each of the four state modes sends different control parameters to a lower-level force-position hybrid controller, which can achieve single or mixed force-position control of the exoskeleton actuator. This control method fully considers the characteristics of human-machine compatible lumbar assistive exoskeleton operation, is simple to implement, and allows for personalized adjustments of assistive parameters based on the actual working conditions of different individuals, greatly improving the control state diversity and human-machine coordination of the lumbar assistive exoskeleton. Attached Figure Description

[0069] Figure 1 This is a flowchart of the steps of the lumbar exoskeleton control method provided by the present invention.

[0070] Figure 2This is a schematic diagram of the inertial measurement unit and foot pressure sensor worn on the human body.

[0071] Figure 3 This is a flowchart illustrating the specific implementation of the intermediate-level finite state machine provided by the present invention.

[0072] Figure 4 This is a block diagram of force-position hybrid control based on FOC current loop drive. Detailed Implementation

[0073] The following will refer to the appendices in the embodiments of the present invention. Figure 1-4 The technical solutions in the embodiments of the present invention are clearly and completely described herein. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0074] Combination Figure 1 The present invention provides an adaptive assist control method for a lumbar exoskeleton robot, comprising the following steps:

[0075] Step 1: Installation of data acquisition equipment and execution unit, and information collection;

[0076] Step 1.1: Installation design of inertial measurement unit, pressure sensor, actuator, and microcontroller;

[0077] See Figure 2 Before the human body performs a lifting action, an inertial measurement unit (IMU) is attached to the waist or back to collect the Euler angles and angular velocities of the waist and back in the sagittal plane during human movement. Two IMUs are attached to the left and right thighs respectively to collect the Euler angles and angular velocities of the legs in the sagittal plane. A pressure sensor is installed on the sole of the foot to collect the plantar pressure. The actuator is a brushless DC motor driven by an FOC current loop, which requires an integrated encoder to obtain its own rotation angle and angular velocity in real time. The microcontroller communicates with the IMU and foot pressure sensors via UART, and with the motor driver board via CAN.

[0078] Step 1.2: Collect individual motion data and work weight data;

[0079] When the human body is being moved, the natural upright position of the human body is defined as the zero position. The inertial measurement unit at the waist and back is used to detect the Euler angle and angular velocity of the human body's waist flexion and extension movements. The inertial measurement unit at the thigh is used to detect the Euler angle and angular velocity of the human body's legs. The plantar pressure sensor detects the total weight. The motor output shaft needs to be fixedly connected to the end of the human-machine compatible exoskeleton actuator. The other end of the actuator is fixedly connected to the outer garment above the human body's back.

[0080] Step 2: Upper-level control system—Constructing the assist parameters for the fully assisted state mode;

[0081] The upper control system is the parameter generator for the fully assisted process of the middle control system. It is responsible for determining the parameters of the fully assisted state mode in the finite state machine of the middle control system based on the individual motion data and work weight data of different wearable exoskeletons. The individual motion data is the lumbar flexion angle value at the lowest point of bending over detected by the inertial measurement unit, and the work weight data is the plantar pressure value at the lowest point of bending over detected by the plantar pressure sensor.

[0082] The specific steps are as follows:

[0083] Step 2.1: Define the fully assisted state:

[0084] The process from when the wearer bends over to lift the object to be moved into a ready state to when they return to an upright position is defined as the fully assisted process of the finite state machine of the middle-level control system. The assistance parameters of this process depend on the parameters determined by the upper-level parameter generator.

[0085] Step 2.2: Construct adaptive human motion assistance parameters;

[0086] The lumbar and back inertial measurement unit detects the Euler angles of the human body in the sagittal plane in real time. Euler angles in the lumbar and back flexion direction are defined as positive values, while Euler angles in the lumbar and back extension direction are defined as negative values. During the process of the wearer bending over from a normal upright position to grasping an object (zero-force tracking mode), the microcontroller records the position of the lowest point of bending as θ. lowest At this time θ lowest To achieve the maximum positive value, the formula for determining the adaptive human motion assistance parameters after the wearer grabs an object and triggers the assist preparation mode from the lowest point of bending over is as follows:

[0087]

[0088] In the formula, K pLowest To adapt to human motion assistance parameters, K ref To help establish a baseline value for the parameter, θ lowest The position of the lowest point of bending over during the wearer's carrying motion, recorded by the microcontroller.

[0089] Step 2.3: Construct adaptive task weight assist parameters;

[0090] The plantar pressure sensor detects the pressure on the human foot in real time. The plantar pressure value is positive. In zero-force tracking mode, the microcontroller records the maximum plantar pressure value F before the wearer grasps an object. ref After the wearer grabs an object and triggers the assist preparation mode from the lowest point of bending over, the microcontroller records the maximum plantar pressure value F after the wearer leaves the ground after grabbing the object.max The formula for determining the adaptive work weight assist parameter is as follows:

[0091]

[0092] In the formula, K pMax F is an adaptive operation weight assist parameter. max F represents the maximum plantar pressure value recorded by the microcontroller after the wearer grasps an object and leaves the ground. ref The maximum plantar pressure value recorded by the microcontroller before the wearer grasps an object.

[0093] Step 2.4: Construct the final assist parameters;

[0094] The final assist parameters determined by the upper-level parameter generator are determined by the product of the adaptive human motion assist parameters and the adaptive work weight assist parameters, as shown in the following formula:

[0095] K pFinal =K pLowest K pMax

[0096] T1 = K pFina l(θ set -θ now )

[0097] In the formula, K pFinal This is the final assist parameter, which plays a role in the fully assisted mode. T1 is the assist torque value when the target is a normal upright position, and θ is the final assist parameter. set The target position is the normal upright position, θ now This is the current location.

[0098] Step 3: Mid-level control system—Constructing a finite state machine;

[0099] See Figure 3 The middle-level control system employs a finite state machine, dividing the system into four state modes based on the angle and angular velocity of the human lumbar and back flexion and extension, as well as the plantar pressure: normal standing, zero-force tracking, assist preparation, and full assist. Each state mode sends different parameters to the lower-level control system. The transitions between state modes are based on the lumbar and back angle, angular velocity, and plantar pressure, as shown in Table 1. θ t For the wearer's real-time flexion angle of the lower back, F t and F t-1 These represent the wearer's current plantar pressure value and the previous plantar pressure value, respectively, ω t The real-time angular velocity of the wearer's lower back extension is represented by f and ω, which are force and angular velocity thresholds that need to be set manually, respectively. In a specific embodiment, f = 10 and ω = 60 can be selected.

[0100] Table 1. Status Modes and Corresponding Thresholds and Instructions

[0101]

[0102]

[0103] The specific steps for constructing a finite state machine are as follows:

[0104] Step 3.1: During initialization, the wearer needs to stand naturally and then power on the control system. The inertial measurement unit, plantar pressure sensor and brushless DC motor will be initialized and enter the working state.

[0105] Step 3.2: The finite state machine enters the normal standing state mode, in which the underlying control system keeps the motor stationary.

[0106] Step 3.3: Determine whether to enter the zero-force tracking mode based on whether the flexion angle of the human lower back is greater than 0°. If yes, proceed to step 3.4. Otherwise, maintain the normal standing mode and continue to judge.

[0107] Step 3.4: The finite state machine enters the zero-force tracking state mode. In this state mode, the underlying control system causes the motor actuator to perform lumbar and back flexion follow-up, that is, the motor actuator tracks the rotation angle of the human body's lumbar and back.

[0108] Step 3.5: Determine whether to enter the assist preparation state mode based on whether the change value of plantar pressure is greater than a fixed threshold. If yes, proceed to step 3.6; otherwise, maintain the zero-force tracking state mode and continue to make judgments.

[0109] Step 3.6: The finite state machine enters the assist preparation state mode. In this state mode, the underlying control system causes the motor actuator to output a small initial torque value, giving the wearer sufficient adaptability and eliminating the slack caused by the flexible structure.

[0110] Step 3.7: Determine whether to enter the fully assisted state mode based on whether the angular velocity of the human lower back extension is greater than a fixed threshold. If yes, proceed to step 3.8; otherwise, maintain the assisted preparation state mode and continue to judge.

[0111] Step 3.8: The finite state machine enters the fully assisted state mode. In this state mode, the underlying control system causes the motor actuator to output the torque command that has been adjusted according to the upper controller, so as to provide the wearer with sufficient assistance.

[0112] Step 3.9: Determine whether to return to the normal standing position mode based on the extension angle of the human waist and back. If yes, proceed to step 3.2. Otherwise, maintain the fully assisted state mode and continue to judge.

[0113] Step 4: Underlying Control System – Constructing a force-position hybrid controller driven by the motor current loop;

[0114] See Figure 4 The underlying control system is a force-position hybrid controller based on motor current loop drive. The direct target is a brushless DC motor based on FOC current loop drive. It can realize pure force control, pure position control and simultaneous force-position control. As can be seen from the following formula, the essence is to make the motor accurately output the required torque. The formula for calculating the torque directly output by the brushless DC motor driven by FOC current loop is:

[0115] T output =K p (P set -P now )+K d (V set -V now )+T set

[0116] In the formula, T output P represents the torque directly output by a brushless DC motor driven by the FOC current loop. set For the set target location, P now V represents the current position detected by the encoder. set For the set target speed, V now K is the current speed detected by the encoder. p K is the stiffness coefficient. d T is the damping coefficient. set The set feedforward torque.

[0117] The lower-level control system can receive parameters sent by the middle-level control system to achieve different execution effects, as shown in Table 2. τ is the set non-zero feedforward torque value, and k is the set non-zero stiffness value. In a specific embodiment, k = 5 and τ = 1 can be selected.

[0118] Table 2 Different command parameters received by the bottom-level force-position hybrid controller

[0119]

[0120] The specific steps for constructing the control strategy of a force-position hybrid controller driven by a motor current loop are as follows:

[0121] Step 4.1: When the finite state machine is in the normal standing state mode, the force-position hybrid controller does not control the motor, so that the motor is in a passive stationary state.

[0122] Step 4.2: When the finite state machine is in zero-force tracking mode, the force-position hybrid controller performs pure position control on the motor and sends the position command as the human lumbar flexion angle detected by the inertial measurement unit.

[0123] Step 4.3: When the finite state machine is in the assist preparation state mode, the force-position hybrid controller performs pure force control on the motor and sends a force command with a small torque value.

[0124] Step 4.4: When the finite state machine is in the fully assisted state mode, the force-position hybrid controller performs simultaneous force and position control on the motor. The force command sent is the torque compensation value, which can eliminate the slack generated by the flexible structure of the exoskeleton. The position command sent is the initial angle, i.e., 0°. At this time, the stiffness coefficient is the final assist parameter K determined in step 2.4. pFinal .

[0125] Step 4.5: In steps 4.2 and 4.4, when using the position control loop, overshoot oscillation may occur when the motor rotates a specific angle. To solve this problem, a certain amount of damping can be added, i.e., the damping coefficient K is set to... d This is a relatively small value, ranging from 0.5 to 2, while the target velocity remains at 0, as shown in the following formula. In a specific embodiment, ξ = 1 can be selected.

[0126] K d =ξ,V set =0;

[0127] This invention provides an adaptive assist control method for a lumbar exoskeleton robot. An upper-level parameter generator generates parameters for the fully assisted state mode in a middle-level finite state machine based on the lumbar flexion angle and plantar pressure values ​​at the lowest point of bending for different wearers. The middle-level finite state machine sets four state modes based on the lumbar flexion-extension angle, angular velocity, and plantar pressure of the human body. A force-position hybrid controller enables force-position control of the exoskeleton motors. This control method has an adaptive effect and can adapt to the working characteristics of the lumbar assist exoskeleton.

Claims

1. An adaptive assist control method for a lumbar exoskeleton robot, characterized in that, Specifically, including, Step 1: Installation of data acquisition equipment and execution unit, and information collection; Step 2: Upper-level control system—Constructing the assist parameters for the fully assisted state mode; Step 2.1: Define the fully assisted state: The process from when the wearer bends over to lift the object to be moved into a ready state to when they return to an upright position is defined as the fully assisted process of the finite state machine of the middle-level control system. The assistance parameters of this process depend on the parameters determined by the upper-level parameter generator. Step 2.2: Construct adaptive human motion assistance parameters; The formula for determining the adaptive human motion assistance parameters is as follows: In the formula, K pLowest To adapt to human motion assistance parameters, K ref To help establish a baseline value for the parameter, θ lowest The position of the lowest point of bending over during the wearer's carrying motion, recorded by the microcontroller; Step 2.3: Construct adaptive task weight assist parameters; The formula for the adaptive work weight assist parameter is as follows: In the formula, K pMax For adaptive operation weight assist parameters, F max F represents the maximum plantar pressure value recorded by the microcontroller after the wearer grasps an object and leaves the ground. ref The maximum plantar pressure value recorded by the microcontroller before the wearer grasps an object; Step 2.4: Construct the final assist parameters; The final assist parameters determined by the upper-level parameter generator are determined by the product of the adaptive human motion assist parameters and the adaptive work weight assist parameters, as shown in the following formula: K pFinal =K pLowest K pMax , T1=K pFinal (θ set -θ now ), In the formula, K pFinal This is the final assist parameter, which plays a role in the fully assisted mode. T1 is the assist torque value when the target position is a normal upright position, and θ is the assist torque value. set The target position is the normal upright position, θ now Current position; Step 3: Mid-level control system—Constructing a finite state machine; Step 4: Underlying Control System - Constructing a force-position hybrid controller driven by the motor current loop.

2. The adaptive assist control method for a lumbar exoskeleton robot according to claim 1, characterized in that, Step 1 specifically includes: Step 1.1: Installation design of inertial measurement unit, pressure sensor, actuator, and microcontroller; Step 1.2: Collect individual motion data and work weight data; The inertial measurement unit in the lower back is used to detect the Euler angles and angular velocities of the human body's lumbar flexion and extension movements. The inertial measurement unit in the thigh is used to detect the Euler angles and angular velocities of the human body's legs. The plantar pressure sensor detects the total weight. The motor output shaft needs to be fixedly connected to the end of the human-machine compatible exoskeleton actuator. The other end of the actuator is fixedly connected to the outer garment above the human body's back.

3. The adaptive assist control method for a lumbar exoskeleton robot according to claim 1, characterized in that, In step 3, the middle-level control system uses a finite state machine to divide the human body into four state modes based on the angle and angular velocity of the lumbar and back flexion and extension and the plantar pressure: normal standing, zero-force tracking, assist preparation and full assistance. Each state mode sends different parameters to the lower-level control system, and the transition between state modes is based on the lumbar and back angle, angular velocity and plantar pressure.

4. The adaptive assist control method for a lumbar exoskeleton robot according to claim 3, characterized in that, The state mode, the threshold for transitioning to the next state, and the instructions sent to the lower layer are as follows: When standing normally, θ t >0°, the motor remains stationary; During zero-force tracking, F t -F t-1 >f, the motor tracks the rotation angle of the lower back; Help with preparation, ω t >ω, the motor outputs a smaller torque; Full power, θ t ≤0°, the adjusted torque output of the motor. Of the four state modes above, θ t For the wearer's real-time flexion angle of the lower back, F t and F t-1 These represent the wearer's current plantar pressure value and the previous plantar pressure value, respectively, ω t The real-time angular velocity of the wearer's lower back extension is given by f and ω, which are the force and angular velocity thresholds that need to be set manually, respectively.

5. The adaptive assist control method for a lumbar exoskeleton robot according to claim 4, characterized in that, Based on the state patterns, the threshold for transitioning to the next state, and the instructions sent to the lower layer, a finite state machine is constructed. The specific steps are as follows: Step 3.1: During initialization, the wearer needs to stand naturally and then power on the control system. The inertial measurement unit, plantar pressure sensor and brushless DC motor are initialized and put into working state. Step 3.2: The finite state machine enters the normal standing state mode, in which the underlying control system keeps the motor stationary; Step 3.3: Determine whether to enter the zero-force tracking mode based on whether the flexion angle of the human lower back is greater than 0°. If yes, proceed to step 3.

4. Otherwise, maintain the normal standing mode and continue to judge. Step 3.4: The finite state machine enters the zero-force tracking state mode. In this state mode, the underlying control system causes the motor actuator to perform lumbar and back flexion follow-up, that is, the motor actuator tracks the rotation angle of the human body's lumbar and back. Step 3.5: Determine whether to enter the assisted preparation state mode based on whether the change value of plantar pressure is greater than a fixed threshold. If yes, proceed to step 3.6; otherwise, maintain the zero-force tracking state mode and continue to judge. Step 3.6: The finite state machine enters the assist preparation state mode. In this state mode, the underlying control system causes the motor actuator to output an initial torque value, giving the wearer sufficient adaptability and eliminating the slack caused by the flexible structure. Step 3.7: Determine whether to enter the fully assisted state mode based on whether the angular velocity of the human lower back extension is greater than a fixed threshold. If yes, proceed to step 3.8; otherwise, maintain the assisted preparation state mode and continue to judge. Step 3.8: The finite state machine enters the fully assisted state mode. In this state mode, the underlying control system causes the motor actuator to output the torque command that has been adjusted according to the upper controller, so as to provide the wearer with sufficient assistance. Step 3.9: Determine whether to return to the normal standing position mode based on the extension angle of the human waist and back. If yes, proceed to step 3.

2. Otherwise, maintain the fully assisted state mode and continue to judge.

6. The adaptive assist control method for a lumbar exoskeleton robot according to claim 1, characterized in that, In step 4, the underlying control system is a force-position hybrid controller based on motor current loop drive. Its direct target is a brushless DC motor driven by FOC current loop, achieving pure force control, pure position control, and simultaneous force-position control. Essentially, it aims to ensure the motor accurately outputs the required torque. The formula for calculating the torque directly output by the FOC current loop driven brushless DC motor is as follows: T output =K p (P set -P now )+K d (V set -V now )+T set In the formula, T output P represents the torque directly output by a brushless DC motor driven by the FOC current loop. set For the set target location, P now V represents the current position detected by the encoder. set For the set target speed, V now K is the current speed detected by the encoder. p K is the stiffness coefficient. d T is the damping coefficient. set The set feedforward torque.

7. The adaptive assist control method for a lumbar exoskeleton robot according to claim 6, characterized in that, In step 4, the underlying force-position hybrid controller receives different instruction parameters: When standing normally, the motor remains stationary, K p =0,K d =0,T set =0; During zero-force tracking, the motor tracks the rotation angle of the lower back, K p =k,p set =θ t ,K d =0,T set =0; During the preparation for power assistance, the motor outputs a smaller torque, K p =0,K d =0,T set =τ; When fully assisted, the motor outputs the adjusted torque, K. p =K pFinal ,P set =0,K d =0,T set =τ; where K pFinal For the final assist parameters; Where τ is the set non-zero feedforward torque value, and k is the set non-zero stiffness value.

8. The adaptive assist control method for a lumbar exoskeleton robot according to claim 6, characterized in that, In step 4, the specific steps for constructing the force-position hybrid controller driven by the motor current loop are as follows: Step 4.1: When the finite state machine is in the normal standing state mode, the force-position hybrid controller does not control the motor, so that the motor is in a passive stationary state; Step 4.2: When the finite state machine is in zero-force tracking mode, the force-position hybrid controller performs pure position control on the motor and sends the position command as the human lumbar flexion angle detected by the inertial measurement unit. Step 4.3: When the finite state machine is in the assist preparation state mode, the force-position hybrid controller performs pure force control on the motor and sends a force command with a small torque value. Step 4.4: When the finite state machine is in the fully assisted state mode, the force-position hybrid controller performs simultaneous force and position control on the motor. The force command sent is the torque compensation value, which is used to eliminate the slack generated by the flexible structure of the exoskeleton. The position command sent is the initial angle, i.e., 0°. At this time, the stiffness coefficient is the final assist parameter K determined in step 2.

4. pFinal .

9. The adaptive assist control method for a lumbar exoskeleton robot according to claim 8, characterized in that, In steps 4.2 and 4.4, if no overshoot oscillation occurs when the motor rotates a specific angle using the position control loop, then it is not necessary to set the damping coefficient and target speed. If overshoot oscillation occurs, then it is necessary to add a damping coefficient and set the target speed to 0. K d =ξ,V set =0, In the formula, ξ is the set non-zero damping value.

Images