A method for planning upstairs moment based on multi-pose information fusion

Abstract

A method for planning stair-climbing torque based on multi-pose information fusion first acquires the pose information of the lower limbs when climbing stairs through a pose sensor. Then, based on the pose information of the lower limbs, the motion information of the human lower limbs when climbing stairs is obtained. Next, kinematic calculations are performed on the entire lower limb pose data to obtain the motion information of the human lower limbs, and then the human motion trajectory is obtained. The vertical velocity of the ankle joint is used as the discrimination criterion. When the velocity is greater than a predetermined threshold, an assist phase torque is output and the device outputs torque to follow the human body according to the interactive force data. When the velocity is less than the predetermined threshold, a follow-up phase torque is used and no assist phase torque is output. Finally, the motion controller MCU outputs torque and the knee joint motor executes it. This invention patent utilizes the ankle joint, which has good periodicity and strong noise resistance, to recognize the motion state and improve the stability of state identification. By filtering the human-computer interaction force and the output torque, the accuracy of torque is improved while optimizing the user comfort of the exoskeleton wearer.

Description

Technical Field

[0001] This invention relates to the field of exoskeleton motion control, specifically to a method for planning stair-climbing torque based on multi-posture information fusion. Background Technology

[0002] Lower limb exoskeleton technology is a technology that connects to the human body through mechanical devices to enhance human mobility and function. When using an exoskeleton to climb stairs, the mechanical structure and electrodynamic system distribute some of the weight and load, reducing pressure on the knee joint and thus lowering the risk of joint pain and injury. Torque planning technology plays a crucial role in this process. Exoskeletons often need to adjust joint torques based on actual movement conditions, such as ground slope, stair inclination, and stride height, to ensure user stability and safety. In posture detection for walking and climbing stairs, the movements of the thigh and lower leg exhibit both randomness and periodicity. Traditional motion state recognition algorithms based on thigh and lower leg motion sensor data cannot accurately determine the movement state when climbing stairs. For example:

[0003] CN 113771040 A proposes a control system and method for a lower limb exoskeleton robot, which recognizes human movement intentions and plans gait by collecting information from sensors. However, this method is mainly designed for walking and lacks the ability to recognize and optimize movement when going up and down stairs.

[0004] CN 115592690 A proposes a lower limb exoskeleton control device and method based on IMU and video stream. It can accurately acquire human gait information using non-invasive means. However, since this method requires video stream to improve and optimize the model, it is not applicable in unknown outdoor environments.

[0005] This invention proposes an upstairs moment planning method based on multi-pose information fusion to solve the above-mentioned technical problems. Summary of the Invention

[0006] A method for planning the upstairs moment based on multi-pose information fusion, the steps of which are as follows:

[0007] First, the posture information of the human lower limbs when climbing stairs is acquired by a posture sensor installed on the lower limbs. Then, the motion information of the human lower limbs when climbing stairs is obtained based on the posture information of the lower limbs. Next, kinematic calculations are performed on the entire lower limb posture data to obtain the motion trajectory of the human lower limbs. The vertical velocity of the ankle joint is used as the discrimination criterion. When the velocity is greater than a predetermined threshold, an assist phase torque is output. When it is less than a predetermined threshold, a follow-up phase torque is used and no assist phase torque is output. Finally, the motion controller MCU outputs the corresponding assist torque, which is executed by the assist servo motor of the knee joint, so that the knee exoskeleton provides assist support to the wearer's knee joint during the process of climbing stairs.

[0008] Preferably, the method for planning the upstairs torque based on multi-posture information fusion embeds IMU sensors at the thighs, calves, and ankle joints on both sides of the exoskeleton wearer to perceive lower limb posture and orientation information in real time.

[0009] Preferably, in the above-mentioned method for planning the upstairs torque based on multi-posture information fusion, the assist structure at the knee joint is embedded with a strain-type interactive force sensor. The real-time interactive force error is obtained by subtracting the sensed human-computer interaction force from the target human-computer interaction force and then summed with the force input by the PD controller and applied to the exoskeleton knee joint motor.

[0010] Preferably, in the method for planning the upstairs torque based on multi-posture information fusion, the motion controller MCU filters the data before outputting the torque command to form the final torque control data, which is then handed over to the knee joint servo motor for execution.

[0011] Preferably, in the method for planning the upstairs moment based on multi-pose information fusion, the posture relationship of the human foot during the upstairs process is as follows:

[0012]

[0013] in, The angle between the thigh and the vertical space. The angle between the thigh and the calf. The angle between the lower leg and the foot is an acute angle. , , The lengths of the thigh, calf, and foot are obtained through a posture sensor.

[0014] Preferably, the method for planning the upstairs moment based on multi-posture information fusion obtains information on joint angular velocity and angular acceleration by first and second smoothing the slope of the pose curve, and performs kinematic calculation on the entire lower limb posture data to obtain the movement trajectory of the human foot.

[0015] Preferably, in the method for planning the upward movement torque based on multi-pose information fusion, the motion controller MCU performs torque planning based on the fused lower limb movement trajectory, and the assist calculation formula is as follows:

[0016]

[0017] in, The derivative of the actual human-computer interaction force. To account for the current position error compared to the time when the object was upright, For the knee joint angular velocity, is a sign function for angular velocity.

[0018] The working principle is as follows:

[0019] During the process of the wearer climbing stairs, the lower limb movement is divided into three stages: lifting the foot, supporting, and landing (pre-support). In these three stages, the speed of the ankle joint end point is detected in real time to determine whether the vertical speed of the ankle joint (i.e., the vertical upward speed) is greater than the set threshold. That is, the speed of the foot determines whether it is in the lifting stage. If it is in the lifting stage, the motion controller MCU obtains the lower limb posture movement trajectory by analyzing and calculating the data transmitted by the IMU sensor. The motion controller MCU outputs the corresponding torque to the knee joint servo motor to execute the lower limb posture movement trajectory.

[0020] The advantages are as follows:

[0021] In human walking and stair climbing posture detection, the movements of the thigh and calf exhibit both randomness and periodicity. Traditional motion state recognition algorithms based on calf and thigh motion sensor data cannot accurately determine the motion state when climbing stairs. This algorithm, however, does not require foot data. The multi-posture information fusion stair climbing torque planning method of this invention identifies the stair climbing motion state through the ankle joint (foot) motion state. The foot has good periodicity and strong noise resistance, so the motion state can be accurately determined by acquiring the foot posture at the ankle joint. Then, after obtaining the lower limb data through posture sensors installed on the lower limbs, corresponding control data is output to make the knee joint's assist torque adapt to the stair climbing requirements. By fusing the sensing data from the strain-type interactive force sensor with the feedback data, the mechanical resistance of the motor in the swinging phase is minimized, the inertia of the equipment is overcome, the user comfort of the exoskeleton is improved, and the equipment's responsiveness is enhanced. The motion controller MCU performs filtering before sending torque data to smooth the force application process and improve the user comfort of the exoskeleton. Attached Figure Description

[0022] The specific embodiments are further described below with reference to the accompanying drawings, wherein:

[0023] Figure 1 This invention relates to a flowchart of an upstairs moment planning method based on multi-pose information fusion;

[0024] Figure 2 This is the kinematic model of a single human leg in Specific Implementation Case 1;

[0025] Figure 3 It is a graph showing the change of the ankle joint distal point with gait;

[0026] Figure 4 This is a graph showing the phase recognition results for the right-side staircase assist.

[0027] Figure 5 This is a curve showing the phase recognition results for the left-side staircase assist.

[0028] Figure 6 This is the control block diagram of the follow-up control of the equipment in the non-assisted state in specific implementation case 2;

[0029] Figure 7 This is the control block diagram for the assist filter processing in specific implementation case 3;

[0030] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Implementation

[0031] Specific implementation case 1:

[0032] A method for planning the upstairs moment based on multi-pose information fusion, the steps of which are as follows:

[0033] First, the posture information of the human lower limbs when climbing stairs is obtained by a posture sensor installed on the lower limbs;

[0034] The motion information of the human lower limbs when climbing stairs is obtained based on the posture information of the lower limbs;

[0035] Kinematic calculations were performed on the entire lower limb posture data to obtain the motion trajectory of the human lower limbs.

[0036] The vertical velocity of the ankle is used as the criterion. When the velocity is greater than a predetermined threshold, the assist phase torque is output. When the velocity is less than the predetermined threshold, the follow-up phase torque is used and no assist phase torque is output. Finally, the corresponding assist torque is output by the motion controller MCU, which is executed by the assist servo motor of the knee joint to realize the exoskeleton assistance for climbing stairs.

[0037] IMU sensors are embedded in the thighs, calves, and ankles on both sides of the exoskeleton wearer to sense lower limb posture and orientation information in real time.

[0038] During the process of an exoskeleton wearer climbing stairs: the hip joint acts as an undriven unit, the knee joint as an active driven unit, and the ankle joint as a passive driven unit. The wearer can freely rotate the ankle joint to drive the foot movement. The motion controller MCU sends and receives sensor data through the CAN communication network, and analyzes and calculates motion commands based on the sensor data of the lower limb IMU sensor to provide the wearer with knee joint assistance during the process of climbing stairs.

[0039] During the process of climbing stairs, the kinematic model of a single human leg is as follows: Figure 2 As shown, the positional relationship of the human foot is as follows:

[0040]

[0041] in, The angle between the thigh and the vertical space. The angle between the thigh and the calf. The angle between the lower leg and the foot is an acute angle. , , The system obtains the lengths of the thigh, calf, and foot using a posture sensor. x and y represent the pose of the foot on one side relative to the origin of the hip joint. This allows the system to obtain the real-time pose information of the human lower limbs (thigh, calf, foot, etc.) in space. The system then performs first and second-order smoothing of the pose curve to obtain information on joint angular velocity and angular acceleration. Finally, it performs kinematic calculations on the entire lower limb pose data to obtain the motion trajectory of the human lower limbs.

[0042] The formulas for the first and second smoothing slopes are based on the least squares method, and the specific expressions are as follows:

[0043]

[0044] in, Corresponding to the sampling time, The curves show the angular velocities and angles of the upper and lower legs. The final calculated end-point pose coordinates are shown below. Figure 3 As shown, the velocity direction of the Y-pose on the opposite side is used as the discrimination criterion. A velocity greater than a predetermined threshold is considered motion phase 1 (assist phase), and a velocity less than the predetermined threshold is considered follow-up phase. The discrimination results for the left and right legs are as follows. Figure 4 and Figure 5 As shown, the speed threshold typically has three judgment conditions: the angle between the foot and the horizontal (e.g., greater than 10 degrees), the ankle joint angular velocity (e.g., greater than 5 degrees / second), and the duration of maintaining the first two states (e.g., 30ms for three detection cycles).

[0045] The motion controller MCU performs torque planning based on the fused human lower limb motion trajectory, and the assist calculation formula is as follows:

[0046]

[0047] in, The derivative of the actual human-computer interaction force. To account for the current position error compared to the time when the object was upright, For the knee joint angular velocity, It is a sign function of angular velocity, with a value of 1 if greater than 0 and -1 if less than 0.

[0048] Specific Implementation Case 2:

[0049] A method for planning the upstairs moment based on multi-pose information fusion, based on specific implementation case 1,

[0050] Furthermore, the assist structure at the knee joint incorporates a strain-type interactive force sensor, which calculates the real-time interactive force error by subtracting the sensed human-machine interaction force from the target human-machine interaction force. This error is then summed with the force input by the PD controller and applied to the exoskeleton knee joint motor.

[0051] Interactive force sensors are mainly used to determine the following relationship between the human body and the exoskeleton. They mainly act on the swing phase of the human gait, and their main purpose is to eliminate the mechanical resistance of the motor in the swing phase and overcome the inertia of the equipment. The output torque is later referred to as follow-up assist.

[0052] During the servo-assisted movement, the human-machine interaction force during real-time motion is low-pass filtered and then subtracted from the target human-machine interaction force to obtain the real-time interaction force error. This error is then summed with the feedforward input by the PD controller and applied to the exoskeleton knee joint motor. The control block diagram for this servo-assisted movement is attached. Figure 6 .

[0053]

[0054] The derivative of the actual human-computer interaction force. Human-computer interaction force error, The joint angular velocity, It is a sign function of angular velocity, with a value of 1 if greater than 0 and -1 if less than 0. The target interaction force is 0. The feedforward part is mainly the actual measured moment of inertia and the rotational friction compensation after the motor is energized.

[0055] Specific Implementation Case 3:

[0056] A method for planning the upstairs moment based on multi-pose information fusion, based on either Implementation Case 1 or Implementation Case 2,

[0057] Furthermore, before outputting the torque command, the motion controller MCU performs a first-order Kalman filter on the data to form the final torque control data, which is then given to the knee joint servo motor for execution. The corresponding assist control block diagram is shown below. Figure 7 .

[0058] After the torque data is filtered, the force application process is smoothed, avoiding sudden assistance that may cause discomfort to the human body, improving the comfort of using the exoskeleton and optimizing the human body's sensation.

[0059] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for planning the upward moment of stairs based on multi-pose information fusion, characterized in that: First, the posture information of the human lower limbs when climbing stairs is acquired by a posture sensor installed on the lower limbs. Then, the motion information of the human lower limbs when climbing stairs is obtained based on the posture information of the lower limbs. Next, kinematic calculations are performed on the entire lower limb posture data to obtain the motion trajectory of the human lower limbs. The vertical velocity of the ankle joint is used as the discrimination criterion. When the velocity is greater than a predetermined threshold, an assist phase torque is output. When it is less than a predetermined threshold, a follow-up phase torque is used and no assist phase torque is output. Finally, the motion controller MCU outputs the corresponding assist torque, which is executed by the assist servo motor of the knee joint, so that the knee exoskeleton provides assist support to the wearer's knee joint during the process of climbing stairs.

2. The method for planning upstairs moment based on multi-pose information fusion as described in claim 1, characterized in that: IMU sensors are embedded in the thighs, calves, and ankles on both sides of the exoskeleton wearer to sense lower limb posture and orientation information in real time.

3. The method for planning upstairs moment based on multi-pose information fusion as described in claim 1, characterized in that: The assist structure at the knee joint is embedded with a strain-type interactive force sensor. The real-time interactive force error is obtained by subtracting the sensed human-computer interaction force from the target human-computer interaction force. This error is then summed with the force input by the PD controller and applied to the exoskeleton knee joint motor.

4. The method for planning stair-climbing moment based on multi-pose information fusion as described in any one of claims 1 to 3, characterized in that: Before outputting torque commands, the motion controller MCU filters the data to form the final torque control data, which is then given to the knee joint servo motor for execution.

5. The method for planning upstairs moment based on multi-pose information fusion as described in claim 1, characterized in that: The positional relationship of the human foot during the process of climbing stairs is as follows: in, The angle between the thigh and the vertical space. The angle between the thigh and the calf. The angle between the lower leg and the foot is an acute angle. , , The lengths of the thigh, calf, and foot are obtained through a posture sensor.

6. The method for planning upstairs moment based on multi-pose information fusion as described in claim 1, characterized in that: By performing first and second smoothing slope calculations on the pose curve, information on joint angular velocity and angular acceleration is obtained. Kinematic calculations are then performed on the entire lower limb posture data to obtain the movement trajectory of the human foot.

7. The method for planning upstairs moment based on multi-pose information fusion as described in claim 1, characterized in that: The motion controller MCU performs torque planning based on the fused human lower limb motion trajectory, and the assist calculation formula is as follows: in, The derivative of the actual human-computer interaction force. To account for the current position error compared to the time when the object was upright, For the knee joint angular velocity, is a sign function for angular velocity.

Images