CN118806566A - A safe walking aid intention detection and protection method for a walking aid - Google Patents

Abstract

The present invention discloses a method for detecting and protecting safe walking intentions for a walker. The sensor monitoring system fuses sensor data, analyzes the user's status, estimates the user's intentions, detects in real time and even predicts the user's falling status, and controls the walker accordingly based on the analysis of the signal. When the walker detects danger, the walker executes corresponding protective measures. The safe walking intention detection system of the present invention increases the participation of both hands in the walking aid mode, is easy to operate, can detect and judge the user's movement intentions, and smoothly and comprehensively assist the user's movement. At the same time, it can identify environmental obstacle information and perform obstacle avoidance movement planning. In addition, it can judge the stability of the user's movement state, detect the tendency to fall, and control the walker's steering and forward response, thereby improving the guarantee of stability and safety for the user.

Description

A safe walking aid intention detection and protection method for a walking aid

Technical Field

The present invention relates to a protection technology for a rehabilitation walking-assisting robot, and in particular to a safety walking-assisting intention detection and protection method for a walking-assisting robot.

Background Art

China is one of the countries with the fastest aging population in the world. It is estimated that by 2050, the number of elderly people will exceed 300 million, accounting for one-fifth of China's total population, which means that China will enter a stage of super-aging. These aging problems will bring huge challenges to the economy, medical care, and elderly care services. At present, the health problems and living environment problems of China's elderly population have become the focus of attention of the whole society. With the aggravation of aging, a considerable number of elderly people have motor disorders characterized by gait instability, which increases the risk of falling while walking. The demand for assisted walking for the elderly has greatly increased. Whether it is at home or outdoors, walking robots have become resources and measures that urgently need to be followed up.

Existing mobility aid technologies have a single real-time monitoring function, low intelligence, and lack of active safety detection and protection functions. Real-time safety assurance and breaking through the limitations on users' independent mobility are effective measures to further improve the convenience and safety of the lives of the elderly.

Summary of the invention

Purpose of the invention: The purpose of the present invention is to provide a safe walking aid intention detection and protection method for a walker. The intention monitoring strategy designed specifically for an intelligent walker increases the involvement of both hands. The walker can follow the user's hands to complete related actions. The control method is more flexible and the processing speed is faster, ensuring safety while assisting walking.

Technical solution: The safe walking intention detection and active protection method for a walker integrates sensor data through a sensor monitoring system, analyzes the user's status, estimates the user's intention, detects and even predicts the user's fall status in real time, and controls the walker's actions accordingly based on the analysis of the signal. When the walker detects danger, the walker executes corresponding protective measures.

Specifically, the safe walking aid intention detection and protection system divides the real-time motion state into initialization and ready waiting state, normal working state and abnormal state. By fusing relevant sensor data, analyzing the user's state, estimating the user's intention, and detecting and predicting the user's fall state in real time, the walking aid is controlled accordingly based on the analysis of the signal, such as speed change, steering, emergency stop, alarm, etc. When the walking aid detects danger, such as the user falling, the state machine switches to the abnormal state, and the walking aid executes the corresponding protection measures, as shown in Figure 1.

Further, the intention detection method includes acceleration intention detection, steering intention detection, uphill movement detection, downhill movement detection, and obstacle avoidance intention detection; the braking intention detection method includes fall detection and safety protection response;

Furthermore, the safe walking aid intention sensing monitoring system includes a thin film pressure sensor, a three-dimensional force sensor, a laser radar and a distance sensor. The force sensor is placed at the handle of the walker to detect the upper limb force signal, the laser radar is placed in front of the walker to detect environmental information, and the distance sensor is placed in front of the walker and aligned with the lower limbs of the human body to detect the movement state of the lower limbs. As shown in Figure 1.

Furthermore, the force sensor includes two types of high-precision force sensor devices for collecting force signals from the user's hands. The force sensor devices include a force sensor matrix handle and a three-dimensional force sensor. The force sensor matrix handle is a composite sensor matrix composed of multiple pressure film sensors installed on the walker handle in a certain arrangement and combination; the three-dimensional force sensor is a six-axis force sensor directly installed on the walker handle; the distance sensor installation position is respectively aligned with the user's waist, knees and upper ankles to detect the movement state of the lower limbs; the laser radar is used to detect environmental obstacle information.

Furthermore, the forward intention detection method is specifically implemented as follows: the force sensor on the handrail of the walker is used to identify the user's intention by sensing the user's force and force distribution characteristics; if the user wants to accelerate forward, the overall force on the handrail is increased so that the force sensor detects the change in the user's force; if the user wants to accelerate forward, the overall force on the handrail is reduced so that the force sensor detects the change in the user's force; at the same time, the current environment is detected by the laser radar, and when it is detected that the user is going uphill, the uphill assistance is implemented, and when it is detected that the user is going downhill, the speed is actively controlled to prevent the speed from dropping too quickly, and when the obstacle distance information is detected, obstacle avoidance steering is implemented, and the motor can be driven by the admittance and impedance controllers to complete the corresponding actions. As shown in Figure 2.

Furthermore, the steering intention detection method is specifically implemented as follows: according to the force interaction information between the user and the robot, as the signal input for controlling the direction of the motor, the steering characteristics of the walker are divided into five labels, namely, large left turn, small left turn, straight, small right turn and large right turn. When the steering force difference between the two handles of the user is close to zero, the steering expectation is more likely to approach zero; when the steering force difference between the two sides of the user slowly increases, the steering expectation begins to accelerate; when the steering force difference reaches a certain range, the steering expectation has entered a critical state and will no longer increase, and the walker will turn at the built-in maximum steering angular velocity. As shown in Figure 3.

Furthermore, the specific implementation of the obstacle avoidance intention detection and safety protection method is to use the obstacle distance information detected by the laser radar as the signal input to control the direction of the motor, and divide the steering characteristics of the walker into 5 labels, namely, large left turn, small left turn, straight, small right turn and large right turn. When no obstacle information is detected, its steering expectation is more likely to approach zero; and as the distance of the obstacle approaches, the steering expectation begins to accelerate and increase; when the distance of the obstacle reaches a certain range, the steering expectation has entered a critical state and no longer increases, and the walker will turn at the built-in maximum steering angular velocity. As shown in Figure 4.

Furthermore, the specific implementation of the fall detection and safety protection response method is to use the force sensor matrix on the walker handle to detect the user's upper limb force signal, use the laser rangefinder sensor to detect the user's lower limb movement signal, perform multi-sensor data fusion through the Kalman filter algorithm, analyze the user's movement speed, and use the improved sequential probability ratio test method to detect whether the user has fallen on the fused data. If the user is detected to have a tendency to fall, the user and the robot are reversed according to the expected position of the stable state to prevent the falling trend until a stable state is reached and braked, thereby forming a buffering movement until the user and the robot return to a stable state. As shown in Figures 5 and 6.

Furthermore, the improved sequential probability ratio test algorithm (PLT-SPRT) is specifically implemented as follows: the sequential probability ratio test (SPRT) approximately satisfies: After taking the logarithm, the threshold becomes ln _A , ln _B , and the decision relationship of the SPRT method is: In fall detection, the null hypothesis is normal walking, and the alternative hypothesis is fall. When the value of the decision function is greater than the threshold ln _A , the test is stopped and a fall judgment is given; when the value of the decision function is less than the threshold ln _B , the test is stopped and a normal state judgment is given. For the fall detection of the walker, the judgment less than the threshold ln _B is optimized to reduce the normal state judgment and further eliminate the detection delay. Then the optimization decision function is redefined as:

Wherein, k is the sampling sequence number. When the decision function Δ(k) is greater than or equal to the lower limit threshold ln _B , no correction is made. When it is less than ln _B , the decision function Δ(k) is corrected to Δ′(k), and negative values are not accumulated.

In addition, preconditions and weights are added to the SPRT test, and a higher weight is given to the upper limbs. The precondition here is the sum of the left and right handle force values. The optimization decision function can be further expressed as:

Δ″(k)=Δ′(k)C+ln _A (1-C)

Among them, C is the state of judging the occurrence of the precondition, and _ri is the state of judging the occurrence of the i-th precondition, that is, the condition that the handle is completely released or the force increases sharply instantly. When the decision function Δ(k) is greater than or equal to the lower threshold ln _B , no correction is required; when it is less than ln _B , the decision function Δ(k) is corrected to Δ′(k), and negative values are not accumulated, as shown in Figure 6.

Compared with the prior art, the present invention has the following beneficial effects: a safe walking assistance intention detection and protection method for a walker, which obtains the user's movement intention and movement state according to the detected force data and leg movement data, thereby assisting the user in walking and monitoring the user's movement state, and predicting falls in real time, thereby triggering a fall protection response to provide anti-fall assistance to the user, thereby providing protection for the safety of the intelligent walker, allowing the walker to follow the user's hands to complete related actions, and having a more flexible control method and a faster processing speed.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of the overall technical solution of a method for detecting safe walking-assistance intention of a walking aid according to the present invention;

FIG2 is a flow chart of a forward and turning intention detection scheme based on force information of a method for detecting safe walking-assisting intention of a walking aid according to the present invention;

FIG3 is a flow chart of a steering intention detection scheme based on a force signal of a method for detecting safe walking-assistance intention of a walking aid according to the present invention;

FIG4 is a flow chart of an obstacle avoidance intention detection scheme of a method for detecting safe walking-assistance intention of a walking aid according to the present invention;

FIG5 is a flow chart of a fall monitoring and protection scheme of a method for detecting safe walking aid intention for a walker according to the present invention;

FIG6 is a flow chart of a fall monitoring process of a method for detecting safe walking aid intention for a walker of the present invention.

DETAILED DESCRIPTION

In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further described below.

The safe walking intention detection and active protection method for the walker in this embodiment is to fuse sensor data through a sensor monitoring system, analyze the user's status, estimate the user's intention, detect and even predict the user's fall status in real time, and control the walker accordingly based on the analysis of the signal. When the walker detects danger, the walker executes corresponding protective measures.

The sensing monitoring system includes a thin film pressure sensor, a three-dimensional force sensor, a laser radar and a ranging sensor; the force sensor is placed at the handle of the walker to detect the upper limb force signal; the laser radar is placed in front of the walker to detect environmental information; the ranging sensor is placed on the front side of the walker and aligned with the lower limbs of the human body to detect the movement state of the lower limbs; the laser radar is installed in front of the walker to detect environmental obstacle information.

The safe walking assistance intention detection and active protection method for a walker includes a forward intention detection method and a braking intention detection method. The forward intention detection method includes acceleration intention detection, steering intention detection, uphill movement detection, downhill movement detection and obstacle avoidance intention detection; the braking intention detection method includes fall monitoring and safety protection response.

The acceleration intention detection method uses a force sensor on the armrest of the walker to identify the user's intention by sensing the user's force and force distribution characteristics; if the user wants to accelerate, the overall force on the armrest is increased, so that the force sensor detects the change in the user's force.

The uphill motion detection and downhill motion detection methods use lidar to detect the current environment. When it is detected that the user is going uphill, it will assist in going uphill. When it is detected that the user is going downhill, it will actively control the speed to prevent the speed from dropping too quickly.

The obstacle avoidance detection method implements obstacle avoidance steering when the obstacle distance information is detected, and drives the motor to complete the corresponding action through the admittance and impedance controller.

The steering intention detection method uses the force interaction information between the user and the walker as the signal input to control the direction of the motor, and divides the steering characteristics of the walker into five labels, namely, a large left turn, a small left turn, straight ahead, a small right turn and a large right turn. When the difference in the steering force of the user's two handles is close to zero, the steering expectation is more likely to approach zero; when the difference in the steering force on both sides of the user slowly increases, the steering expectation begins to accelerate; when the difference in the steering force reaches a certain range, the steering expectation has entered a critical state and will no longer increase, and the walker will turn at the built-in maximum steering angular velocity.

The force information features involved are for the handlebar structure composed of a matrix of 8 thin film pressure sensors. The handlebar force information is analyzed according to a pixel bitmap of ordinary display accuracy through machine learning analysis. When a thin film pressure sensor senses the user's force, the signal generated by its own deformation passes through the amplification circuit and the ADC circuit and enters the main control chip. The signal sampled by the main control chip can be considered as the RGB color value of the pixel, which corresponds to the percentage of the force signal measured by the walker force sensor in the full range. The 16 thin film pressure sensors form an array of 16 pixels. The sum of the 16 pixel points corresponding to the thin film pressure sensors carried by the left and right force sensor handles is compared to determine the steering force difference between the left and right handlebars.

The obstacle avoidance intention detection and safety protection method uses the obstacle distance information detected by the lidar as the signal input to control the direction of the motor, and divides the steering characteristics of the walker into 5 labels, namely, large left turn, small left turn, straight, small right turn and large right turn. When no obstacle information is detected, its steering expectation is more likely to approach zero; as the distance of the obstacle approaches, the steering expectation begins to accelerate; when the distance of the obstacle reaches a certain range, the steering expectation has entered a critical state and will no longer increase, and the walker will turn at the built-in maximum steering angular velocity.

The fall monitoring and safety protection response method uses the force sensor matrix on the walker handle to detect the user's upper limb force signal, uses the laser ranging sensor to detect the user's lower limb movement signal, performs multi-sensor data fusion through the Kalman filter algorithm, analyzes the user's movement speed, and uses the improved sequential probability ratio test method to detect whether the user has fallen on the fused data. If it is detected that the user has a tendency to fall, the user and the robot are reversed according to the expected position of the stable state to prevent the falling trend until a stable state is reached and brakes are applied, thereby forming a buffering movement until the user and the robot return to a stable state. The fall detection method determines whether a fall has occurred based on the improved sequential probability ratio test algorithm (PLT-SPRT). By adding preconditions and weights to the sequential probability ratio test (SPRT), the precondition is made to be the sum of the left and right handle force values, that is, the handles are completely released or the force increases significantly in an instant, and a higher weight is given to the upper limbs to optimize the decision function. When the decision function Δ(k) is greater than or equal to the lower threshold lnB, no correction is made. When it is less than lnB, the decision function Δ(k) is corrected to Δ′(k). Negative values will not be accumulated, further improving the detection delay and reducing the false positive rate.

The above is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any technician in the relevant technical field, without departing from the scope of the technical solution of the present invention, makes any form of equivalent replacement or modification to the technical solution and technical content disclosed in the present invention, which does not depart from the content of the technical solution of the present invention and still falls within the protection scope of the present invention.

Claims (10)

1. A method for detecting and actively protecting safe walking intentions of a walker, characterized in that sensor data is integrated through a sensor monitoring system to analyze the user's status, estimate the user's intention, detect and even predict the user's fall status in real time, and perform corresponding action control on the walker based on the analysis of the signal. When the walker detects danger, the walker executes corresponding protective measures. 2. The method for safe walking aid intention detection and active protection according to claim 1 is characterized in that the sensing monitoring system includes a thin film pressure sensor, a three-dimensional force sensor, a laser radar and a ranging sensor; the force sensor is placed at the handle of the walker to detect the upper limb force signal; the laser radar is placed in front of the walker to detect environmental information; the ranging sensor is placed on the front side of the walker and aligned with the lower limbs of the human body to detect the movement state of the lower limbs; the laser radar is installed in front of the walker to detect environmental obstacle information. 3. The safe walking assistance intention detection and active protection method for a walking aid according to claim 2 is characterized in that it includes a forward intention detection method and a braking intention detection method, the forward intention detection method includes acceleration intention detection, steering intention detection, uphill movement detection, downhill movement detection and obstacle avoidance intention detection; the braking intention detection method includes fall monitoring and safety protection response. 4. The method for detecting safe walking aid intention and active protection for a walker according to claim 3 is characterized in that the acceleration intention detection method adopts a force sensor on the armrest of the walker to identify the user's intention by sensing the user's force and force distribution characteristics; if the user wants to accelerate, the overall force on the armrest is increased, so that the force sensor detects the change in the user's force. 5. The method for detecting safe walking assistance intention and active protection for a walker according to claim 3 is characterized in that the uphill movement detection and downhill movement detection methods detect the current environment through a laser radar, and when it is detected that the user is going uphill, assisting uphill is implemented, and when it is detected that the user is going downhill, active speed control is performed to prevent the speed from dropping too quickly; the obstacle avoidance detection method, when the obstacle distance information is detected, implements obstacle avoidance steering, and drives the motor through the admittance and impedance controller to complete the corresponding action. 6. The method for safe walking intention detection and active protection for a walker according to claim 3 is characterized in that the steering intention detection method, based on the force interaction information between the user and the walker, is used as the signal input to control the direction of travel of the motor, and divides the steering characteristics of the walker into 5 labels, namely, a large left turn, a small left turn, a straight line, a small right turn and a large right turn. When the difference in the steering force of the two handles of the user is close to zero, the steering expectation is more likely to approach zero; when the difference in the steering force on both sides of the user slowly increases, the steering expectation begins to accelerate; when the difference in the steering force reaches a certain range, the steering expectation has entered a critical state and will no longer increase, and the walker will turn at the built-in maximum steering angular velocity. 7. The method for detecting safe walking aid intention and actively protecting a walker according to any one of claims 3 to 6 is characterized in that the force information feature involved is for the handle structure composition characteristics of a matrix of 8 thin film pressure sensors, and the handle force information is analyzed according to a pixel bitmap of ordinary display accuracy through a machine learning analysis method. When a certain thin film pressure sensor senses the force of the user, the signal generated by its own deformation passes through an amplifying circuit and an ADC circuit and enters the main control chip. The signal sampled by the main control chip can be considered as the RGB color value of the pixel, corresponding to the percentage of the force signal measured by the walker force sensor in the full range. The 16 thin film pressure sensors form an array of 16 pixels, and the sum of the thin film pressure sensors carried by the left and right force sensor handles corresponding to the 16 pixels is compared to determine the steering force difference between the left and right handles. 8. The method for detecting and actively protecting a walker with a safe walking intention according to claim 3 is characterized in that the obstacle avoidance intention detection and safety protection method uses obstacle distance information detected by a laser radar as a signal input to control the direction of travel of the motor, and divides the walking aid's steering characteristics into five labels, namely, a large left turn, a small left turn, straight ahead, a small right turn, and a large right turn. When no obstacle information is detected, its steering expectation is more likely to approach zero; and as the distance of the obstacle approaches, the steering expectation begins to accelerate; when the distance of the obstacle reaches a certain range, the steering expectation has entered a critical state and will no longer increase, and the walker will turn at the built-in maximum steering angular velocity. 9. The method for detecting and actively protecting a walker with a safe walking intention according to claim 3 is characterized in that the fall monitoring and safety protection response method uses a force sensor matrix on the handle of the walker to detect the upper limb force signal of the user, uses a laser ranging sensor to detect the lower limb motion signal of the user, performs multi-sensor data fusion through a Kalman filter algorithm, analyzes the user's movement speed, and uses an improved sequential probability ratio test method to detect whether the user falls on the fused data. If it is detected that the user has a tendency to fall, reverse movement is performed according to the expected position of the stable state of the user and the robot to prevent the falling tendency until a stable state is reached and braking is performed, thereby forming a buffering movement until the user and the robot return to a stable state. 10. The method for detecting and actively protecting walking aids with safe walking intentions according to claim 9 is characterized in that the fall detection method determines whether a fall has occurred based on an improved sequential probability ratio test algorithm, by adding preconditions and weights to the sequential probability ratio test, wherein the handle being completely released or the instantaneous force increase amplitude is used as a precondition, and the precondition is the sum of the superposition of the left and right handle force values, and a higher weight is given to the upper limbs to optimize the decision function, and when the decision function Δ(k) is greater than or equal to the lower threshold lnB, no correction is made, and when it is less than lnB, the decision function Δ(k) is corrected to Δ′(k), and negative values are not accumulated, thereby further improving the detection delay and reducing the misjudgment rate.