Robot-based operation fall monitoring and early warning method and system

Abstract

The present invention relates to the field of artificial intelligence technology, and discloses a robot-based operation fall monitoring and early warning method, including: collecting the operation data of the robot, smoothing the operation data to obtain smoothed data of the operation data; generating the acceleration change rate of the robot according to the smoothed data, and generating the inclination angle change value of the robot according to the smoothed data; collecting the acceleration change rate and the inclination angle change value as the state change value of the robot; collecting the robot's environmental data, and generating the update weight of the state change value according to the environmental data; jointly judging the robot according to the preset fall judgment logic, the update weight and the state change value to obtain the robot's fall early warning; generating the robot's response mechanism according to the fall early warning. The present invention also proposes a robot-based operation fall monitoring and early warning system. The present invention can improve the accuracy of the robot-based operation fall monitoring and early warning.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a robot-based operation fall monitoring and early warning method and system. Background Art

[0002] With the advancement of artificial intelligence, sensor technology, and robot control systems, the accuracy of robot fall warning systems has become increasingly important. If the accuracy of fall detection is insufficient, it will not only affect the completion of the robot's mission, but may also pose a safety hazard to the user and the surrounding environment.

[0003] With the advancement of artificial intelligence, sensor technology, and robot control systems, the accuracy of robot fall warning systems has become increasingly important. If the accuracy of fall detection is insufficient, it will not only affect the completion of the robot's mission, but may also pose a safety hazard to the user and the surrounding environment. The robot's operating environment is very complex. Uneven ground, obstacles, ambient lighting, etc. will affect the robot's motion trajectory and stability. Fall warning not only depends on the robot's own motion parameters, but also needs to consider factors in the surrounding environment. Existing systems often find it difficult to fully and accurately consider all these factors, resulting in monitoring errors. Therefore, how to improve the efficiency of robot-based operation fall monitoring and warning has become an urgent problem to be solved. Summary of the invention

[0004] The present invention provides a robot-based operation fall monitoring and early warning method and system, the main purpose of which is to solve the problem of low accuracy in the robot-based operation fall monitoring and early warning.

[0005] To achieve the above purpose, the present invention provides a robot-based operation fall monitoring and early warning method, comprising:

[0006] Collecting operation data of the robot, and performing smoothing processing on the operation data to obtain smoothed data of the operation data;

[0007] generating an acceleration change rate of the robot according to the smoothed data, and generating an inclination angle change value of the robot according to the smoothed data;

[0008] Collecting the acceleration change rate and the tilt angle change value as the state change value of the robot;

[0009] Collecting environmental data of the robot, and generating an update weight of the state change value according to the environmental data;

[0010] The robot is jointly judged according to the preset fall judgment logic, the update weight and the state change value to obtain a fall warning of the robot, wherein the preset fall judgment logic is:

[0011] ;

[0012] in, is a binary variable, is the rate of change of acceleration, is the threshold value of the acceleration rate of change, is the tilt angle change value, is the threshold of the tilt angle change value, Yes or, is the update weight corresponding to the acceleration change rate, is the update weight corresponding to the tilt angle change value;

[0013] A response mechanism of the robot is generated according to the fall warning.

[0014] Optionally, the performing smoothing on the operating data to obtain smoothed data of the operating data includes:

[0015] The operation data is smoothed using a preset sliding average algorithm to obtain smoothed data of the operation data, wherein the preset sliding average algorithm is:

[0016] ;

[0017] in, is the smoothed data, yes The running data at the moment, is a time stamp, is the size of the smoothing window.

[0018] Optionally, the performing smoothing on the operating data to obtain smoothed data of the operating data includes:

[0019] The operation data is smoothed using a preset weighted moving average algorithm to obtain smoothed data of the operation data, wherein the preset weighted moving average algorithm is:

[0020] ;

[0021] in, is the smoothed data of the operating data, is the attenuation factor, yes The running data at the moment, It is a time stamp.

[0022] Optionally, the performing smoothing on the operating data to obtain smoothed data of the operating data includes:

[0023] The operation data is smoothed using a preset bilinear transformation algorithm to obtain smoothed data of the operation data, wherein the preset bilinear transformation algorithm is:

[0024] ;

[0025] in, is the smoothed data of the operating data, is the time step, yes The running data at the moment, It is a time stamp.

[0026] Optionally, generating the acceleration change rate of the robot according to the smoothed data includes:

[0027] The acceleration rate of change of the robot is generated according to a preset acceleration rate generation algorithm and the smoothing data, wherein the preset acceleration rate generation algorithm is:

[0028] ;

[0029] in, is the rate of change of acceleration of the robot, It's time The robot's acceleration is The weight in direction, It's time The robot's acceleration is The weight in direction, It's time The robot's acceleration is The weight in direction, It is a time stamp.

[0030] Optionally, generating the tilt angle change value of the robot according to the smoothed data includes:

[0031] The tilt angle change value of the robot is generated according to a preset tilt angle change value generation algorithm and the smoothing data, wherein the preset tilt angle change value generation algorithm is:

[0032] ;

[0033] in, is the change in the robot's tilt angle, yes The angle between the robot and the ground at that moment, It is a time stamp.

[0034] Optionally, generating an update weight of the state change value according to the environmental data includes:

[0035] Extracting features from the environmental data to obtain data features of the environmental data;

[0036] Performing feature classification on the data features to obtain the data type of the environmental data;

[0037] An update weight for the state change value is generated according to the data type.

[0038] Optionally, the extracting features of the environmental data to obtain data features of the environmental data includes:

[0039] Filling missing values ​​in the environmental data to obtain filled data for the environmental data;

[0040] Performing anomaly detection on the filling data, and removing abnormal values ​​from the filling data according to the detection result of the anomaly detection to obtain standard data of the filling data;

[0041] Performing normalization processing on the standard data to obtain normalized data of the standard data;

[0042] The data features of the environmental data are generated according to the normalized data.

[0043] Optionally, the performing feature classification on the data features to obtain the data type of the environmental data includes:

[0044] Performing seasonal decomposition on the data features, and screening the data features for climate features according to the decomposition results of the seasonal decomposition to obtain climate features of the data features;

[0045] Performing feature clustering on the data features, and dividing the data features into static and dynamic features according to the clustering results of the feature clustering to obtain static features and dynamic features of the data features;

[0046] The data type of the environmental data is determined according to the climate characteristics, the static characteristics, and the dynamic characteristics.

[0047] In order to solve the above problems, the present invention also provides a robot-based operation fall monitoring and early warning system, the system comprising:

[0048] A data smoothing module is used to collect the operation data of the robot, and perform smoothing on the operation data to obtain smoothed data of the operation data;

[0049] A change rate generating module, used to generate an acceleration change rate of the robot according to the smooth data, and to generate a tilt angle change value of the robot according to the smooth data;

[0050] A state change value collection module, used for collecting the acceleration change rate and the tilt angle change value as the state change value of the robot;

[0051] An update weight generation module, used for collecting environmental data of the robot and generating an update weight of the state change value according to the environmental data;

[0052] A joint determination module is used to perform a joint determination on the robot according to a preset fall determination logic, the update weight and the state change value to obtain a fall warning of the robot, wherein the preset fall determination logic is:

[0053] ;

[0054] in, is a binary variable, is the rate of change of acceleration, is the threshold value of the acceleration rate of change, is the tilt angle change value, is the threshold of the tilt angle change value, Yes or, is the update weight corresponding to the acceleration change rate, is the update weight corresponding to the tilt angle change value;

[0055] A response mechanism generation module is used to generate a response mechanism of the robot according to the fall warning.

[0056] The present invention comprehensively considers the acceleration change rate and tilt angle change value of the robot, and does not rely solely on a single indicator to judge the state of the robot. This comprehensive analysis of multi-dimensional data can more comprehensively and accurately reflect the dynamic state of the robot, thereby improving the accuracy of the early warning. By smoothing the robot's operating data, the noise in the data can be eliminated, and the impact of external disturbances and sensor errors on the early warning system can be reduced. By collecting the robot's environmental data and combining these environmental data to dynamically adjust the update weights of the acceleration change rate and the tilt angle change value, adaptive adjustments can be made according to specific environmental changes. In this way, the early warning system can respond more accurately under different environmental conditions, and the fall judgment logic Thresholds of the acceleration change rate and the tilt angle change value are used and combined through a logical operator ("or") to make the determination of a fall more flexible. By generating a response mechanism for the robot based on a fall warning, the robot can be effectively prevented from causing further damage or failure after a fall occurs. By combining multiple factors (acceleration, tilt angle change, environmental data, etc.) for joint determination, the system can comprehensively evaluate whether the robot has fallen. This multi-factor determination method improves the accuracy of the system and reduces the misjudgment or missed judgment that may be caused by a single factor. Therefore, the present invention proposes a robot-based operation fall monitoring and early warning method and system, which can solve the problem of low accuracy of robot-based operation fall monitoring and early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of a flow chart of a robot-based operation fall monitoring and early warning method provided in one embodiment of the present invention;

[0058] Figure 2 A functional module diagram of a robot-based operation fall monitoring and early warning system provided in one embodiment of the present invention;

[0059] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0060] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0061] The embodiment of the present application provides a robot-based operation fall monitoring and early warning method. The execution subject of the robot-based operation fall monitoring and early warning method includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the robot-based operation fall monitoring and early warning method can be executed by software or hardware installed on a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), as well as basic cloud computing services such as big data and artificial intelligence platforms.

[0062] Reference Figure 1 FIG. 1 is a flow chart of a robot-based operation fall monitoring and early warning method provided in an embodiment of the present invention. In this embodiment, the robot-based operation fall monitoring and early warning method includes:

[0063] S1. Collecting operation data of the robot, and performing smoothing processing on the operation data to obtain smoothed data of the operation data.

[0064] In an embodiment of the present invention, the collection of the robot's operating data includes: collecting acceleration data, tilt angle data, angular velocity data and position data of each axis of the robot, wherein the acceleration data can be obtained through an accelerometer to monitor the robot's motion state and its changes in different directions; collecting the robot's tilt angle relative to the ground or horizontal plane through a tilt sensor or a gyroscope, and these data can help determine whether the robot is in a tilted or fallen state; measuring the robot's angular velocity through a gyroscope, especially when the robot is rotating or tilting, the angular velocity data can reflect the speed of change of the robot's posture; obtaining the robot's position information through a GPS module or other positioning sensors, these data help to understand the robot's running trajectory and its positional relationship in the environment.

[0065] In the embodiment of the present invention, the step of smoothing the operating data to obtain smoothed data of the operating data includes:

[0066] The operation data is smoothed using a preset sliding average algorithm to obtain smoothed data of the operation data, wherein the preset sliding average algorithm is:

[0067] ;

[0068] in, is the smoothed data, yes The running data at the moment, is a time stamp, is the size of the smoothing window.

[0069] In the embodiment of the present invention, the size of the smoothing window, that is, the past The number of data points.

[0070] In detail, the preset sliding average algorithm is simple to calculate, easy to understand and implement, and it can effectively eliminate random noise in the running data, especially when the data changes relatively slowly.

[0071] In detail, Indicates at a point in time The smoothed data obtained by the sliding average algorithm is the original data The result after smoothing is Indicates at a point in time The original running data, is the current time stamp, that is, the time point at which you want to calculate the smoothed data. is the size of the smoothing window, that is, the number of historical data points used to calculate the current smoothed value during the sliding average process, is within the time window The sum of the data points, from arrive , and then divided by , to achieve averaging.

[0072] In detail, suppose the following original data: D(1)=5, D(2)=6, D(3)=7, D(4)=8, D(5)=9. If we choose the window size n=3 and calculate the smoothed data at t=4: .

[0073] In detail, the preset sliding average algorithm is used to smooth the running data. Its core is to use the original data within a window for averaging, thereby reducing noise and fluctuations. The window size is a key parameter that determines the smoothing effect: the larger the window, the stronger the smoothing effect, but it may also cause lag; the smaller the window, the weaker the smoothing effect, but the faster the response to data changes.

[0074] In the embodiment of the present invention, the step of smoothing the operating data to obtain smoothed data of the operating data includes:

[0075] The operation data is smoothed using a preset weighted moving average algorithm to obtain smoothed data of the operation data, wherein the preset weighted moving average algorithm is:

[0076] ;

[0077] in, is the smoothed data of the operating data, is the attenuation factor, yes The running data at the moment, It is a time stamp.

[0078] In detail, the preset weighted moving average algorithm assigns different weights to data at different time points, and generally the weight of data at a close distance is larger, and the weight of data far from the current time point is smaller.

[0079] In detail, the decay factor indicates the degree to which the weight decays over time.

[0080] In detail, the weighted moving average algorithm responds faster than the ordinary sliding average algorithm because the recent samples have a larger weight and can better reflect the real-time changes of the data. By adjusting the attenuation factor, the emphasis on historical data can be controlled to adapt to different data characteristics.

[0081] In detail, Indicates at a point in time Smoothed data on the running data The result after smoothing is is the attenuation factor, also called the smoothing coefficient. The larger the value, the more recent the data. The higher the importance attached to , the smaller the influence of past data on the smoothing result; The smaller the value, the more accurate the past data is. The larger the weight of , the smoothing result changes more slowly and has a larger lag; is the smoothed data of the previous moment.

[0082] In detail, each new smoothed data They are all calculated by weighted average. Specifically, it is the raw data at the current moment. and the smoothed data of the previous moment According to weight and In this way, the impact of the current raw data on the smoothed value is limited, and the impact of past data on the smoothed value will gradually weaken over time.

[0083] In detail, the algorithm has a recursive nature, that is, each new smoothing value depends on the smoothing value at the previous moment. This recursive relationship enables the algorithm to respond to new data relatively quickly while maintaining a certain degree of smoothing effect and avoiding excessive fluctuations.

[0084] In the embodiment of the present invention, the step of smoothing the operating data to obtain smoothed data of the operating data includes:

[0085] The operation data is smoothed using a preset bilinear transformation algorithm to obtain smoothed data of the operation data, wherein the preset bilinear transformation algorithm is:

[0086] ;

[0087] in, is the smoothed data of the operating data, is the time step, yes The running data at the moment, It is a time stamp.

[0088] In detail, the preset bilinear transformation algorithm can smooth the signal well while retaining the trend of the signal, can adapt to different time steps, and performs well in dynamically changing data.

[0089] In detail, Represents the time step, usually the time interval between adjacent data points. This parameter is used to adjust the degree of smoothing.

[0090] In detail, Represents the current time data And the data of the first two moments and Perform weighted averaging, where the data at the current moment The weight received is 1, and the data at the previous moment The weight received is 2, and the data of the first two moments The weight received is 1.

[0091] In detail, It controls the overall scaling of the result after weighted combination. This coefficient will affect the strength of the smoothing effect. The sum of squares , which enables the algorithm to adjust the smoothing effect when the time step is large.

[0092] In detail, the size of the time step directly affects the degree of the smoothing process. A larger time step will weaken the influence of the current data point, thereby increasing the reliance on historical data points. This algorithm adds the first two data points in the smoothing process to avoid relying solely on the previous data point, thereby improving the algorithm's smoothing effect on data trends.

[0093] In detail, the algorithm uses a weighted approach to combine the raw data, where the weight of the current moment data is 1, the weight of the previous moment data is 2, and the weight of the data two moments before is 1. This weighted approach can reduce high-frequency noise in the data and smooth out sudden fluctuations. The weighted value helps balance the contribution of current data and past data, avoiding the impact of over-reliance on current data.

[0094] In detail, the role of the normalization factor is to adjust the size of the smoothed data to ensure that the overall change of the smoothed data is not too drastic. As the time step increases, the normalization factor will increase, thereby reducing data fluctuations.

[0095] S2. Generate the acceleration change rate of the robot according to the smooth data, and generate the tilt angle change value of the robot according to the smooth data.

[0096] In an embodiment of the present invention, generating the acceleration change rate of the robot according to the smoothed data includes:

[0097] The acceleration rate of change of the robot is generated according to a preset acceleration rate generation algorithm and the smoothing data, wherein the preset acceleration rate generation algorithm is:

[0098] ;

[0099] in, is the rate of change of acceleration of the robot, It's time The robot's acceleration is The weight in direction, It's time The robot's acceleration is The weight in direction, It's time The robot's acceleration is The weight in direction, It is a time stamp.

[0100] In detail, Indicates time The rate of change of the robot's acceleration, that is, the amount of change in acceleration over time, measures the speed at which the robot's acceleration changes in all directions. , , Respectively represent the robot at time When, along The acceleration components of the axes. Typically, the robot acceleration is affected by multiple forces in these three directions, so the acceleration vector is usually expressed in three-dimensional space.

[0101] In an embodiment of the present invention, generating the tilt angle change value of the robot according to the smoothed data includes:

[0102] The tilt angle change value of the robot is generated according to a preset tilt angle change value generation algorithm and the smoothing data, wherein the preset tilt angle change value generation algorithm is:

[0103] ;

[0104] in, is the change in the robot's tilt angle, yes The angle between the robot and the ground at that moment, It is a time stamp.

[0105] In detail, Indicates the change in the robot's tilt angle at the moment, which measures the change in the robot's tilt angle relative to the ground; Indicates that the robot is at time The angle between the ground and the ground, usually expressed in degrees or radians. This angle is usually measured by sensors (such as accelerometers and gyroscopes). Indicates that the robot is at time The angle with the ground.

[0106] In detail, we need to know the robot at the time and The tilt angle and , these angle values ​​can be calculated using smoothed sensor data; the absolute value operation ensures that a positive value is obtained as the change in tilt angle regardless of the direction of the angle change (increase or decrease).

[0107] Specifically, the preset tilt angle change value generation algorithm measures the change in the robot's tilt angle relative to the ground. A large value means that the robot's tilt angle has changed a lot, usually indicating that the robot has experienced a large tilt or center of gravity change; a small value means that the robot's tilt angle has changed a lot, usually indicating that the robot has experienced a large tilt or center of gravity change; A value of 0 indicates that the robot maintains a relatively stable posture.

[0108] S3. Collect the acceleration change rate and the tilt angle change value as the state change value of the robot.

[0109] In the embodiment of the present invention, the state change value is an indicator to measure the degree of change of the robot state at a certain moment compared with the previous moment, and is usually used to describe the dynamic changes of the robot's acceleration and posture during the movement. Gathering the acceleration change rate and the tilt angle change value means that we combine the two key change information of the robot during the movement process to comprehensively evaluate the robot's movement state.

[0110] In detail, the acceleration rate reflects the rapid changes in the robot's motion state, such as acceleration, deceleration, or sudden speed changes.

[0111] In detail, the tilt angle change value represents the change in the robot's posture, which is usually measured by the tilt angle of the robot relative to the ground.

[0112] S4. Collect environmental data of the robot, and generate an update weight for the state change value according to the environmental data.

[0113] In an embodiment of the present invention, assuming that an intelligent robot is navigating in a complex environment, environmental data collection may include: position data, posture data, and environmental obstacle data, wherein the position coordinates of the robot in the environment are obtained through GPS, inertial navigation system (INS), visual sensors (such as SLAM technology), laser radar (LiDAR), etc., and the position data is the basis for the robot to perform positioning and path planning; sensors such as accelerometers, gyroscopes, and magnetometers are used to obtain the robot's pitch angle, roll angle, and yaw angle, and these data can reflect the robot's posture changes; the position, shape, size, and other information of obstacles in the environment are detected through laser radar (LiDAR), ultrasonic sensors, cameras (computer vision), and other devices. The laser radar can generate high-precision point cloud data of the robot's surroundings, thereby helping the robot to identify surrounding obstacles; the three-dimensional structure information of the robot's surroundings is obtained through a depth camera or a stereo vision system, and these data help the robot understand the spatial distance and the relative position of obstacles.

[0114] In detail, environmental data also includes lighting data, which uses optical sensors (such as LDR, CCD, CMOS sensors, etc.) to obtain the light intensity in the environment. This is very important for indoor and outdoor navigation, especially for judging the working efficiency of visual sensors, because changes in lighting will affect the image processing effect of the computer vision system, and thus affect the robot's visual perception ability.

[0115] In the embodiment of the present invention, the step of generating the update weight of the state change value according to the environmental data includes:

[0116] Extracting features from the environmental data to obtain data features of the environmental data;

[0117] Performing feature classification on the data features to obtain the data type of the environmental data;

[0118] An update weight for the state change value is generated according to the data type.

[0119] In detail, suppose there is a set of environmental data, which includes the following:

[0120] Temperature: 20°C, 21°C, 22°C, 19°C

[0121] Humidity: 65%, 60%, 63%, 70%

[0122] Light intensity: 100 lux, 200 lux, 150 lux, 130 lux.

[0123] In detail, the extracting features of the environmental data to obtain data features of the environmental data includes:

[0124] Filling missing values ​​in the environmental data to obtain filled data for the environmental data;

[0125] Performing anomaly detection on the filling data, and removing abnormal values ​​from the filling data according to the detection result of the anomaly detection to obtain standard data of the filling data;

[0126] Performing normalization processing on the standard data to obtain normalized data of the standard data;

[0127] The data features of the environmental data are generated according to the normalized data.

[0128] In detail, assuming that in the actual collected data, humidity is missing at the third time point, the missing value can be filled using mean filling or other methods. For example:

[0129] Original humidity data: 65%, 60%, missing, 70%

[0130] Humidity data after filling: 65%, 60%, 66%, 70%.

[0131] In detail, outlier detection is performed on the padded data to identify whether there is any data that deviates from the normal range. Suppose an outlier is detected in the temperature data, and the temperature suddenly changes to 50°C at a certain moment, which is an anomaly.

[0132] Further, after anomaly detection, the temperature of 50°C is marked as abnormal and removed, and may be restored to the normal temperature range using interpolation.

[0133] In detail, in order to eliminate the influence of different data units, all environmental data are normalized so that the value of each feature data is in the same range, for example, between [0, 1]. The normalized data is:

[0134] Temperature normalization: [0.5, 0.6, 0.7, 0.4]

[0135] Humidity normalization: [0.5, 0.4, 0.45, 0.6]

[0136] Light intensity normalization: [0.1, 0.2, 0.15, 0.13].

[0137] In general, through the above-mentioned steps of filling, outlier removal, normalization, etc., normalized environmental data are generated as the data features of the robot, which serve as the basis for subsequent feature classification.

[0138] In detail, the feature classification of the data features to obtain the data type of the environmental data includes:

[0139] Performing seasonal decomposition on the data features, and screening the data features for climate features according to the decomposition results of the seasonal decomposition to obtain climate features of the data features;

[0140] Performing feature clustering on the data features, and dividing the data features into static and dynamic features according to the clustering results of the feature clustering to obtain static features and dynamic features of the data features;

[0141] The data type of the environmental data is determined according to the climate characteristics, the static characteristics, and the dynamic characteristics.

[0142] In detail, the static characteristics include: ground slope, distance to obstacles, etc.; dynamic characteristics include: robot speed, acceleration, tilt angle, etc.; climate characteristics include: humidity, temperature, wind speed, etc.

[0143] In detail, in the process of robot perception and environmental data analysis, in order to better understand and process environmental data, it is usually necessary to perform feature classification, cluster analysis and climate feature screening on the data. This process not only helps to understand environmental changes from multiple dimensions, but also improves the robot's ability to adapt to environmental changes.

[0144] Specifically, data classification and feature extraction can be completed through methods such as seasonal decomposition, feature clustering, and climate feature screening, thereby dividing the data into three categories: static features, dynamic features, and climate features.

[0145] In detail, seasonal decomposition refers to decomposing time series data into multiple components, usually including trend components, seasonal components, and residual components. Through seasonal decomposition, robots can better understand the laws and patterns of data changes over time, and make more intelligent decisions.

[0146] For example, in environmental data, variables such as temperature, humidity, and wind speed may fluctuate regularly with seasonal changes. Through seasonal decomposition, the seasonal components of these data can be extracted and climate characteristics can be identified.

[0147] In detail, feature clustering is a method of dividing data features into different categories according to similarity. Through clustering, data with similar features can be identified, thus providing a basis for subsequent decision-making and analysis. The process of feature clustering can use various algorithms, such as K-means clustering, hierarchical clustering, DBSCAN, etc.

[0148] Specifically, static features refer to environmental data features that remain stable and do not change easily over a certain period of time. These features play a greater role in the long-term decision-making of robots because they usually do not change significantly over the long term in the environment. Common static features include:

[0149] Ground slope: The inclination of the ground or terrain affects the robot's ability to move on different terrains.

[0150] Obstacle distance: The static distance of obstacles in the environment relative to the robot, usually obtained through sensors such as lidar and ultrasound.

[0151] The shape of environmental objects: such as the shape of walls, the structure of buildings, etc., which usually do not change in a short period of time.

[0152] In detail, dynamic features refer to data features that change with time, environment, or robot state. These features usually affect the robot's real-time decision-making, path planning, or control strategy. Common dynamic features include: robot speed: the speed at which the robot moves in the environment, affecting path planning and motion control; robot acceleration: acceleration affects the smoothness of the robot's movement and its response to external disturbances; tilt angle: the robot's posture angle, which may affect its stability and task execution ability; changes in the robot's distance to obstacles: during movement, changes in the distance to obstacles will affect the execution of the obstacle avoidance algorithm.

[0153] In detail, through feature clustering, the robot can better identify which features are constant parts of the environment and which are dynamically changing over time or with external conditions, thereby making more accurate decisions.

[0154] S5. Perform a joint judgment on the robot according to a preset fall judgment logic, the update weight and the state change value to obtain a fall warning of the robot.

[0155] In the embodiment of the present invention, the step of jointly judging the robot according to the preset fall judgment logic, the update weight and the state change value to obtain a fall warning of the robot includes:

[0156] When the product of the acceleration change rate and the update weight corresponding to the acceleration change rate is greater than the threshold value of the acceleration change rate, generating an acceleration warning of the robot according to the acceleration change rate;

[0157] When the product of the acceleration change rate and the update weight corresponding to the acceleration change rate is less than or equal to the threshold of the acceleration change rate, determining whether the product of the inclination angle change value and the update weight corresponding to the inclination angle change value is greater than the threshold of the inclination angle change value;

[0158] When the product of the tilt angle change value and the update weight corresponding to the tilt angle change value is greater than the threshold value of the tilt angle change value, generating a tilt warning of the robot according to the tilt angle change value;

[0159] A fall warning of the robot is generated according to the acceleration warning and the tilt warning.

[0160] In detail, each measurement value (such as acceleration rate) has an update weight associated with it, and these weights are adjusted based on the real-time monitoring status.

[0161] In detail, when the product of the acceleration change rate and its corresponding update weight is greater than the set acceleration change rate threshold, it means that the acceleration changes dramatically and there may be a risk of falling, so an acceleration warning is generated.

[0162] In detail, if the acceleration warning is not triggered, the system will next determine whether it is necessary to generate a tilt warning based on the tilt angle change value; when the product of the acceleration change rate and the update weight is less than or equal to the threshold, it means that the acceleration change is not significant, and the tilt angle change value is then determined; whether the product of the tilt angle change value and the update weight corresponding to the tilt angle change value is greater than the set tilt angle change value threshold, if greater, a tilt warning is generated.

[0163] In detail, the preset fall determination logic is:

[0164] ;

[0165] in, is a binary variable, is the rate of change of acceleration, is the threshold value of the acceleration rate of change, is the tilt angle change value, is the threshold of the tilt angle change value, Yes or, is the update weight corresponding to the acceleration change rate, is the update weight corresponding to the tilt angle change value.

[0166] In detail, if the acceleration warning (based on the acceleration change rate) has been triggered, there is no need to further determine the tilt angle, and a fall warning is directly generated.

[0167] If the acceleration warning is not triggered, it is further determined whether a tilt warning is generated, and finally a final fall warning is generated based on the acceleration warning and the tilt warning.

[0168] In detail, It is a binary variable indicating whether a fall warning occurs (1 for occurrence and 0 for nonoccurrence).

[0169] S6. Generate a response mechanism of the robot according to the fall warning.

[0170] In the embodiment of the present invention, when the robot detects that the acceleration change rate or the tilt angle change exceeds a set threshold, a fall warning signal is generated through the fall determination logic mentioned above.

[0171] In detail, once the fall warning is triggered, the robot will perform a series of actions or processes according to the set response strategy, which may include the following response methods: When the fall warning signal is issued, the robot may reduce the risk of further falls by reducing the movement speed, automatically stopping, or performing self-stabilization operations. For example, the robot can increase standing stability by adjusting the joint angles and enhancing the support force of the drive to prevent a complete fall; if the robot is in motion, the response mechanism may activate the emergency braking mechanism to immediately slow down or stop all movements of the robot. Avoid exacerbating the potential risk of falling by reducing speed or stopping completely; to ensure that people around know that the robot is in a dangerous state, the robot can issue a warning through a sound alarm, a visual signal (such as a flashing light), or other communication methods. The prompt may convey a fall warning to the operator through a built-in speaker or other external device.

[0172] The present invention comprehensively considers the acceleration change rate and the tilt angle change value of the robot, and does not rely solely on a single indicator to judge the state of the robot. This comprehensive analysis of multi-dimensional data can more comprehensively and accurately reflect the dynamic state of the robot, thereby improving the accuracy of the early warning. By smoothing the robot's operating data, the noise in the data can be eliminated, and the impact of external disturbances and sensor errors on the early warning system can be reduced. By collecting the robot's environmental data and combining these environmental data to dynamically adjust the update weights of the acceleration change rate and the tilt angle change value, adaptive adjustments can be made according to specific environmental changes. In this way, the early warning system can respond more accurately under different environmental conditions, and the fall judgment logic can be more accurate. The thresholds of the acceleration change rate and the tilt angle change value are used and combined through a logical operator ("or") to make the determination of a fall more flexible. By generating a response mechanism for the robot based on a fall warning, the robot can be effectively prevented from causing further damage or failure after a fall occurs. By combining multiple factors (acceleration, tilt angle change, environmental data, etc.) for joint determination, the system can comprehensively evaluate whether the robot has fallen. This multi-factor determination method improves the accuracy of the system and reduces the misjudgment or missed judgment that may be caused by a single factor. Therefore, the present invention proposes a robot-based operation fall monitoring and early warning method, which can solve the problem of low accuracy of the robot-based operation fall monitoring and early warning.

[0173] like Figure 2 , which is a functional module diagram of a robot-based operation fall monitoring and early warning system provided in one embodiment of the present invention.

[0174] The robot-based operation fall monitoring and early warning system 100 of the present invention can be installed in an electronic device. According to the functions to be implemented, the robot-based operation fall monitoring and early warning system 100 can include a data smoothing module 101, a change rate generation module 102, a state change value collection module 103, an update weight generation module 104, a joint determination module 105, and a response mechanism generation module 106. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0175] In this embodiment, the functions of each module / unit are as follows:

[0176] The data smoothing module 101 is used to collect the operation data of the robot, and perform smoothing on the operation data to obtain smoothed data of the operation data;

[0177] The change rate generating module 102 is used to generate the acceleration change rate of the robot according to the smooth data, and to generate the tilt angle change value of the robot according to the smooth data;

[0178] The state change value collection module 103 is used to collect the acceleration change rate and the tilt angle change value into the state change value of the robot;

[0179] The update weight generation module 104 is used to collect environmental data of the robot and generate an update weight of the state change value according to the environmental data;

[0180] The joint determination module 105 is used to perform a joint determination on the robot according to a preset fall determination logic, the update weight and the state change value to obtain a fall warning of the robot, wherein the preset fall determination logic is:

[0181] ;

[0182] in, is a binary variable, is the rate of change of acceleration, is the threshold value of the acceleration rate of change, is the tilt angle change value, is the threshold of the tilt angle change value, Yes or, is the update weight corresponding to the acceleration change rate, is the update weight corresponding to the tilt angle change value;

[0183] The response mechanism generating module 106 is used to generate a response mechanism of the robot according to the fall warning.

[0184] In the several embodiments provided by the present invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0185] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0186] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0187] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0188] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A robot-based running fall monitoring and early warning method, characterized in that: The method comprises: Collecting operation data of the robot, and performing smoothing processing on the operation data to obtain smoothed data of the operation data; generating an acceleration change rate of the robot according to the smoothed data, and generating an inclination angle change value of the robot according to the smoothed data; Collecting the acceleration change rate and the tilt angle change value as the state change value of the robot; Collect environmental data of the robot, and generate update weights of the state change values ​​according to the environmental data, wherein the generating update weights of the state change values ​​according to the environmental data comprises: extracting features from the environmental data to obtain data features of the environmental data, performing seasonal decomposition on the data features, performing climate feature screening on the data features according to the decomposition results of the seasonal decomposition to obtain climate features of the data features, performing feature clustering on the data features, performing dynamic and static feature division on the data features according to the clustering results of the feature clustering to obtain static features and dynamic features of the data features, determining the data type of the environmental data according to the climate features, the static features, and the dynamic features, and generating update weights of the state change values ​​according to the data type; The robot is jointly judged according to the preset fall judgment logic, the update weight and the state change value to obtain a fall warning of the robot, wherein the preset fall judgment logic is: ； in, is a binary variable, is the rate of change of acceleration, is the threshold value of the acceleration rate of change, is the tilt angle change value, is the threshold of the tilt angle change value, Yes or, is the update weight corresponding to the acceleration change rate, is the update weight corresponding to the tilt angle change value; A response mechanism of the robot is generated according to the fall warning.

2. The robot-based operation fall monitoring and early warning method according to claim 1, characterized in that: The step of smoothing the operating data to obtain smoothed data of the operating data includes: The operation data is smoothed using a preset sliding average algorithm to obtain smoothed data of the operation data, wherein the preset sliding average algorithm is: ； in, is the smoothed data, yes The running data at the moment, is a time stamp, is the size of the smoothing window.

3. The robot-based operation fall monitoring and early warning method according to claim 1, characterized in that: The step of smoothing the operating data to obtain smoothed data of the operating data includes: The operation data is smoothed using a preset weighted moving average algorithm to obtain smoothed data of the operation data, wherein the preset weighted moving average algorithm is: ； in, is the smoothed data of the operating data, is the attenuation factor, yes The running data at the moment, It is a time stamp.

4. The robot-based operation fall monitoring and early warning method according to claim 1, characterized in that: The step of smoothing the operating data to obtain smoothed data of the operating data includes: The operation data is smoothed using a preset bilinear transformation algorithm to obtain smoothed data of the operation data, wherein the preset bilinear transformation algorithm is: ； in, is the smoothed data of the operating data, is the time step, yes The running data at the moment, Is a time stamp.

5. The robot-based operation fall monitoring and early warning method according to claim 1, characterized in that: Generating the acceleration change rate of the robot according to the smoothed data includes: The acceleration rate of change of the robot is generated according to a preset acceleration rate generation algorithm and the smoothing data, wherein the preset acceleration rate generation algorithm is: ； in, is the rate of change of acceleration of the robot, It's time The robot's acceleration is The weight in direction, It's time The robot's acceleration is The weight in direction, It's time The robot's acceleration is The weight in direction, It is a time stamp.

6. The robot-based operation fall monitoring and early warning method according to claim 1, characterized in that: Generating the tilt angle change value of the robot according to the smoothed data includes: The tilt angle change value of the robot is generated according to a preset tilt angle change value generation algorithm and the smoothing data, wherein the preset tilt angle change value generation algorithm is: ； in, is the change in the robot's tilt angle, yes The angle between the robot and the ground at that moment, It is a time stamp.

7. The robot-based operation fall monitoring and early warning method according to claim 1, characterized in that: The extracting features of the environmental data to obtain data features of the environmental data includes: Filling missing values ​​in the environmental data to obtain filled data for the environmental data; Performing anomaly detection on the filling data, and removing abnormal values ​​from the filling data according to the detection result of the anomaly detection to obtain standard data of the filling data; Performing normalization processing on the standard data to obtain normalized data of the standard data; The data features of the environmental data are generated according to the normalized data.

8. A robot-based running fall monitoring and early warning system, characterized in that: The system comprises: A data smoothing module is used to collect the operation data of the robot, and perform smoothing on the operation data to obtain smoothed data of the operation data; A change rate generating module, used to generate an acceleration change rate of the robot according to the smooth data, and to generate a tilt angle change value of the robot according to the smooth data; A state change value collection module, used for collecting the acceleration change rate and the tilt angle change value as the state change value of the robot; An update weight generation module is used to collect environmental data of the robot and generate update weights of the state change values ​​according to the environmental data, wherein the generation of update weights of the state change values ​​according to the environmental data includes: extracting data features of the environmental data, performing seasonal decomposition on the data features, performing climate feature screening on the data features according to the decomposition results of the seasonal decomposition to obtain climate features of the data features, performing feature clustering on the data features, performing dynamic and static feature division on the data features according to the clustering results of the feature clustering to obtain static features and dynamic features of the data features, determining the data type of the environmental data according to the climate features, the static features and the dynamic features, and generating update weights of the state change values ​​according to the data type; A joint determination module is used to perform a joint determination on the robot according to a preset fall determination logic, the update weight and the state change value to obtain a fall warning of the robot, wherein the preset fall determination logic is: ； in, is a binary variable, is the rate of change of acceleration, is the threshold value of the acceleration rate of change, is the tilt angle change value, is the threshold of the tilt angle change value, Yes or, is the update weight corresponding to the acceleration change rate, is the update weight corresponding to the tilt angle change value; A response mechanism generation module is used to generate a response mechanism of the robot according to the fall warning.

Images