Motion control method applied to stair climbing of sweeping robot

Abstract

The invention relates to the field of robot control, and discloses a motion control method applied to a floor sweeping robot to climb stairs, which comprises the following steps: analyzing a section height difference in real-time detection data to determine a stair climbing trigger condition of the floor sweeping robot; analyzing correlation factors of the body pitch angle and the leg expansion and contraction amount of the sweeping robot to balance so as to calculate the horizontal stability of the body of the sweeping robot; analyzing leg stretching characteristics and wheel driving characteristics of the sweeping robot at different step heights so as to calculate an action adaptation coefficient of the sweeping robot during operation; the terrain physical attribute of the current step environment of the sweeping robot is analyzed, and the action fitness of cooperation of legs and wheels of the sweeping robot is analyzed; and stair climbing action control of the sweeping robot is executed to complete stair climbing operation, if it is detected that the stair climbing trigger condition is met again, stair climbing processing continues to be executed, and otherwise, stair climbing control is quitted. According to the invention, the accuracy of movement control applied to the sweeping robot to go upstairs can be improved.

Description

Technical Field

[0001] This invention relates to a motion control method for a sweeping robot climbing stairs, belonging to the field of robot control. Background Technology

[0002] The technology of robotic vacuum cleaners has become increasingly mature, and significant progress has been made in the fields of intelligence and automation. Its core advantages are reflected in many aspects: In terms of navigation, mainstream products are equipped with LiDAR or visual SLAM technology, which can achieve accurate mapping and path planning, and support functions such as zoned cleaning and virtual wall settings.

[0003] However, existing robotic vacuum cleaners lack the wheel and leg structure to overcome obstacles of varying heights, such as stairs, resulting in persistent blind spots in stairwells and failing to meet users' needs for thorough cleaning of the entire house. Summary of the Invention

[0004] This invention provides a motion control method for a robotic vacuum cleaner to climb stairs, the main purpose of which is to improve the accuracy of motion control for a robotic vacuum cleaner to climb stairs.

[0005] To achieve the above objectives, the present invention provides a motion control method for a sweeping robot climbing stairs, comprising:

[0006] The system collects elevation sequence data of the terrain in front of the robot vacuum cleaner, including real-time detection data and historical stair climbing elevation records. It analyzes the segment elevation differences in the real-time detection data to determine the stair climbing trigger conditions of the robot vacuum cleaner.

[0007] Based on the aforementioned stair-climbing triggering conditions, the correlation factors between the robot's body pitch angle and leg extension amount and balance are analyzed to calculate the robot's body horizontal stability.

[0008] Based on the historical stair-climbing elevation records, the leg extension and wheel drive characteristics of the sweeping robot at different step heights are analyzed to calculate the motion adaptation coefficient of the sweeping robot during operation.

[0009] The terrain physical properties of the current step environment of the robot vacuum cleaner are analyzed, and the motion adaptability of the robot vacuum cleaner's legs and wheels is analyzed based on the terrain physical properties.

[0010] Based on the robot's horizontal stability, the action adaptation coefficient, and the action adaptability, the robot vacuum cleaner performs stair climbing control to complete the stair climbing operation. If the stair climbing trigger condition is detected again, the stair climbing process continues; otherwise, the stair climbing control is exited.

[0011] Optionally, the analysis of the segment elevation difference in the real-time detection data includes:

[0012] The elevation values ​​in the real-time detection data are sorted to obtain an ordered elevation sequence;

[0013] The ordered elevation sequence is subjected to adjacent value pairing processing to obtain elevation value pairs;

[0014] Calculate the difference between each pair of elevation values ​​to obtain the original elevation difference;

[0015] The original elevation difference is processed to be undirected, resulting in an undirected elevation difference;

[0016] Based on the non-directional elevation difference, the segment elevation difference in the real-time detection data is obtained.

[0017] Optionally, the step of analyzing the segment elevation differences in the real-time detection data to determine the stair-climbing trigger conditions of the sweeping robot includes:

[0018] Analyze the detection segments corresponding to the elevation differences in the sections, and based on the detection segments, perform clustering processing on the elevation differences in the sections to obtain the clustered elevation differences in the sections;

[0019] Continuity detection is performed on the distribution segments corresponding to the elevation differences of the clustered segments to obtain continuous segment groups;

[0020] The number of segments in the continuous segment group is counted to obtain the number of valid segments;

[0021] The number of valid segments is compared with the preset minimum number of consecutive segments to obtain the result that the quantity meets the standard;

[0022] When the quantity meets the requirements, the flatness of the overall distribution of the continuous segment group is verified to obtain the terrain verification result.

[0023] If the flatness verification result is consistent with the staircase terrain characteristics, then the stair-climbing trigger condition is generated.

[0024] Optionally, the step of analyzing the correlation factors between the robot's body pitch angle and leg extension / retraction amount and balance based on the stair-climbing trigger condition, in order to calculate the robot's body horizontal stability, includes:

[0025] When the stair-climbing trigger condition is met, the motion characteristic data of the sweeping robot is acquired and captured in real time to obtain the raw motion data.

[0026] The original motion parameters are subjected to noise filtering to obtain the target motion parameters;

[0027] The pitch angle and leg extension / retraction in the target motion parameters are correlated and decomposed to obtain the balancing correlation factors;

[0028] Based on the aforementioned balance-related factors, a balance determination model for horizontal posture is constructed.

[0029] Using the aforementioned balance determination model, the horizontal stability of the sweeping robot is calculated.

[0030] Optionally, calculating the horizontal stability of the robot vacuum cleaner using the balance determination model includes:

[0031] Extract the pitch angle deviation term and leg extension deviation term from the balance determination model;

[0032] Time series sampling was performed on the pitch angle deviation term and the leg extension deviation term to obtain multi-time deviation data;

[0033] Based on the multi-time deviation data and preset weighting coefficients, a steady-state calculation equation for the sweeping robot is constructed.

[0034] Based on the stability calculation equation, the horizontal stability of the robot vacuum cleaner is calculated using the following formula:

[0035]

[0036] Where A represents the robot vacuum's horizontal stability, B a Let B0 represent the pitch angle at time a in the multi-time deviation data, and α represent the target pitch angle. B D represents the weighting coefficient for the pitch angle. a This represents the leg extension / retraction amount at time a in the multi-time deviation data, where D0 represents the target extension / retraction amount, and k D The weighting coefficient represents the leg extension / retraction amount, where a represents the sampling time and n represents the number of sampling times.

[0037] Optionally, the step of analyzing the leg extension and wheel drive characteristics of the sweeping robot at different step heights based on the historical stair-climbing elevation records includes:

[0038] The historical stair-climbing elevation records are grouped by height interval to obtain stair-climbing action segments within each interval;

[0039] The leg trajectory is segmented from the segment of climbing the stairs in the interval to obtain the leg extension and contraction sequence;

[0040] The extreme values ​​of the extension and contraction amounts in the leg extension and contraction sequence are statistically analyzed to obtain the leg extension length characteristics;

[0041] The wheel rotation speed is stripped from the segment of the stair-climbing motion in the section to obtain the wheel drive curve;

[0042] The power consumption characteristics of the wheel are obtained by performing power consumption integration on the wheel drive curve.

[0043] By combining the leg extension length characteristics and the wheel power consumption characteristics, the leg extension and wheel drive characteristics of the sweeping robot at different step heights are generated.

[0044] Optionally, the step of performing power consumption integration on the wheel drive curve to obtain wheel power consumption characteristics includes:

[0045] Voltage and current components are separated from the wheel drive curve to obtain a set of electrical power elements;

[0046] The set of electrical power elements is divided into action state units to obtain a power consumption calculation unit;

[0047] The power consumption density of the power consumption calculation unit is calculated to obtain the state power consumption spectrum;

[0048] By performing rotational speed-power coupling correlation on the state power consumption spectrum, the driving energy efficiency characteristics are obtained;

[0049] The drive energy efficiency characteristics are calibrated and fused within a high-range interval to generate wheel power consumption characteristics.

[0050] Optionally, the analysis of the robot vacuum's leg extension and wheel drive characteristics at different step heights to calculate the robot vacuum's motion adaptation coefficient during operation includes:

[0051] The leg extension and retraction features and the wheel drive features are subjected to feature decomposition processing to obtain the leg extension and retraction decomposition features and the wheel drive decomposition features;

[0052] Based on the decomposed features of the leg extension and retraction, the amount of motion change of the sweeping robot at adjacent time points is calculated to obtain the amount of leg change;

[0053] Based on the wheel drive decomposition characteristics, the motion change of the sweeping robot at adjacent time points is calculated to obtain the wheel change.

[0054] The hardware information of the sweeping robot under the decomposition features of the leg extension and the wheel drive is queried, and the real-time working condition information of the sweeping robot is collected.

[0055] Combining the hardware information and the real-time operating condition information, the robot vacuum cleaner's corresponding motion execution efficiency is calculated under the leg extension decomposition feature and the wheel drive decomposition feature, thus obtaining the leg execution efficiency and the wheel execution efficiency.

[0056] Combining the leg execution efficiency, the wheel execution efficiency, the leg variation, and the wheel variation, the motion adaptation coefficient of the sweeping robot during operation is calculated using the following formula:

[0057]

[0058] Where μ represents the motion adaptation coefficient of the robot vacuum cleaner during operation, and γ d ΔG represents the motion coordination weight of the leg extension / retraction decomposition feature and the wheel drive decomposition feature in the d-th dimension. d ΔH represents the change in leg length and width in the d-th dimension of the leg extension and contraction decomposition feature. d η represents the wheel-related changes in the d-th dimension of the wheel drive decomposition features. leg (d) represents the leg execution efficiency of the leg extension / retraction decomposition feature in the d-th dimension, η wheel (d) represents the wheel execution efficiency of the wheel drive decomposition feature in the d-th dimension, δ d This represents the motion baseline weight of the leg extension / retraction decomposition feature and the wheel drive decomposition feature in the d-th dimension, where d represents the feature dimension of the leg extension / retraction decomposition feature and the wheel drive decomposition feature, and m represents the number of dimensions of the leg extension / retraction decomposition feature and the wheel drive decomposition feature.

[0059] Optionally, the step of analyzing the motion adaptability of the robot vacuum cleaner's legs and wheels based on the terrain physical properties includes:

[0060] Extract key terrain parameters from the terrain physical attributes and determine the surface action characteristics corresponding to the terrain physical attributes;

[0061] Based on the key terrain parameters and the surface action characteristics, the motion requirement patterns of the robot's legs and wheels are constructed.

[0062] Based on the aforementioned motion requirement pattern, calculate the theoretical ratio between the leg extension / retraction and the wheel drive force;

[0063] Record the leg wheel coordination data during the actual operation of the sweeping robot, and calculate the motion coordination deviation value of the sweeping robot by combining the leg wheel coordination data with the theoretical coordination ratio value;

[0064] Based on the aforementioned motion coordination deviation value, the motion adaptability of the robot vacuum cleaner's legs and wheels is analyzed.

[0065] Optionally, the step of controlling the robot vacuum cleaner to climb stairs based on the robot's horizontal stability, the motion adaptation coefficient, and the motion adaptability to complete the stair-climbing operation further includes:

[0066] Measure the total mass, horizontal distance between the legs, and pitch inertia of the robotic vacuum cleaner.

[0067] Based on the horizontal stability of the machine body, the vertical acceleration and pitch acceleration corresponding to the sweeping robot are determined;

[0068] Based on the robot's total mass, the horizontal distance between the leg centers, the pitch inertia, the vertical acceleration, and the pitch acceleration, the resultant forces on the legs and wheels of the sweeping robot are calculated using the following dynamic equilibrium formula:

[0069]

[0070] Among them, F leg F represents the resultant force of the legs. wheel M represents the resultant force on the wheels, and M represents the total mass of the robot. Indicates vertical acceleration. Pitch acceleration, l represents the horizontal distance between the leg and the center of gravity, g represents the acceleration due to gravity, I θ Indicates pitch angle inertia;

[0071] By combining the combined force of the legs and the combined force of the wheels, the robot vacuum cleaner performs the stair-climbing motion control to complete the stair-climbing operation.

[0072] Compared to the problems described in the background technology, this invention, by analyzing the segmental elevation differences in the real-time detection data, can distinguish interference factors such as stair steps and ground protrusions, providing a core basis for determining the stair-climbing trigger conditions. This significantly reduces the probability of false or missed triggers, ensuring efficient robot operation in complex terrain. Based on the stair-climbing trigger conditions, this invention analyzes the correlation between the robot's body pitch angle and leg extension / retraction on balance, clarifying their influence mechanism on the balance state. This facilitates subsequent calculations of the robot's horizontal stability. Furthermore, based on historical stair-climbing elevation records, this invention analyzes the robot's leg extension / retraction characteristics and wheel drive characteristics at different step heights, uncovering the adaptation patterns between movement and terrain in historical data. This provides a data foundation for subsequent calculations of movement adaptation coefficients, enabling timely identification of different step heights. Furthermore, this invention analyzes the terrain physical attributes of the current stair environment of the robotic vacuum cleaner. Based on these attributes, it analyzes the motion adaptability of the robot's legs and wheels, clarifying the impact of terrain on the leg and wheel motion coordination. This provides a basis for subsequent control of the robot's stair-climbing motion. Based on the robot's horizontal stability, motion adaptation coefficient, and motion adaptability, this invention controls the robot's stair-climbing motion to complete the stair-climbing operation. It comprehensively considers the robot's balance, the degree of motion matching with the terrain, and the adaptability of component coordination during the stair-climbing process, allowing for adjustments to the motion output and effectively improving the stability and smoothness of the stair-climbing process. This provides a strong guarantee for the robotic vacuum cleaner to efficiently complete stair-climbing operations, ensuring reliable operation even in complex stair environments. Therefore, the motion control method for robotic vacuum cleaners climbing stairs provided in this invention can improve the accuracy of motion control for robotic vacuum cleaners climbing stairs. Attached Figure Description

[0073] Figure 1 This is a flowchart illustrating a motion control method for a sweeping robot climbing stairs, provided in an embodiment of the present invention.

[0074] Figure 2 This is a schematic diagram of the terrain physical attribute analysis and processing flow in a motion control method for a sweeping robot going up stairs, provided by the present invention.

[0075] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0076] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0077] This application provides a motion control method for a robotic vacuum cleaner to climb stairs. The execution entity of this motion control method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the motion control method for a robotic vacuum cleaner to climb stairs 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.

[0078] Reference Figure 1 The diagram shown is a flowchart illustrating a motion control method for a robotic vacuum cleaner to climb stairs, according to an embodiment of the present invention. In this embodiment, the motion control method for a robotic vacuum cleaner to climb stairs includes:

[0079] S1. Collect elevation sequence data of the terrain in front of the sweeping robot. The elevation sequence data includes real-time detection data and historical stair climbing elevation records. Analyze the segment elevation differences in the real-time detection data to determine the stair climbing trigger conditions of the sweeping robot.

[0080] This invention analyzes the segmental elevation differences in the real-time detection data to distinguish between interfering factors such as stair steps and ground protrusions, providing a core basis for determining the stair-climbing trigger conditions. This significantly reduces the probability of false or missed triggers, ensuring efficient robot operation in complex terrain. The elevation sequence data is a continuous set of data reflecting changes in the terrain elevation in front of the robot, serving as fundamental information for terrain judgment, such as the elevation values ​​from LiDAR scanning and ultrasonic ranging. The real-time detection data is the terrain elevation information currently detected by the robot, such as ground elevation changes within 1 meter of the current location. The historical stair-climbing elevation record is the terrain elevation change data stored during the robot's past stair-climbing processes, such as the sequence of step heights recorded when climbing stairs in the past. The segmental elevation difference is the height difference between adjacent terrain segments in the real-time detection data, a key feature for determining whether it is a staircase. For example, the height difference between two consecutive detection segments is 12 centimeters. Optionally, the elevation sequence data of the terrain in front of the robot can be collected by a front-mounted depth sensing sensor (such as a TOF or RGB-D camera).

[0081] As an embodiment of the present invention, the analysis of the segment elevation difference in the real-time detection data includes:

[0082] The elevation values ​​in the real-time detection data are sorted to obtain an ordered elevation sequence;

[0083] The ordered elevation sequence is subjected to adjacent value pairing processing to obtain elevation value pairs;

[0084] Calculate the difference between each pair of elevation values ​​to obtain the original elevation difference;

[0085] The original elevation difference is processed to be undirected, resulting in an undirected elevation difference;

[0086] Based on the non-directional elevation difference, the segment elevation difference in the real-time detection data is obtained.

[0087] The ordered elevation sequence refers to the sequence formed by arranging the elevation values ​​in the real-time detection data according to the detection order. It is the basis for subsequent data processing. For example, a sequence of elevation values ​​such as "80cm, 82cm, 95cm, 96cm" arranged in chronological order. The elevation value pair is a pair of data obtained by combining two adjacent values ​​in the ordered elevation sequence. It is the direct basis for calculating the elevation difference. For example, "(80cm, 82cm), (82cm, 95cm)". The original elevation difference is the result obtained by calculating the difference between two values ​​in the elevation value pair. It reflects the height change of adjacent detection points. For example, "82cm-80cm=2cm", "95cm-82cm=13cm". The non-directional elevation difference is the data that only reflects the magnitude of the height difference after the original elevation difference is processed to be non-directional, eliminating the influence of positive and negative directions. For example, the original elevation difference "-5cm" is processed to become "5cm".

[0088] Optionally, the elevation values ​​in the real-time detection data can be sorted using a sequential arrangement algorithm to obtain an ordered elevation sequence. For example, the `sort` function in C++ can be used to sort the elevation values ​​according to the detection timestamp, resulting in ordered elevation sequences such as "75cm, 78cm, 89cm, 90cm" arranged chronologically. Alternatively, adjacent values ​​in the ordered elevation sequence can be paired using an adjacent combination algorithm to obtain elevation value pairs. For example, the `zip` function in Python can be used to combine the ordered elevation sequence with a sequence offset by one position, resulting in elevation value pairs such as "(75cm, 78cm), (78cm, 89cm)". Finally, a difference operation algorithm can be used to calculate each value in the elevation value pair. The original elevation difference can be obtained by comparing the differences between the two values. For example, using Java's arithmetic operators, the difference between the preceding and following terms can be calculated for each elevation value pair to obtain original elevation differences such as "3cm" and "11cm". The original elevation difference can be processed into an undirected elevation difference by using an absolute value conversion algorithm to obtain an undirected elevation difference. For example, the JavaScript method Math.abs() can be used to convert the original elevation difference "-4cm" into an undirected elevation difference of "4cm". The segment elevation difference in the real-time detection data can be obtained based on the undirected elevation difference using a data filtering algorithm. For example, using Python's list comprehension, data that meets the segment division criteria can be filtered from the undirected elevation differences to obtain segment elevation differences that reflect the height differences of different segments.

[0089] This invention analyzes the elevation differences in the real-time detection data to determine the stair-climbing trigger conditions of the sweeping robot, effectively avoiding misjudging ground protrusions and other interference factors as stairs, significantly improving the accuracy of stair-climbing trigger judgment, and ensuring that the robot can flexibly navigate complex terrain and efficiently complete cleaning tasks. The stair-climbing trigger condition is a command signal that starts the stair-climbing function when all verifications related to the elevation difference are met.

[0090] As an embodiment of the present invention, the step of analyzing the segment elevation difference in the real-time detection data to determine the stair-climbing trigger condition of the sweeping robot includes:

[0091] Analyze the detection segments corresponding to the elevation differences in the sections, and based on the detection segments, perform clustering processing on the elevation differences in the sections to obtain the clustered elevation differences in the sections;

[0092] Continuity detection is performed on the distribution segments corresponding to the elevation differences of the clustered segments to obtain continuous segment groups;

[0093] The number of segments in the continuous segment group is counted to obtain the number of valid segments;

[0094] The number of valid segments is compared with the preset minimum number of consecutive segments to obtain the result that the quantity meets the standard;

[0095] When the quantity meets the requirements, the flatness of the overall distribution of the continuous segment group is verified to obtain the terrain verification result.

[0096] If the flatness verification result is consistent with the staircase terrain characteristics, then the stair-climbing trigger condition is generated.

[0097] The detection segment refers to the terrain unit divided by spatial intervals in real-time detection data, which is the basic range for elevation difference analysis. For example, each 0.2-meter length is divided into a detection segment. The clustered segment elevation difference is a set of elevation differences with similar characteristics formed after clustering processing, such as a group of elevation difference data concentrated in the range of 12-14 cm. The distribution segment is the spatial distribution range of the clustered segment elevation differences, reflecting the continuity of the location of the elevation differences, such as a continuous area from 0.5 meters to 1.5 meters in front of the robot. The continuous segment group is a set of continuously distributed segments confirmed after continuous detection, such as five consecutive detection segments belonging to the same cluster of elevation differences. The number of effective segments is the number of detection segments included in the continuous segment group. The total number of test sections reflects the scale of the terrain features, such as the effective number consisting of 8 consecutive test sections; the minimum number of consecutive sections is a pre-set standard for determining whether it is a staircase, such as a minimum of 4 consecutive sections; the quantity compliance result is the judgment conclusion after comparing the effective number of sections with the minimum number of consecutive sections, which is divided into "meets the requirements" or "does not meet the requirements"; the terrain verification result is the terrain feature judgment conclusion obtained after flatness verification, such as "meets the staircase terrain features" or "does not meet the staircase terrain features"; the staircase terrain features are the unique height difference distribution attributes of the staircase, which are manifested as continuous and uniform height differences, such as the height difference of multiple consecutive sections being stable at 12-14cm.

[0098] Optionally, the analysis of the detection segments corresponding to the elevation differences, and the clustering of the elevation differences based on the detection segments, can be implemented using a clustering algorithm, such as the K-means algorithm from Python's scikit-learn library, setting the number of clusters to 2 (stairs and non-stairs), grouping the elevation differences into groups, and finally obtaining the clustered elevation differences; the continuity detection of the distribution segments corresponding to the clustered elevation differences can be implemented using continuity analysis techniques, such as using Java's array traversal method to determine whether the distribution positions of the clustered elevation differences are continuous, counting the number of consecutively occurring segments, and finally obtaining continuous segment groups; the counting of the number of segments in the continuous segment groups can be implemented using counting techniques, such as using JavaScript's array length property to count the number of segments contained in the continuous segment group. The final number of valid segments is obtained. Comparing the number of valid segments with the preset minimum number of consecutive segments can be achieved using comparison techniques, such as using C# conditional statements to compare the number of valid segments with the minimum number of consecutive segments and output the result indicating that the number meets the requirements. When the result indicates that the number meets the requirements, the flatness verification of the overall distribution of the consecutive segment group can be achieved using a flatness algorithm, such as using Python's NumPy library to calculate the standard deviation of the elevation difference within the consecutive segment group. If the standard deviation is less than a preset value, the flatness is determined, and the terrain verification result is obtained. If the flatness verification result conforms to the characteristics of staircase terrain, the generation of stair-climbing trigger conditions can be achieved using instruction generation techniques, such as using C++ function calls. When the verification is successful, a start signal is sent to the main control robot control system module, ultimately generating the stair-climbing trigger conditions.

[0099] S2. Based on the stair-climbing trigger condition, analyze the correlation factors between the robot's body pitch angle and leg extension amount and balance, so as to calculate the robot's body horizontal stability.

[0100] This invention analyzes the correlation between the robot vacuum cleaner's pitch angle and leg extension / retraction amount and its balance based on the stair-climbing trigger condition. This clarifies the influence mechanism of these two factors on the balance state, facilitating subsequent calculations of the robot vacuum cleaner's horizontal stability. The pitch angle is the tilt angle of the robot vacuum cleaner's body around its horizontal axis, reflecting its pitch posture. The leg extension / retraction amount is the extension length of the robot vacuum cleaner's leg actuators, directly affecting the body's support height and balance adjustment capability. The balance correlation factors are the intrinsic relationship between the pitch angle and leg extension / retraction amount, which interact to maintain the robot's balance. The horizontal stability is a quantitative indicator measuring the robot vacuum cleaner's overall stability in the vertical and pitch directions; a lower value indicates stronger stability.

[0101] As an embodiment of the present invention, the step of analyzing the correlation factors of the robot vacuum cleaner's body pitch angle and leg extension / retraction on balance based on the stair-climbing trigger condition, in order to calculate the robot vacuum cleaner's body horizontal stability, includes:

[0102] When the stair-climbing trigger condition is met, the motion characteristic data of the sweeping robot is acquired and captured in real time to obtain the raw motion data.

[0103] The original motion parameters are subjected to noise filtering to obtain the target motion parameters;

[0104] The pitch angle and leg extension / retraction in the target motion parameters are correlated and decomposed to obtain the balancing correlation factors;

[0105] Based on the aforementioned balance-related factors, a balance determination model for horizontal posture is constructed.

[0106] Using the aforementioned balance determination model, the horizontal stability of the sweeping robot is calculated.

[0107] The motion characteristic data refers to the real-time collected information on the robot's pitch angle and leg extension when triggered by climbing stairs, including dynamic data such as angle changes and extension length. The original motion parameters are the initial records obtained after capturing the motion characteristic data, which may contain interference information such as sensor noise. The target motion parameters are the data that retains the effective features after filtering out noise from the original motion parameters. The balance correlation factors refer to the intrinsic relationship between the robot's pitch angle and leg extension, which affects the balance state. For example, if the pitch angle is too large, the leg extension needs to be adjusted accordingly to maintain balance. The balance judgment model is a mathematical model that integrates balance correlation factors and is used to quantitatively analyze the horizontal stability of the robot. The robot's horizontal stability is a quantitative index calculated by the model that reflects the comprehensive stability of the robot in the vertical and pitch directions. The lower the value, the stronger the stability.

[0108] Furthermore, the motion characteristic data of the robotic vacuum cleaner can be captured in real time using multi-sensor fusion technology to obtain raw motion parameters. For example, an IMU (Inertial Measurement Unit) and a displacement sensor can be used to simultaneously collect pitch angle and leg extension. The raw motion parameters can be filtered for noise using a Kalman filter algorithm to obtain target motion parameters. For example, a state-space model can be constructed to separate effective signals from noise interference. The pitch angle and leg extension in the target motion parameters can be correlated and decomposed using a multiple linear regression method to obtain balance correlation factors. For example, the correlation coefficient matrix between the two can be calculated to determine the influence weights. A balance determination model for horizontal posture can be constructed based on the balance correlation factors using a neural network algorithm. For example, an LSTM network can be used to learn the feature patterns of historical balance data. The mean square error can be calculated using the balance determination model to determine the horizontal stability of the robotic vacuum cleaner. For example, the deviation between the model prediction and the actual stability threshold can be compared.

[0109] Furthermore, as an optional embodiment of the present invention, the step of calculating the horizontal stability of the robot vacuum cleaner using the balance determination model includes:

[0110] Extract the pitch angle deviation term and leg extension deviation term from the balance determination model;

[0111] Time series sampling was performed on the pitch angle deviation term and the leg extension deviation term to obtain multi-time deviation data;

[0112] Based on the multi-time deviation data and preset weighting coefficients, a steady-state calculation equation for the sweeping robot is constructed.

[0113] Based on the stability calculation equation, the body level stability of the sweeping robot is calculated.

[0114] The pitch angle deviation term refers to the difference between the actual pitch angle and the target pitch angle (e.g., 0° in a horizontal state); the leg extension deviation term refers to the difference between the actual leg extension and the target extension (e.g., the reference length for horizontal support); the multi-time deviation data is the collection and recording of pitch angle deviation and leg extension deviation at different time points; the weighting coefficient is a value set according to the robot's structural characteristics, reflecting the degree of influence of pitch angle and leg extension on balance, for example, the weighting coefficient of pitch angle is set to 0.6, and the weighting coefficient of leg extension is set to 0.4.

[0115] Furthermore, the pitch angle deviation and leg extension deviation terms in the balance determination model can be extracted through model parameter analysis, for example, by separating the corresponding deviation calculation nodes from the output layer of the neural network model; the pitch angle deviation and leg extension deviation terms can be sampled in time series using timestamps to obtain multi-time deviation data, for example, by collecting deviation data for 1 second at 10ms intervals; based on the multi-time deviation data, a steady-state calculation equation can be constructed using the analytic hierarchy process combined with preset weight coefficients, for example, by determining the weight ratio of pitch angle and leg extension through expert scoring.

[0116] Furthermore, as another embodiment, the horizontal stability of the robot vacuum cleaner is calculated using the following formula, based on the stability calculation equation:

[0117]

[0118] Where A represents the robot vacuum's horizontal stability, B a Let B0 represent the pitch angle at time a in the multi-time deviation data, and α represent the target pitch angle. B D represents the weighting coefficient for the pitch angle. a This represents the leg extension / retraction amount at time a in the multi-time deviation data, where D0 represents the target extension / retraction amount, and k D The weighting coefficient represents the leg extension / retraction amount, where a represents the sampling time and n represents the number of sampling times.

[0119] It should be noted that this formula considers the horizontal stability of the robot as the weighted average of the pitch angle deviation and the leg extension / retraction deviation at various moments. The weighting coefficient reflects the degree of influence of each on the balance. For example, the larger the pitch angle deviation or the larger the leg extension / retraction deviation, the higher the stability value and the worse the stability. In practical applications, such as when a robot vacuum cleaner is climbing stairs, if the pitch angle deviates from the target value by 1° (α) at ​​a certain moment... B =0.6), the leg extension / extension deviates from the target value by 2cm (k D =0.4), then the steady-state contribution at that moment is 1×0.6+2×0.4=1.4. After sampling 100 times, the average is taken to obtain the overall steady-state. By adjusting the weight coefficient (e.g., setting α according to the robot structure), the steady-state is obtained. B =0.7, k D =0.3), which can be adapted to the balance characteristics of different models, reducing the steady state calculation error to ±0.05, effectively identifying subtle unstable trends, and providing data support for subsequent stair climbing action control.

[0120] S3. Based on the historical stair-climbing elevation records, analyze the leg extension and wheel drive characteristics of the sweeping robot at different step heights to calculate the motion adaptation coefficient of the sweeping robot during operation.

[0121] This invention analyzes the leg extension and wheel drive characteristics of the sweeping robot at different step heights based on the historical stair climbing elevation records. It can uncover the adaptation patterns between the movements and the terrain in the historical data, providing a data foundation for subsequent calculation of the movement adaptation coefficient and timely identifying the optimal movement mode at different step heights.

[0122] The historical stair-climbing elevation record is the elevation data stored by the robot vacuum cleaner during its past stair-climbing process, including information such as step height, body posture, leg extension and retraction amount, and wheel driving force. The leg extension and retraction characteristics are the variation law of leg extension and retraction amount with step height, covering dynamic performance such as extension and retraction length and speed. The wheel driving characteristics are the variation law of wheel driving force and rotation speed with step height.

[0123] As an embodiment of the present invention, the step of analyzing the leg extension and wheel drive characteristics of the sweeping robot at different step heights based on the historical stair-climbing elevation records includes:

[0124] The historical stair-climbing elevation records are grouped by height interval to obtain stair-climbing action segments within each interval;

[0125] The leg trajectory is segmented from the segment of climbing the stairs in the interval to obtain the leg extension and contraction sequence;

[0126] The extreme values ​​of the extension and contraction amounts in the leg extension and contraction sequence are statistically analyzed to obtain the leg extension length characteristics;

[0127] The wheel rotation speed is stripped from the segment of the stair-climbing motion in the section to obtain the wheel drive curve;

[0128] The power consumption characteristics of the wheel are obtained by performing power consumption integration on the wheel drive curve.

[0129] By combining the leg extension length characteristics and the wheel power consumption characteristics, the leg extension and wheel drive characteristics of the sweeping robot at different step heights are generated.

[0130] The defined data includes: the stair-climbing motion segment, which is a complete record of the robot's leg and wheel movements within a height range; the leg extension sequence, which is a time-series data chain extracted from the motion segment reflecting the change in leg extension length over time; the leg extension length feature, which is a statistical set of the maximum extension, minimum extension, and the difference between them in the leg extension sequence, used to characterize the range of leg movements; the wheel drive curve, which is a curve extracted from the motion segment showing the change in wheel speed over time; and the wheel power consumption feature, which is the total energy consumption obtained by integrating the wheel drive curve over time, used to reflect the energy demand of the wheel drive.

[0131] Furthermore, height intervals can be grouped using an equidistant division method, for example, dividing intervals by a height difference of 1.5cm to ensure that each interval contains no fewer than 5 sets of valid stair-climbing records; leg trajectory segmentation of stair-climbing motion segments within intervals can be performed using a sliding window method, for example, extracting leg extension and retraction data with a window size of 0.2 seconds to form a leg extension and retraction sequence; extreme value filtering can be used to statistically analyze the leg extension and retraction sequence, for example, selecting the maximum and minimum values ​​in the sequence and calculating the difference between them as the core indicator of leg extension length; wheel speed separation can be performed on stair-climbing motion segments within intervals using signal filtering, for example, using low-pass filtering to remove noise and retaining the effective change curve of wheel speed; and feature matrix splicing can be used to combine leg extension length features and wheel power consumption features, for example, constructing a matrix with height intervals as rows and feature indicators as columns to form feature sets corresponding to different step heights.

[0132] Furthermore, as an optional embodiment of the present invention, the step of performing power consumption integration on the wheel drive curve to obtain wheel power consumption characteristics includes:

[0133] Voltage and current components are separated from the wheel drive curve to obtain a set of electrical power elements;

[0134] The set of electrical power elements is divided into action state units to obtain a power consumption calculation unit;

[0135] The power consumption density of the power consumption calculation unit is calculated to obtain the state power consumption spectrum;

[0136] By performing rotational speed-power coupling correlation on the state power consumption spectrum, the driving energy efficiency characteristics are obtained;

[0137] The drive energy efficiency characteristics are calibrated and fused within a high-range interval to generate wheel power consumption characteristics.

[0138] The electrical power element set is a collection of voltage and current time series sequences separated from the wheel drive curve and aligned with the time axis, used to provide basic electrical parameters for power consumption calculation. The motion state unit segmentation is a process of dividing the electrical power element set into different motion stages based on the acceleration changes of the wheel rotation speed (e.g., the critical point between acceleration and constant speed, and constant speed and deceleration). The power consumption calculation unit is a structured data unit containing the time interval, voltage sequence, and current sequence of a specific motion stage (acceleration / constant speed / deceleration). The state power consumption spectrum is a collection of unit-time power consumption values ​​(reflecting energy intensity) and power consumption fluctuation coefficients (reflecting energy stability) calculated by each power consumption calculation unit. The rotation speed power consumption coupling correlation is a process of proportionally calculating the average rotation speed and corresponding power consumption value of each motion stage to quantify the unit rotation speed energy consumption efficiency. The drive energy efficiency characteristics are a feature set containing the unit rotation speed energy consumption coefficient of each motion stage. The height interval calibration fusion is a process of binding the drive energy efficiency characteristics with the corresponding step height interval and integrating them into a cross-interval energy efficiency law.

[0139] Furthermore, the wheel drive curve can be separated into voltage and current components using a filtering decoupling method. For example, a bandpass filter can be used to extract the voltage signal from 10-50Hz and the current signal from 5-30Hz, forming a time-aligned set of electrical power elements. The action state unit can be segmented using an acceleration threshold method, for example, setting the absolute value of the rotational speed acceleration to ≤0.5 rad / s. 2 To determine the constant velocity critical point, the electrical power element set is divided into three units: acceleration (acceleration > 0.5), constant velocity (acceleration between -0.5 and 0.5), and deceleration (acceleration < -0.5), thus obtaining the power consumption calculation unit. The unit power consumption density can be calculated using the sliding window method. For example, the sum of the products of voltage and current is calculated with a window of 0.1 seconds, and then divided by the window duration to obtain the power consumption per unit time. The fluctuation coefficient is obtained by calculating the ratio of the standard deviation to the mean of the power consumption within the window, forming the state power consumption spectrum. The speed-power consumption coupling relationship can be calculated by linear fitting. For example, a linear regression is performed on the mean speed and mean power consumption of each unit, and the slope is taken as the energy consumption coefficient per unit speed (e.g., the slope of the acceleration unit is 0.8 W·s / rad), thus obtaining the drive energy efficiency characteristics. Height interval calibration and fusion can be performed by interval index matching. For example, each step height interval (e.g., 5-8cm, 8-11cm) is used as an index, and the energy efficiency coefficient of each unit within the corresponding interval is bound. After calculating the interval mean, a "height-energy consumption" correspondence table is formed, generating the wheel power consumption characteristics.

[0140] This invention analyzes the leg extension and wheel drive characteristics of the robotic vacuum cleaner at different step heights to calculate the motion adaptation coefficient of the robotic vacuum cleaner during operation. This allows for matching the leg and wheel motion combinations at different step heights, reducing motion stuttering or slipping during stair climbing, and significantly improving the robotic vacuum cleaner's adaptability and operational stability in the face of diverse stair terrains. The motion adaptation coefficient is a quantitative indicator that measures the appropriate matching of the leg extension and wheel drive motion combinations to the corresponding step heights; a higher value indicates a better degree of matching between the motion and the terrain.

[0141] As an embodiment of the present invention, the analysis of the leg extension and retraction characteristics and wheel drive characteristics of the sweeping robot at different step heights, in order to calculate the motion adaptation coefficient of the sweeping robot during operation, includes:

[0142] The leg extension and retraction features and the wheel drive features are subjected to feature decomposition processing to obtain the leg extension and retraction decomposition features and the wheel drive decomposition features;

[0143] Based on the decomposed features of the leg extension and retraction, the amount of motion change of the sweeping robot at adjacent time points is calculated to obtain the amount of leg change;

[0144] Based on the wheel drive decomposition characteristics, the motion change of the sweeping robot at adjacent time points is calculated to obtain the wheel change.

[0145] The hardware information of the sweeping robot under the decomposition features of the leg extension and the wheel drive is queried, and the real-time working condition information of the sweeping robot is collected.

[0146] Combining the hardware information and the real-time operating condition information, the robot vacuum cleaner's corresponding motion execution efficiency is calculated under the leg extension decomposition feature and the wheel drive decomposition feature, thus obtaining the leg execution efficiency and the wheel execution efficiency.

[0147] By combining the leg execution efficiency, the wheel execution efficiency, the leg variation, and the wheel variation, the motion adaptation coefficient of the sweeping robot during operation is calculated.

[0148] The leg extension / retraction decomposition feature is a set of sub-features obtained by breaking down the leg extension / retraction feature according to motion dimensions (such as extension / retraction amplitude, extension / retraction frequency, and extension / retraction acceleration), used to finely describe the leg's motion characteristics under different motion parameter dimensions; the wheel drive decomposition feature is a set of sub-features obtained by breaking down the wheel drive feature according to drive parameters (such as rotational speed, torque, and steering angle), used to finely describe the wheel's drive characteristics under different power output dimensions; the leg change is the numerical difference between corresponding sub-features at adjacent moments in the leg extension / retraction decomposition feature, used to quantify the degree of change in leg movement during the dynamic process; the wheel change is the numerical difference between corresponding sub-features at adjacent moments in the wheel drive decomposition feature, used to quantify the degree of change in wheel drive during the dynamic process; The hardware information refers to the set of hardware parameters for the robot vacuum's legs and wheels, including inherent hardware parameters such as the rated power of the leg motors, the rated torque of the wheel motors, the transmission ratio of the leg extension mechanism, and the diameter of the wheel drive wheels. The real-time operating condition information refers to the operating status parameters collected in real time during the operation of the robot vacuum, including dynamic operating parameters such as the real-time battery voltage, real-time motor current, real-time body tilt angle, and real-time ground friction. The action execution efficiency is a quantitative evaluation value of the leg or wheel action execution effect, combining the hardware information and the real-time operating condition information, reflecting the comprehensive performance of energy conversion efficiency and action accuracy during action execution (leg execution efficiency reflects the energy utilization rate and extension accuracy of the leg extension action, while wheel execution efficiency reflects the power output efficiency and path tracking accuracy of the wheel drive action).

[0149] Furthermore, the leg extension and retraction features and wheel drive features can be decomposed using multidimensional parameter decomposition to obtain the decomposed features of leg extension and retraction and wheel drive. For example, the leg extension and retraction features can be decomposed into three sub-features: extension amplitude, extension frequency, and extension acceleration; the wheel drive features can be decomposed into three sub-features: rotational speed, torque, and steering angle. The change in motion can be calculated based on the decomposed features of leg extension and retraction and wheel drive using the difference between adjacent time points, resulting in the change in leg and wheel motion. For example, the difference between the leg extension and retraction amplitude at time t and time t-1 can be calculated as the amplitude change, and the difference between the wheel rotational speed at time t and time t-1 can be calculated as the rotational speed change. The hardware parameters under the decomposed features of leg extension and retraction and wheel drive can be obtained through hardware parameter database queries or embedded system parameter reading. Information, such as the rated power of the leg motor retrieved from the product design document, and the diameter of the wheel drive wheel read through the robot control system; real-time operating information of the sweeping robot can be collected through multi-sensor data fusion, such as using a voltage sensor to collect the real-time battery voltage, a current sensor to collect the real-time motor current, and a tilt sensor to collect the real-time tilt angle of the body; the motion execution efficiency can be calculated by combining the energy conversion efficiency and motion accuracy coupling formula with the hardware information and real-time operating information to obtain the leg execution efficiency and wheel execution efficiency, for example, leg execution efficiency = (theoretical extension range × real-time voltage) / (actual extension range × rated voltage) × (extension positioning error correction coefficient), wheel execution efficiency = (theoretical driving torque × real-time current) / (actual driving torque × rated current) × (path tracking error correction coefficient).

[0150] Furthermore, as another embodiment, by combining the leg execution efficiency, the wheel execution efficiency, the leg variation, and the wheel variation, the motion adaptation coefficient of the sweeping robot during operation is calculated, including:

[0151]

[0152] Where μ represents the motion adaptation coefficient of the robot vacuum cleaner during operation, and γ d ΔG represents the motion coordination weight of the leg extension / retraction decomposition feature and the wheel drive decomposition feature in the d-th dimension. d ΔH represents the change in leg length and width in the d-th dimension of the leg extension and contraction decomposition feature. d η represents the wheel-related changes in the d-th dimension of the wheel drive decomposition features. leg (d) represents the leg execution efficiency of the leg extension / retraction decomposition feature in the d-th dimension, η wheel (d) represents the wheel execution efficiency of the wheel drive decomposition feature in the d-th dimension, δ dThis represents the motion baseline weight of the leg extension / retraction decomposition feature and the wheel drive decomposition feature in the d-th dimension, where d represents the feature dimension of the leg extension / retraction decomposition feature and the wheel drive decomposition feature, and m represents the number of dimensions of the leg extension / retraction decomposition feature and the wheel drive decomposition feature.

[0153] It should be noted that this formula quantifies the robot vacuum's motion adaptability in complex terrain interaction scenarios (such as crossing thresholds, climbing slopes, turning into carpet seams, etc.) into a weighted coupling result of "coordination strength - execution efficiency" for each motion dimension (such as leg extension range, wheel drive speed, body posture angle, etc.). The weight coefficients reflect the differences in priority between leg and wheel actions under different terrains. For example, the stronger the terrain resistance, the higher the coordination weight of leg extension dynamically, or when the change in wheel drive fluctuates abnormally due to a sharp increase in ground resistance, the numerical deviation of the motion adaptability coefficient will be amplified simultaneously, intuitively exposing the matching defects between the robot's actions and the terrain.

[0154] S4. Analyze the terrain physical attributes of the current step environment of the sweeping robot, and based on the terrain physical attributes, analyze the motion adaptability of the sweeping robot's legs and wheels.

[0155] This invention analyzes the terrain physical attributes of the current stair environment of the robotic vacuum cleaner. Based on these attributes, it analyzes the motion adaptability of the robot's legs and wheels, clarifying the impact of terrain on the coordinated leg and wheel movements. This provides a basis for subsequent control of the robot's stair-climbing actions. The terrain physical attributes refer to the physical characteristics of the stair environment in which the robot is located, including stair height, surface material, and slope inclination. The motion adaptability refers to the suitability of the coordinated leg and wheel movements to the current stair environment terrain. Furthermore, the analysis of the terrain physical attributes of the current stair environment of the robotic vacuum cleaner... The analysis steps are as follows: First, raw data of the step environment is collected using a combination of multiple sensors (such as infrared ranging sensors and vision cameras) to obtain initial data containing information such as step height, surface texture, and slope angle. Then, the initial data is denoised to remove invalid data caused by sensor errors. Next, feature extraction is performed on the processed data to obtain effective information reflecting key terrain features. Then, through data quantization and transformation, the extracted feature information is converted into specific numerical indicators. Finally, these numerical indicators are integrated to complete a comprehensive description of the terrain's physical properties. For a more intuitive understanding of the analysis and processing steps of the robotic vacuum cleaner in dealing with step environments in this application, please refer to [reference needed]. Figure 2 The diagram shown is a schematic representation of the analysis and processing flow of a sweeping robot in response to a current stepped environment, as provided by this invention. It should be noted that in this invention... Figure 2The flowchart presented is only for the analysis and processing of the robot vacuum cleaner in the context of step environments, and is not limited to the analysis and processing of the robot vacuum cleaner in different actual application scenarios.

[0156] As an embodiment of the present invention, the step of analyzing the motion adaptability of the robot vacuum cleaner's legs and wheels based on the terrain physical properties includes:

[0157] Extract key terrain parameters from the terrain physical attributes and determine the surface action characteristics corresponding to the terrain physical attributes;

[0158] Based on the key terrain parameters and the surface action characteristics, the motion requirement patterns of the robot's legs and wheels are constructed.

[0159] Based on the aforementioned motion requirement pattern, calculate the theoretical ratio between the leg extension / retraction and the wheel drive force;

[0160] Record the leg wheel coordination data during the actual operation of the sweeping robot, and calculate the motion coordination deviation value of the sweeping robot by combining the leg wheel coordination data with the theoretical coordination ratio value;

[0161] Based on the aforementioned motion coordination deviation value, the motion adaptability of the robot vacuum cleaner's legs and wheels is analyzed.

[0162] The key terrain parameters are core physical quantities selected from the physical properties of the terrain that directly affect the coordination between the legs and wheels, such as step height, tread width, and surface friction coefficient. These serve as the basic inputs for analyzing motion adaptability. The surface action characteristics are the mechanical feedback characteristics of the terrain on the robot, including the distribution of support force, the magnitude of friction, and the component force generated by the slope. Examples include the low friction characteristics of a smooth surface and the concentrated support characteristics at the edge of a step. The motion requirement pattern is a framework of coordination rules that the legs and wheels should adopt, formed by combining the key terrain parameters and surface action characteristics. For example, "appropriate wheel driving force under high-friction terrain." The robot employs a coordinated mode of "reduced leg extension and retraction"; the theoretical coordination ratio is the optimal numerical ratio of leg extension and retraction to wheel driving force derived from the motion requirement mode. For example, when the step height is 15cm, the theoretical ratio is "8cm extension corresponds to 10N driving force"; the leg and wheel coordination data are real-time corresponding data of leg extension and retraction and wheel driving force recorded when the robot actually climbs stairs, including the dynamic change process; the motion coordination deviation value is the quantitative result of the difference between the actual coordination data and the theoretical coordination ratio, calculated using the relative error formula "(actual ratio - theoretical ratio) / theoretical ratio". The smaller the deviation value, the more accurate the coordination.

[0163] Furthermore, key terrain parameters can be extracted from the physical attributes of the terrain using multi-sensor fusion technology. For example, lidar can be used to acquire step height, visual sensors can identify tread width, and tactile sensors can measure friction coefficient. Data can be integrated and core parameters can be filtered using Python's Pandas library. The surface interaction characteristics corresponding to the physical attributes of the terrain can be determined through mechanical simulation analysis. For example, ANSYS software can be used to simulate the contact mechanics between the robot and the ground, calculating the distribution of support reaction force and friction force under different terrains. Combining the key terrain parameters and the surface interaction characteristics, motion demand patterns can be constructed through association rule mining. For example, the Apriori algorithm can be used to mine association rules such as "friction coefficient < 0.3 → wheel drive force increased by 20% + leg extension increased by 15%" from historical terrain-motion data. Based on the motion demand patterns, proportional derivation formulas can be used to calculate... The theoretical matching ratio is calculated, for example, based on the pattern that "the extension and contraction increases linearly with the step height, and the driving force changes exponentially with the friction coefficient" in the model, to derive a specific numerical ratio; the leg and wheel matching data of the sweeping robot in actual operation can be recorded through a real-time data acquisition module, for example, by deploying encoders to record the leg extension and contraction, and torque sensors to collect the wheel driving force, and storing the data synchronously at 10ms intervals; combining the leg and wheel matching data and the theoretical matching ratio, the action matching deviation value can be obtained through an error calculation model, for example, by using MATLAB's curve fitting tool to compare the actual and theoretical curves and calculating the root mean square error as the deviation value; based on the action matching deviation value, the action fitness can be analyzed through normalization processing, for example, by using the conversion formula "1 - absolute value of deviation value" to map the deviation value to a fitness index of 0-1, when the deviation value is 0, the fitness is 1 (perfect fit).

[0164] S5. Based on the horizontal stability of the robot body, the action adaptation coefficient, and the action adaptability, execute the stair-climbing action control of the sweeping robot to complete the stair-climbing operation. If the stair-climbing trigger condition is detected again, continue to execute the stair-climbing process; otherwise, exit the stair-climbing control.

[0165] This invention controls the stair-climbing motion of a robotic vacuum cleaner based on the robot's horizontal stability, motion adaptation coefficient, and motion adaptability. This allows for comprehensive consideration of the robot's balance, the degree of matching between its movements and the terrain, and the adaptability of its components during the stair-climbing process. This enables adjustments to the motion output, effectively improving the stability and smoothness of the stair-climbing process. Consequently, this provides a strong guarantee for the robotic vacuum cleaner to efficiently complete stair-climbing operations, ensuring its reliable operation even in complex stairwell environments.

[0166] As an embodiment of the present invention, the step of controlling the robot vacuum cleaner to climb stairs based on the horizontal stability of the robot body, the motion adaptation coefficient, and the motion adaptability to complete the stair climbing operation further includes:

[0167] Measure the total mass, horizontal distance between the legs, and pitch inertia of the robotic vacuum cleaner.

[0168] Based on the horizontal stability of the machine body, the vertical acceleration and pitch acceleration corresponding to the sweeping robot are determined;

[0169] Based on the robot's total mass, the horizontal distance between the leg centers, the pitch inertia, the vertical acceleration, and the pitch acceleration, the resultant forces on the legs and wheels of the sweeping robot are calculated using the following dynamic equilibrium formula:

[0170]

[0171] Among them, F leg F represents the resultant force of the legs. wheel M represents the resultant force on the wheels, and M represents the total mass of the robot. Indicates vertical acceleration. Pitch acceleration, l represents the horizontal distance between the leg and the center of gravity, g represents the acceleration due to gravity, I θ Indicates pitch angle inertia;

[0172] By combining the combined force of the legs and the combined force of the wheels, the robot vacuum cleaner performs the stair-climbing motion control to complete the stair-climbing operation.

[0173] The total mass of the robot refers to the overall mass of the sweeping robot, a fundamental physical quantity affecting vertical force balance, and is calibrated to 3.5 kg using a weight sensor. The horizontal distance between the leg centers is the horizontal distance from the rotation center of the leg joints to the robot's center of gravity, a key geometric parameter for pitch moment balance, and is measured to be 15 cm using structural design drawings. The pitch inertia is the moment of inertia of the robot body about the pitch axis, reflecting the inertial characteristics of the robot's pitch motion, and is measured to be 0.12 kg·m using dynamic experiments. 2 The vertical acceleration mentioned is the acceleration of the fuselage in the vertical direction, reflecting the change in the fuselage's motion state in the vertical direction. For example, it is detected as 0.5 m / s² by an IMU sensor. 2 The pitch acceleration is the angular acceleration of the fuselage about its lateral axis, reflecting the rate of change of the fuselage's pitch attitude; for example, it is detected by an IMU sensor as 0.2 rad / s. 2 The combined force of the legs is the resultant force of the support and driving force of the leg joints on the fuselage, which directly affects the balance in the vertical and pitch directions; the combined force of the wheels is the resultant force of the traction and support force of the wheels on the ground, which works in conjunction with the combined force of the legs to maintain the stability of the fuselage.

[0174] Furthermore, the total mass of the robotic vacuum cleaner can be measured using a high-precision weight sensor, such as a strain gauge weight sensor, by reading serial port data using Python's PySerial library to obtain a mass value accurate to 0.01 kg; the horizontal distance between the leg centers can be determined by laser ranging combined with structural design drawings, for example, by using a laser displacement sensor to measure the horizontal distance between the leg joints and the center of gravity marker, and correcting it with structural parameters from CAD drawings to obtain a distance value accurate to 0.1 cm; the pitch angle inertia can be determined through dynamic experiments and data fitting, for example, by applying a known torque to the robot's pitch axis. Angular acceleration data is collected, and the inertia value is obtained by fitting using the least squares method. Based on the horizontal stability of the robot body, the vertical acceleration and pitch acceleration are determined by an IMU sensor, such as an MPU6050 sensor. Data fusion is performed using the ROS imu_filter_madgwick package to output stable acceleration data. Combining the resultant force of the legs and the resultant force of the wheels, the robot's stair-climbing motion is controlled by the motor drive module. For example, the resultant force value is converted into current commands for the leg servo motors and wheel drive motors, and sent to the actuator via the CAN bus to achieve precise torque output.

[0175] It should be understood that if the climbing trigger condition is detected again, it means that there are still steps ahead that meet the climbing requirements, and the climbing action of the subsequent stairs needs to be completed. In this case, the climbing process will continue; otherwise, the climbing control will be exited.

[0176] Compared to the problems described in the background technology, this invention, by analyzing the segmental elevation differences in the real-time detection data, can distinguish interference factors such as stair steps and ground protrusions, providing a core basis for determining the stair-climbing trigger conditions. This significantly reduces the probability of false or missed triggers, ensuring efficient robot operation in complex terrain. Based on the stair-climbing trigger conditions, this invention analyzes the correlation between the robot's body pitch angle and leg extension / retraction on balance, clarifying their influence mechanism on the balance state. This facilitates subsequent calculations of the robot's horizontal stability. Furthermore, based on historical stair-climbing elevation records, this invention analyzes the robot's leg extension / retraction characteristics and wheel drive characteristics at different step heights, uncovering the adaptation patterns between movement and terrain in historical data. This provides a data foundation for subsequent calculations of movement adaptation coefficients, enabling timely identification of different step heights. Furthermore, this invention analyzes the terrain physical attributes of the current stair environment of the robotic vacuum cleaner. Based on these attributes, it analyzes the motion adaptability of the robot's legs and wheels, clarifying the impact of terrain on the leg and wheel motion coordination. This provides a basis for subsequent control of the robot's stair-climbing motion. Based on the robot's horizontal stability, motion adaptation coefficient, and motion adaptability, this invention controls the robot's stair-climbing motion to complete the stair-climbing operation. It comprehensively considers the robot's balance, the degree of motion matching with the terrain, and the adaptability of component coordination during the stair-climbing process, allowing for adjustments to the motion output and effectively improving the stability and smoothness of the stair-climbing process. This provides a strong guarantee for the robotic vacuum cleaner to efficiently complete stair-climbing operations, ensuring reliable operation even in complex stair environments. Therefore, the motion control method for robotic vacuum cleaners climbing stairs provided in this invention can improve the accuracy of motion control for robotic vacuum cleaners climbing stairs.

[0177] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0178] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A motion control method for a robotic vacuum cleaner climbing stairs, characterized in that, The method includes: The system collects elevation sequence data of the terrain in front of the robot vacuum cleaner, including real-time detection data and historical stair climbing elevation records. It analyzes the segment elevation differences in the real-time detection data to determine the stair climbing trigger conditions of the robot vacuum cleaner. Based on the aforementioned stair-climbing triggering conditions, the correlation factors between the robot's body pitch angle and leg extension amount and balance are analyzed to calculate the robot's body horizontal stability. Based on the historical stair-climbing elevation records, the leg extension and wheel drive characteristics of the sweeping robot at different step heights are analyzed to calculate the motion adaptation coefficient of the sweeping robot during operation. The terrain physical properties of the current step environment of the robot vacuum cleaner are analyzed, and the motion adaptability of the robot vacuum cleaner's legs and wheels is analyzed based on the terrain physical properties. Based on the robot's horizontal stability, the action adaptation coefficient, and the action adaptability, the robot vacuum cleaner performs stair climbing control to complete the stair climbing operation. If the stair climbing trigger condition is detected again, the stair climbing process continues; otherwise, the stair climbing control is exited.

2. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, The analysis of the segment elevation difference in the real-time detection data includes: The elevation values ​​in the real-time detection data are sorted to obtain an ordered elevation sequence; The ordered elevation sequence is subjected to adjacent value pairing processing to obtain elevation value pairs; Calculate the difference between each pair of elevation values ​​to obtain the original elevation difference; The original elevation difference is processed to be undirected, resulting in an undirected elevation difference; Based on the non-directional elevation difference, the segment elevation difference in the real-time detection data is obtained.

3. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, The analysis of the segment elevation differences in the real-time detection data to determine the stair-climbing trigger conditions of the sweeping robot includes: Analyze the detection segments corresponding to the elevation differences in the sections, and based on the detection segments, perform clustering processing on the elevation differences in the sections to obtain the clustered elevation differences in the sections; Continuity detection is performed on the distribution segments corresponding to the elevation differences of the clustered segments to obtain continuous segment groups; The number of segments in the continuous segment group is counted to obtain the number of valid segments; The number of valid segments is compared with the preset minimum number of consecutive segments to obtain the result that the quantity meets the standard; When the quantity meets the requirements, the flatness of the overall distribution of the continuous segment group is verified to obtain the terrain verification result. If the flatness verification result is consistent with the staircase terrain characteristics, then the stair-climbing trigger condition is generated.

4. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, Based on the stair-climbing trigger condition, the analysis of the correlation factors between the robot's body pitch angle and leg extension / retraction and balance, in order to calculate the robot's horizontal stability, includes: When the stair-climbing trigger condition is met, the motion characteristic data of the sweeping robot is acquired and captured in real time to obtain the raw motion data. The original motion parameters are subjected to noise filtering to obtain the target motion parameters; The pitch angle and leg extension / retraction in the target motion parameters are correlated and decomposed to obtain the balancing correlation factors; Based on the aforementioned balance-related factors, a balance determination model for horizontal posture is constructed. Using the aforementioned balance determination model, the horizontal stability of the sweeping robot is calculated.

5. The motion control method for a sweeping robot climbing stairs as described in claim 4, characterized in that, The step of calculating the horizontal stability of the sweeping robot using the aforementioned balance determination model includes: Extract the pitch angle deviation term and leg extension deviation term from the balance determination model; Time series sampling was performed on the pitch angle deviation term and the leg extension deviation term to obtain multi-time deviation data; Based on the multi-time deviation data and preset weighting coefficients, a steady-state calculation equation for the sweeping robot is constructed. Based on the stability calculation equation, the horizontal stability of the robot vacuum cleaner is calculated using the following formula: Where A represents the robot vacuum's horizontal stability, B a Let B0 represent the pitch angle at time a in the multi-time deviation data, and α represent the target pitch angle. B D represents the weighting coefficient for the pitch angle. a This represents the leg extension / retraction amount at time a in the multi-time deviation data, where D0 represents the target extension / retraction amount, and k D The weighting coefficient represents the leg extension / retraction amount, where a represents the sampling time and n represents the number of sampling times.

6. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, The analysis of the robot vacuum's leg extension and wheel drive characteristics at different step heights, based on the historical stair-climbing elevation records, includes: The historical stair-climbing elevation records are grouped by height interval to obtain stair-climbing action segments within each interval; The leg trajectory is segmented from the segment of climbing the stairs in the interval to obtain the leg extension and contraction sequence; The extreme values ​​of the extension and contraction amounts in the leg extension and contraction sequence are statistically analyzed to obtain the leg extension length characteristics; The wheel rotation speed is stripped from the segment of the stair-climbing motion in the section to obtain the wheel drive curve; The power consumption characteristics of the wheel are obtained by performing power consumption integration on the wheel drive curve. By combining the leg extension length characteristics and the wheel power consumption characteristics, the leg extension and wheel drive characteristics of the sweeping robot at different step heights are generated.

7. The motion control method for a sweeping robot climbing stairs as described in claim 6, characterized in that, The step of performing power consumption integration on the wheel drive curve to obtain wheel power consumption characteristics includes: Voltage and current components are separated from the wheel drive curve to obtain a set of electrical power elements; The set of electrical power elements is divided into action state units to obtain a power consumption calculation unit; The power consumption density of the power consumption calculation unit is calculated to obtain the state power consumption spectrum; By performing rotational speed-power coupling correlation on the state power consumption spectrum, the driving energy efficiency characteristics are obtained; The drive energy efficiency characteristics are calibrated and fused within a high-range interval to generate wheel power consumption characteristics.

8. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, The analysis of the robot vacuum's leg extension and wheel drive characteristics at different step heights is used to calculate the robot vacuum's motion adaptation coefficient during operation, including: The leg extension and retraction features and the wheel drive features are subjected to feature decomposition processing to obtain the leg extension and retraction decomposition features and the wheel drive decomposition features; Based on the decomposed features of the leg extension and retraction, the amount of motion change of the sweeping robot at adjacent time points is calculated to obtain the amount of leg change; Based on the wheel drive decomposition characteristics, the motion change of the sweeping robot at adjacent time points is calculated to obtain the wheel change. The hardware information of the sweeping robot under the decomposition features of the leg extension and the wheel drive is queried, and the real-time working condition information of the sweeping robot is collected. Combining the hardware information and the real-time operating condition information, the robot vacuum cleaner's corresponding motion execution efficiency is calculated under the leg extension decomposition feature and the wheel drive decomposition feature, thus obtaining the leg execution efficiency and the wheel execution efficiency. Combining the leg execution efficiency, the wheel execution efficiency, the leg variation, and the wheel variation, the motion adaptation coefficient of the sweeping robot during operation is calculated using the following formula: Where μ represents the motion adaptation coefficient of the robot vacuum cleaner during operation, and γ d ΔG represents the motion coordination weight of the leg extension / retraction decomposition feature and the wheel drive decomposition feature in the d-th dimension. d ΔH represents the change in leg length and width in the d-th dimension of the leg extension and contraction decomposition feature. d η represents the wheel-related changes in the d-th dimension of the wheel drive decomposition features. leg (d) represents the leg execution efficiency of the leg extension / retraction decomposition feature in the d-th dimension, η wheel (d) represents the wheel execution efficiency of the wheel drive decomposition feature in the d-th dimension, δ d This represents the motion baseline weight of the leg extension / retraction decomposition feature and the wheel drive decomposition feature in the d-th dimension, where d represents the feature dimension of the leg extension / retraction decomposition feature and the wheel drive decomposition feature, and m represents the number of dimensions of the leg extension / retraction decomposition feature and the wheel drive decomposition feature.

9. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, The analysis of the motion adaptability of the robot vacuum cleaner's legs and wheels based on the terrain physical properties includes: Extract key terrain parameters from the terrain physical attributes and determine the surface action characteristics corresponding to the terrain physical attributes; Based on the key terrain parameters and the surface action characteristics, the motion requirement patterns of the robot's legs and wheels are constructed. Based on the aforementioned motion requirement pattern, calculate the theoretical ratio between the leg extension / retraction and the wheel drive force; Record the leg wheel coordination data during the actual operation of the sweeping robot, and calculate the motion coordination deviation value of the sweeping robot by combining the leg wheel coordination data with the theoretical coordination ratio value; Based on the aforementioned motion coordination deviation value, the motion adaptability of the robot vacuum cleaner's legs and wheels is analyzed.

10. The motion control method for a sweeping robot climbing stairs as described in claim 1, characterized in that, The method of controlling the robot vacuum cleaner to climb stairs based on the horizontal stability of the machine body, the action adaptation coefficient, and the action adaptability to complete the stair climbing operation also includes: Measure the total mass, horizontal distance between the legs, and pitch inertia of the robotic vacuum cleaner. Based on the horizontal stability of the machine body, the vertical acceleration and pitch acceleration corresponding to the sweeping robot are determined; Based on the robot's total mass, the horizontal distance between the leg centers, the pitch inertia, the vertical acceleration, and the pitch acceleration, the resultant forces on the legs and wheels of the sweeping robot are calculated using the following dynamic equilibrium formula: Among them, F leg F represents the resultant force of the legs. wheel M represents the resultant force on the wheels, and M represents the total mass of the robot. Indicates vertical acceleration. Pitch acceleration, l represents the horizontal distance between the leg and the center of gravity, g represents the acceleration due to gravity, I θ Indicates pitch angle inertia; By combining the combined force of the legs and the combined force of the wheels, the robot vacuum cleaner performs the stair-climbing motion control to complete the stair-climbing operation.

Images