A method for evaluating the comfort of carrying a nursing robot

Abstract

This invention provides a method for evaluating the comfort of a nursing robot's back-holding mechanism. First, the nursing robot performs a back-holding experiment, using the trajectory of the chest-support posture change as the baseline trajectory. Next, the chest-support posture parameters are used as a single variable in the back-holding experiment to determine the chest-support angle as the key parameter affecting back-holding comfort. Then, multiple trajectory points are randomly selected from the baseline trajectory, and the chest-support angle is changed to collect electromyographic signals of the latissimus dorsi muscle surface and human-robot contact force, obtaining sample data. The constructed comfort evaluation model is trained using this sample data. Next, a subjective comfort evaluation experiment is designed to obtain the latissimus dorsi muscle activation threshold range corresponding to each trajectory point on the baseline trajectory. Finally, the chest-support movement is controlled according to the baseline trajectory, and the human-robot contact force is collected and input into the model. The chest-support angle is adjusted to ensure that the predicted latissimus dorsi muscle activation value is within the threshold range, guaranteeing user comfort. This method provides a reference for the quantitative analysis of the comfort of the person being cared for during a nursing robot's back-holding process.

Description

Technical Field

[0001] This invention belongs to the field of robotics engineering technology, specifically relating to a method for evaluating the comfort of a nursing robot being carried. Background Technology

[0002] With the increasing aging of the population, the number of disabled or semi-disabled elderly people is growing daily. Given the insufficient care resources, care robots have emerged. The carrying and transferring actions of care robots refer to activities such as transferring immobile individuals from bed to wheelchair or vice versa. Because the robot and the individual are in rigid contact, higher requirements are placed on the comfort of use. User comfort has become a key factor influencing the development of care robots and is the foundation for optimizing robot performance.

[0003] Currently, the evaluation of the comfort of nursing robots is mostly based on objectively measured body pressure indicators, combined with subjective comfort questionnaires to obtain evaluation results, and then establish the relationship between the user's subjective feelings and body pressure indicators. However, comfort is a comprehensive feeling of pain, fatigue, and muscle strength, and body pressure indicators alone cannot fully characterize the comfort level of the person being cared for. The discomfort of being carried is also related to biomechanical factors such as joint angles, muscle contraction, and pressure distribution, all of which affect the comfort of use. In order to improve the shortcomings of the existing comfort evaluation system for nursing robots, this application proposes a method for evaluating the comfort of carrying a nursing robot, which can be used to guide the carrying and transfer process of the nursing robot. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the technical problem this invention aims to solve is to provide a method for evaluating the comfort of being carried by a nursing robot.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for evaluating the comfort of a nursing robot carried on the back, the nursing robot comprising a mobile chassis, a robotic arm, and a chest rest; one end of the robotic arm is rotatably mounted on the mobile chassis, and the chest rest is rotatably mounted on the end of the robotic arm, the posture of the chest rest being adjusted by the robotic arm; characterized in that the method includes the following steps:

[0007] Step 1: The nursing robot is instructed to carry the subject from the starting position to the target position in a back-carrying experiment; the positional change trajectory of the chest rest is obtained during the back-carrying experiment, and the positional change trajectory of the chest rest is used as the basic trajectory; the positional change of the chest rest includes the position coordinates and angle of the chest rest.

[0008] Step 2: Conduct a back-hugging experiment with the parameters of the chest-support posture as a single variable to determine that the key parameter affecting the comfort of back-hugging is the chest-support angle.

[0009] Step 3: The subjects were asked to perform maximum voluntary contraction of the latissimus dorsi muscle and collect the maximum activation of the latissimus dorsi muscle. Multiple trajectory points were randomly selected from the basic trajectory, and the chest-rest angle at each trajectory point was changed. Electromyographic signals of the latissimus dorsi muscle surface and human-machine contact force were collected at each trajectory point and at each chest-rest angle to obtain sample data.

[0010] A comfort evaluation model is established based on a neural network. The input and output of the comfort evaluation model are human-machine contact force and latissimus dorsi activation, respectively. The human-machine contact force includes axillary contact force and chest contact force. The formula for calculating latissimus dorsi activation (Act) is as follows:

[0011]

[0012] Wherein, RMS represents the electromyographic signal on the surface of the latissimus dorsi muscle. max Indicates the maximum activation level of the latissimus dorsi muscle;

[0013] The comfort evaluation model is trained using sample data to obtain the trained comfort evaluation model;

[0014] Step 4: Design a subjective comfort evaluation experiment. Have the subject's chest rest against the chest rest, remain still, and stand with both feet on the moving platform, legs relaxed. Randomly select at least three trajectory points from the basic trajectory, changing the chest rest angle at each point. Use a questionnaire to subjectively evaluate the comfort level at each chest rest angle, with levels including very uncomfortable, uncomfortable, somewhat comfortable, comfortable, and very comfortable. Obtain the chest rest angle corresponding to each subject's "very comfortable" level. Collect the surface electromyographic signal of the latissimus dorsi muscle at this chest rest angle and calculate the latissimus dorsi muscle activation level corresponding to the "very comfortable" level. Summarize the latissimus dorsi muscle activation levels corresponding to all subjects' "very comfortable" levels, and use the 95th percentile of the "very comfortable" level latissimus dorsi muscle activation as the comfort state latissimus dorsi muscle activation threshold range. Based on the maximum and minimum values ​​of the comfort state latissimus dorsi muscle activation threshold, use interpolation to obtain the maximum and minimum value change curves of the comfort state latissimus dorsi muscle activation threshold, thus obtaining the comfort state latissimus dorsi muscle activation threshold range corresponding to any trajectory point on the basic trajectory.

[0015] Step 5: During the process of the nursing robot carrying the patient, the chest rest movement is controlled according to the basic trajectory, and the human-machine contact force is collected in real time. The collected human-machine contact force is input into the trained comfort evaluation model to obtain the latissimus dorsi activation prediction value. If the latissimus dorsi activation prediction value is within the comfort state latissimus dorsi activation threshold range, it indicates that the patient is in a comfortable state and no adjustment of the chest rest angle is required. If it is not within the comfort state latissimus dorsi activation threshold range, it indicates that the patient is in an uncomfortable state. In this case, the chest rest angle is finely adjusted so that the latissimus dorsi activation prediction value is within the comfort state latissimus dorsi activation threshold range to ensure the patient's comfort.

[0016] Furthermore, during the back-hugging experiment in step one, the subject was required to have their chest pressed against the chest rest after being picked up by the nursing robot, with their armpits placed on the armpit support structures on both sides of the chest rest, and their feet flat on the moving chassis, with their body in a natural and relaxed state and their legs not exerting any force. First, the starting and target positions of the chest rest were determined, and then the chest rest position was adjusted according to the subject's subjective comfort during the back-hugging process, and the basic trajectory was obtained through interpolation.

[0017] Furthermore, in step two, multiple trajectory points are randomly selected from the basic trajectory. The parameters of the chest-leaning posture are used as single influencing parameters, and the parameter values ​​are changed to conduct a back-hugging experiment. The human-machine contact force and latissimus dorsi surface electromyography signals are collected during the back-hugging experiment. The parameters corresponding to the most significant changes in human-machine contact force and latissimus dorsi surface electromyography signals are regarded as key parameters affecting back-hugging comfort.

[0018] Furthermore, based on the extreme learning machine, a comfort evaluation model is established, using latissimus dorsi muscle activation as a comfort evaluation index. The comfort evaluation model is expressed as follows:

[0019]

[0020] Where, β r G(a) represents the relationship between the hidden layer and the output layer. r ,b r (x) represents the output of the hidden layer, a r b r Let x and f represent the input weights and threshold values ​​of the hidden layer neurons, respectively, r = 1, 2, ..., L, where L represents the number of hidden layer neurons. L (x) represent the model input and output, respectively;

[0021] The comfort evaluation model has two input nodes, so the model input is represented as follows:

[0022] x = [Y] F ,X F ] T

[0023] Among them, Y F X F These represent the contact force under the armpit and the contact force at the chest, respectively.

[0024] The output of the comfort evaluation model is represented as follows:

[0025] f L (x)=[Act] T

[0026] Where T represents the matrix transpose.

[0027] Compared with the prior art, the beneficial effects of the present invention are:

[0028] This invention starts with two objective physiological information sources: human-machine contact force and bioelectrical signals. It utilizes a neural network to establish a mapping relationship between human-machine contact force and surface electromyography (EMG) signals, avoiding the complex and cumbersome process of real-time acquisition of EMG signals during the carrying process. Simultaneously, it combines the subjective evaluation of the patient to obtain the muscle activation threshold range, providing a reference for the quantitative analysis of patient comfort during the carrying process by the nursing robot. The evaluation method of this invention is simple, feasible, and highly practical, providing a good basis for evaluating the comfort of nursing robots and laying the foundation for performance optimization. This invention's method is not for comfort evaluation of a specific typical working condition, but can be extended to the comfort evaluation of human-machine interactions between nursing robots and patients. Attached Figure Description

[0029] Figure 1 This is a simplified diagram of the nursing robot of the present invention;

[0030] Figure 2 This is a schematic diagram of the basic trajectory of the present invention. Detailed Implementation

[0031] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments, but this is not intended to limit the scope of protection of this application.

[0032] Figure 1The schematic diagram of the nursing robot used in this invention includes a mobile chassis, an upper robotic arm L2, a lower robotic arm L1, a chest rest, and electric cylinders. The lower end of the lower robotic arm L1 is rotatably mounted on the mobile chassis. The lower end of the upper robotic arm L2 is rotatably connected to the upper end of the lower robotic arm L1. The chest rest is rotatably mounted on the end of the upper robotic arm L2. The lower robotic arm L1 is controlled by electric cylinder one, the upper robotic arm L2 by electric cylinder two, and the chest rest by electric cylinder three. The two robotic arms are used to adjust the horizontal and vertical coordinates of the chest rest in space, and electric cylinder three is used to adjust the angle between the chest rest and the horizontal direction. In use, the patient's chest rests naturally against the chest rest, and both feet are placed on the mobile chassis. The patient is in a relaxed state. The nursing robot picks up the patient from the starting position and transports them to the target position by carrying them on its back. After the patient is picked up, the position of the chest rest is adjusted in real time to ensure the patient's comfort.

[0033] The method for evaluating the comfort of carrying a nursing robot according to the present invention includes the following steps:

[0034] Step 1: Have the nursing robot perform a back-carrying experiment, picking up the subject from the starting position and transporting them to the target position; obtain the posture change trajectory of the chest support during the back-carrying experiment, and use the posture change trajectory of the chest support as the basic trajectory; the starting position and the target position can be a nursing bed, wheelchair, etc.

[0035] After being lifted by the robot, the subject's chest was pressed against the backrest, with their armpits resting on the underarm support structures on both sides of the backrest. Their hands were naturally gripping the auxiliary handles at the front of the backrest, and their feet were naturally flat on the moving chassis. The subject was required to be in a relaxed state, with no force exerted by their legs, and foot pressure sensors were used to detect whether any force was exerted by their legs. First, the initial and target poses of the backrest during the back-holding experiment were determined, corresponding to trajectory points P1 and P5 on the basic trajectory. The pose of the backrest was denoted as (X, Y, C), where X and Y represent the horizontal and vertical coordinates of the backrest, respectively, and C represents the angle of the backrest, i.e., the angle between the backrest and the horizontal direction. During the experiment, the pose was adjusted according to the subject's subjective feelings. The chest-supported posture ensures the comfort of the subjects. In this embodiment, a coordinate system is established with the center point of the left front wheel of the mobile chassis as the origin. Based on a large amount of experimental data, five trajectory points, P1 (723.44mm, 709.98mm, 47.82°), P2 (708.38mm, 771.93mm, 44.22°), P3 (697.29mm, 796.10mm, 47.24°), P4 (681.86mm, 796.54mm, 46.77°) and P5 (670.65mm, 782.93mm, 45.65°), are selected, and the basic trajectory is obtained by interpolation.

[0036] Step 2: Change the chest-support posture to determine the key parameters that affect the comfort of carrying the person.

[0037] Force sensors were installed under the subject's armpit and on the front of the chest. The armpit force sensor was installed at the point where the subject's armpit contacts the nursing robot to collect the armpit contact force. The front of the chest, where the contact area between the subject and the nursing robot is larger, used an array of force sensors and was installed at the point where the subject's front of the chest contacts the chest to collect the chest contact force. The armpit contact force and the chest contact force are collectively referred to as the human-machine contact force. Electromyography (EMG) sensors were installed on the subject's latissimus dorsi muscles to collect the surface EMG signals of the latissimus dorsi muscles.

[0038] Three trajectory points were randomly selected from the base trajectory. Three parameters of the chest-back posture were fine-tuned near these selected points. The fine-tuning range for the horizontal axis of the chest-back posture was (X-20, X+10) (mm), the fine-tuning range for the vertical axis was (Y-20, Y+10) (mm), and the fine-tuning range for the chest-back angle was (C-10°, C+10°). These three parameters of the chest-back posture were used as single influencing parameters in a back-hug experiment. Simultaneously, the human-machine contact force and surface electromyography (EMG) signals of the latissimus dorsi muscle were collected during the experiment. The temporal characteristics of surface EMG signals can reflect changes in the signal over time and have the advantages of simple and fast calculation. Commonly used indicators include integrated electromyography (iEMG) and root mean square (RMS). In this embodiment, the RMS value was selected as the surface EMG signal indicator.

[0039] The integrated electromyographic value represents the degree of electrical activity in muscle fibers. It reflects how the amplitude of the surface electromyographic signal changes with movement. Its calculation formula is as follows:

[0040] The root mean square (RMS) value represents the change in surface electromyography (EMG) signal per unit time, reflecting the degree of muscle activation. Its calculation formula is as follows: T represents the length of the surface electromyography signal, and EMG(t) represents the amplitude of the surface electromyography signal at time t;

[0041] Experimental results show that the changes in human-machine contact force and electromyographic signals on the latissimus dorsi surface are most significant when the chest-back angle is changed. Therefore, the chest-back angle is considered to be a key parameter affecting the comfort of being carried.

[0042] Step 3: First, select several participants with an average height of 170.4 cm and an average weight of 64.4 kg. Participants are instructed to perform maximum voluntary contraction (MVC) calibration of the latissimus dorsi. This involves participants, with assistance, completing a prescribed MVC calibration movement. Participants are required to stand with their feet shoulder-width apart, lower legs perpendicular to the floor, thighs bent at a 45-degree angle, and upper body leaning forward at a 30-degree angle. The participants hold the assistant's hands, bending their elbows and using their elbows to push backward to maintain the current body posture, stimulating the maximum contraction force of the latissimus dorsi. Simultaneously, the RMS (Recovery Mean Squared Value) of the latissimus dorsi muscle is recorded. max Multiple trajectory points were randomly selected from the base trajectory. The chest-rest angle at each trajectory point was adjusted to 35°, 45°, 55°, and 65°, respectively. Electromyographic (RMS) signals of the latissimus dorsi surface and human-machine contact force were collected from the subjects at each selected trajectory point and chest-rest angle to obtain training sample data. The latissimus dorsi activation level (Act) was calculated using the following formula:

[0043]

[0044] Then, based on the Extreme Learning Machine (ELM), a comfort evaluation model is established, using latissimus dorsi activation as the comfort evaluation index. Therefore, the output of the comfort evaluation model is latissimus dorsi activation, and the comfort evaluation model can be expressed as:

[0045]

[0046] Where, β r G(a) represents the relationship between the hidden layer and the output layer. r ,b r (x) represents the output of the hidden layer, a r b r Let x and f represent the input weights and threshold values ​​of the hidden layer neurons, respectively, r = 1, 2, ..., L, where L represents the number of hidden layer neurons. L (x) represent the model input and output, respectively;

[0047] The comfort evaluation model has two input nodes, so the model input is represented as follows:

[0048] x = [Y] F ,X F ] T

[0049] Y F X F These represent the axillary contact force and the chest contact force, respectively. Therefore, the comfort evaluation model reflects the mapping relationship between human-machine contact force and latissimus dorsi muscle activation.

[0050] The output of the comfort evaluation model is represented as follows:

[0051] f L (x)=[Act] T

[0052] To account for the differences among subjects, the human-machine contact force was made dimensionless by dividing the subject’s axillary contact force and chest contact force by their respective body weights.

[0053] Finally, the comfort evaluation model is trained using the aforementioned sample data to obtain the trained comfort evaluation model.

[0054] Step 4: Design a subjective comfort evaluation experiment. Have the subject's chest rest against the chest rest, remaining still with both feet on the moving platform, legs relaxed naturally. The subject does not need to actively exert any force. Randomly select at least three trajectory points from the base trajectory, changing the chest rest angle at each point within the range of (C-θ~5°, C+θ~5°). Use a questionnaire to subjectively evaluate the comfort level at each chest rest angle, with levels including very uncomfortable, uncomfortable, somewhat comfortable, comfortable, and very comfortable. Obtain the chest rest angle corresponding to each subject's "very comfortable" level, and collect data on the latissimus dorsi muscles at that chest rest angle. Facial electromyography (EMG) signals, with a signal length of 5 seconds, were used to calculate the latissimus dorsi activation level at the "very comfortable" level. The latissimus dorsi activation levels of all subjects at the "very comfortable" level were summarized, and the 95th percentile of the "very comfortable" level latissimus dorsi activation level was taken as the threshold range of latissimus dorsi activation in the comfortable state. Based on the threshold range of latissimus dorsi activation in the comfortable state corresponding to the selected trajectory points, the maximum and minimum values ​​of the latissimus dorsi activation threshold in the comfortable state were obtained. The maximum and minimum value change curves of the latissimus dorsi activation threshold in the comfortable state were obtained by interpolation. Therefore, a threshold range of latissimus dorsi activation in the comfortable state can be obtained for each trajectory point on the basic trajectory.

[0055] Step 5: During the process of the nursing robot carrying the patient, the chest rest movement is controlled according to the basic trajectory obtained in Step 1. The human-machine contact force is collected in real time and input into the trained comfort evaluation model to obtain the latissimus dorsi activation prediction value. If the latissimus dorsi activation prediction value is within the comfort state latissimus dorsi activation threshold range, it indicates that the patient is in a comfortable state and no adjustment of the chest rest angle is required. If it is not within the comfort state latissimus dorsi activation threshold range, it indicates that the patient is in an uncomfortable state. Then, the chest rest angle is finely adjusted by the electric cylinder to bring the latissimus dorsi activation prediction value within the comfort state latissimus dorsi activation threshold range, thereby ensuring the comfort of the nursing robot during the process of carrying the patient.

[0056] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. A method for evaluating the comfort of a nursing robot carried on the back, the nursing robot comprising a mobile chassis, a robotic arm, and a chest rest; one end of the robotic arm is rotatably mounted on the mobile chassis, and the chest rest is rotatably mounted on the end of the robotic arm, the posture of the chest rest being adjusted by the robotic arm; characterized in that, The method includes the following steps: Step 1: The nursing robot is instructed to carry the subject from the starting position to the target position in a back-carrying experiment; the positional change trajectory of the chest rest is obtained during the back-carrying experiment, and the positional change trajectory of the chest rest is used as the basic trajectory; the positional change of the chest rest includes the position coordinates and angle of the chest rest. Step 2: Conduct a back-hugging experiment with the parameters of the chest-support posture as a single variable to determine that the key parameter affecting the comfort of back-hugging is the chest-support angle. Step 3: The subjects were asked to perform maximum voluntary contraction of the latissimus dorsi muscle and collect the maximum activation of the latissimus dorsi muscle. Multiple trajectory points were randomly selected from the basic trajectory, and the chest-rest angle at each trajectory point was changed. Electromyographic signals of the latissimus dorsi muscle surface and human-machine contact force were collected at each trajectory point and at each chest-rest angle to obtain sample data. A comfort evaluation model is established based on a neural network. The input and output of the comfort evaluation model are human-machine contact force and latissimus dorsi activation, respectively. The human-machine contact force includes axillary contact force and chest contact force. The formula for calculating latissimus dorsi activation (Act) is as follows: wherein RMS denotes the surface electromyography signal of the latissimus dorsi muscle, RMS max denotes the maximum activation of the latissimus dorsi muscle; The comfort evaluation model is trained using sample data to obtain the trained comfort evaluation model; Step 4: Design a subjective comfort evaluation experiment. Have the subject's chest rest against the chest rest, remain still, and stand with both feet on the moving platform, legs relaxed. Randomly select at least three trajectory points from the base trajectory, changing the chest rest angle at each point. Use a questionnaire to subjectively evaluate the comfort level at each chest rest angle, with levels including very uncomfortable, uncomfortable, somewhat comfortable, comfortable, and very comfortable. Obtain the chest rest angle corresponding to each subject's "very comfortable" level. Collect the latissimus dorsi surface electromyography (EMG) signal at this chest rest angle and calculate the latissimus dorsi activation level corresponding to the "very comfortable" level. Summarize the latissimus dorsi activation levels corresponding to all subjects' "very comfortable" levels, and use the 95th percentile of the "very comfortable" level latissimus dorsi activation as the comfort state latissimus dorsi activation threshold range. Based on the maximum and minimum values ​​of the comfort state latissimus dorsi activation threshold, use interpolation to obtain the maximum and minimum value change curves of the comfort state latissimus dorsi activation threshold, thus obtaining the comfort state latissimus dorsi activation threshold range corresponding to each trajectory point on the base trajectory. Step 5: During the process of the nursing robot carrying the patient, the chest rest movement is controlled according to the basic trajectory, and the human-machine contact force is collected in real time. The collected human-machine contact force is input into the trained comfort evaluation model to obtain the latissimus dorsi activation prediction value. If the latissimus dorsi activation prediction value is within the comfort state latissimus dorsi activation threshold range, it indicates that the patient is in a comfortable state and no adjustment of the chest rest angle is required. If it is not within the comfort state latissimus dorsi activation threshold range, it indicates that the patient is in an uncomfortable state. In this case, the chest rest angle is finely adjusted so that the latissimus dorsi activation prediction value is within the comfort state latissimus dorsi activation threshold range to ensure the patient's comfort.

2. The method for evaluating the comfort of carrying a nursing robot according to claim 1, characterized in that, In the back-hugging experiment in step one, the subject was required to be picked up by the nursing robot with their chest pressed against the chest rest, their armpits placed on the armpit support structures on both sides of the chest rest, and their feet flat on the moving chassis. The body was in a natural and relaxed state, and the legs did not exert any force. First, the starting and target positions of the chest rest were determined. Then, during the back-hugging process, the chest rest position was adjusted according to the subject's subjective comfort. The basic trajectory was obtained by interpolation.

3. The method for evaluating the comfort of carrying a nursing robot according to claim 1, characterized in that, In step two, multiple trajectory points are randomly selected from the basic trajectory. The parameters of the chest-leaning posture are used as the single influencing parameter, and the parameter values ​​are changed to conduct a back-hugging experiment. The human-machine contact force and latissimus dorsi surface electromyography (EMG) signals are collected during the back-hugging experiment. The parameters corresponding to the most significant changes in human-machine contact force and latissimus dorsi surface EMG signals are regarded as the key parameters affecting back-hugging comfort.

4. The method for evaluating the comfort of carrying a nursing robot according to any one of claims 1 to 3, characterized in that, Based on the Extreme Learning Machine, a comfort evaluation model is established, using latissimus dorsi muscle activation as the comfort evaluation index. The comfort evaluation model is expressed as follows: where β r represents the relationship between the hidden layer and the output layer, G(a r ,b r ,x) represents the output of the hidden layer, a r ,b r represent the input weight of the hidden layer neuron and the threshold value of the hidden layer neuron, respectively, r = 1, 2, …, L, L represents the number of hidden layer neurons, x, f L (x) represent the model input and output, respectively; The comfort evaluation model has two input nodes, so the model input is represented as follows: x = [Y F ,X F ] T wherein Y F , X F respectively represent the underarm contact force and the chest contact force; The output of the comfort evaluation model is represented as follows: f L (x)=[Act] T Where T represents the matrix transpose.

Images