Quality Detection Method and Equipment for Data Acquisition of Humanoid Robots

Abstract

An embodiment of the present application provides a quality inspection method and device for data collection of a humanoid robot. The method includes: obtaining the collected data of the humanoid robot, where the collected data includes an expected value sequence and an actual value sequence of the target parameters of the humanoid robot, determining a latency evaluation value and a continuity evaluation value of the collected data according to the expected value sequence and the actual value sequence, and determining the quality inspection result of the collected data according to the latency evaluation value and the continuity evaluation value. The quality inspection method for data collection of a humanoid robot provided by the embodiment of the present application analyzes two time series, namely the expected value sequence and the actual value sequence, quantifies the latency and continuity errors of the two sequences, obtains the latency evaluation value and the continuity evaluation value, and then determines whether the collected data is qualified based on the evaluation value, which can realize automatic quality inspection, replace manual visual inspection, reduce costs, and improve the detection efficiency.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of humanoid robots, and in particular, to a quality detection method and device for data collection of humanoid robots. Background Art

[0002] Quality inspection of the collected data of humanoid robots is a key link to ensure the accuracy, integrity and consistency of the data, which can avoid inaccurate robot control or function failure caused by data deviation; high-quality quality inspection data provides reliable input for algorithm training and system optimization, ensuring stable, safe and efficient human-computer interaction of humanoid robots in complex scenarios.

[0003] In the related art, manual quality inspection is usually used to inspect the collected data of humanoid robots. However, the manual quality inspection method has high cost and poor efficiency. Summary of the Invention

[0004] The embodiments of the present application provide a quality detection method and device for data collection of humanoid robots to reduce costs and improve efficiency.

[0005] In a first aspect, the embodiments of the present application provide a quality detection method for data collection of humanoid robots, including:

[0006] Obtaining the collected data of the humanoid robot; the collected data includes an expected value sequence and an actual value sequence of target parameters of the humanoid robot;

[0007] Determining a latency evaluation value and a continuity evaluation value of the collected data according to the expected value sequence and the actual value sequence;

[0008] Determining a quality inspection result of the collected data according to the latency evaluation value and the continuity evaluation value.

[0009] In a possible design, the determining the latency evaluation value of the collected data according to the expected value sequence and the actual value sequence includes:

[0010] Determining an error value and a similarity value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence;

[0011] Determining the latency evaluation value of the collected data according to the error value and the similarity value between the expected value sequence and the actual value sequence.

[0012] In a possible design, the determining the error value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence includes:

[0013] Determine multiple task phases according to the task characteristics corresponding to the collected data;

[0014] Segment the expected value sequence and the actual value sequence respectively according to the multiple task phases to obtain multiple expected value subsequences and actual value subsequences corresponding to the multiple expected value subsequences respectively;

[0015] For each of the expected value subsequences, determine a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence;

[0016] Determine an error value between the expected value sequence and the actual value sequence according to the first error values corresponding to the multiple expected value subsequences.

[0017] In a possible design, the determining a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence includes:

[0018] Based on a preset error algorithm, determine a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; there are at least two preset error algorithms among the preset error algorithms corresponding to the multiple expected value subsequences.

[0019] In a possible design, the determining an error value between the expected value sequence and the actual value sequence according to the first error values corresponding to the multiple expected value subsequences includes:

[0020] Allocate corresponding weights to the first error values corresponding to the multiple expected value subsequences according to the number of elements of the multiple expected value subsequences;

[0021] Perform a weighted sum on the multiple first error values according to the weights corresponding to the multiple first error values to obtain a first result;

[0022] Determine an error value between the expected value sequence and the actual value sequence according to the first result.

[0023] In a possible design, the determining a similarity value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence includes:

[0024] Determine multiple task phases according to the task characteristics corresponding to the collected data;

[0025] According to multiple said task stages, segment the expected value sequence and the actual value sequence respectively to obtain multiple expected value subsequences and actual value subsequences corresponding to the multiple expected value subsequences respectively;

[0026] For each of the expected value subsequences, determine a first distance between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence;

[0027] Determine the similarity between the expected value sequence and the actual value sequence according to the first distances corresponding to the multiple expected value subsequences.

[0028] In a possible design, the determining the similarity between the expected value sequence and the actual value sequence according to the first distances corresponding to the multiple expected value subsequences includes:

[0029] According to the number of elements of the multiple expected value subsequences, assign corresponding weights to the first distance values corresponding to the multiple expected value subsequences;

[0030] Perform a weighted sum on the multiple first distance values according to the weights corresponding to the multiple first distance values to obtain a second result;

[0031] Determine the similarity between the expected value sequence and the actual value sequence according to the second result.

[0032] In a possible design, the determining the continuity evaluation value of the collected data according to the expected value sequence and the actual value sequence includes:

[0033] For each element among the multiple expected value elements in the expected value sequence, determine the adjacent element difference and the first derivative corresponding to the element according to the element, the adjacent element of the element, and the time interval between the element and the adjacent element of the element, and determine the first continuity evaluation value corresponding to the element according to the adjacent element difference and the first derivative;

[0034] For each element among the multiple actual value elements in the actual value sequence, determine the adjacent element difference and the first derivative corresponding to the element according to the element, the adjacent element of the element, and the time interval between the element and the adjacent element of the element, and determine the second continuity evaluation value corresponding to the element according to the adjacent element difference and the first derivative;

[0035] Determine the continuity evaluation value of the collected data according to the multiple first continuity evaluation values corresponding to the expected value sequence and the multiple second continuity evaluation values corresponding to the actual value sequence.

[0036] In a possible design, determining the quality inspection result of the collected data according to the latency evaluation value and the continuity evaluation value includes:

[0037] If the latency evaluation value is less than a first preset value, or the continuity evaluation value is less than a second preset value, it is determined that the quality inspection result of the collected data is unqualified.

[0038] In a second aspect, an embodiment of the present application provides a quality inspection device for humanoid robot data collection, including:

[0039] An acquisition module, configured to acquire the collected data of the humanoid robot; the collected data includes an expected value sequence and an actual value sequence of target parameters of the humanoid robot;

[0040] A first determination module, configured to determine a latency evaluation value and a continuity evaluation value of the collected data according to the expected value sequence and the actual value sequence;

[0041] A second determination module, configured to determine the quality inspection result of the collected data according to the latency evaluation value and the continuity evaluation value.

[0042] In a third aspect, an embodiment of the present application provides a quality inspection device for humanoid robot data collection, including: at least one processor and a memory;

[0043] The memory stores computer execution instructions;

[0044] The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the method described in the first aspect and various possible designs of the first aspect above.

[0045] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer execution instructions are stored, and when the processor executes the computer execution instructions, the method described in the first aspect and various possible designs of the first aspect above is implemented.

[0046] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method described in the first aspect and various possible designs of the first aspect above is implemented.

[0047] The quality inspection method and device for humanoid robot data acquisition provided in this embodiment include obtaining the acquisition data of the humanoid robot. The acquisition data includes the expected value sequence and the actual value sequence of the target parameters of the humanoid robot. According to the expected value sequence and the actual value sequence, the latency evaluation value and the continuity evaluation value of the acquisition data are determined, and the quality inspection result of the acquisition data is determined according to the latency evaluation value and the continuity evaluation value. The quality inspection method for humanoid robot data acquisition provided in the embodiments of the present application analyzes two time series, namely the expected value sequence and the actual value sequence, quantifies the latency and continuity errors of the two sequences, obtains the latency evaluation value and the continuity evaluation value, and then determines whether the acquisition data is qualified based on the evaluation value, which can realize automatic quality inspection, replace manual visual inspection, reduce costs, and improve the detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0049] Figure 1 It is a schematic diagram of the scenario for the quality inspection of humanoid robot data acquisition provided in the embodiments of the present application;

[0050] Figure 2 It is a schematic flowchart of the quality inspection method for humanoid robot data acquisition provided in the embodiments of the present application Figure 1 ;

[0051] Figure 3 It is a schematic flowchart of calculating the latency evaluation value between two sequences provided in the embodiments of the present application;

[0052] Figure 4a It is a schematic flowchart of calculating the continuity evaluation value of the actual value sequence provided in the embodiments of the present application;

[0053] Figure 4b It is a schematic flowchart of calculating the continuity evaluation value of the expected value sequence provided in the embodiments of the present application;

[0054] Figure 5 It is a schematic flowchart of the quality inspection method for humanoid robot data acquisition provided in the embodiments of the present application Figure 2 ;

[0055] Figure 6 It is a schematic structural diagram of the quality inspection device for humanoid robot data acquisition provided in the embodiments of the present application;

[0056] Figure 7 It is a schematic hardware structure diagram of the quality inspection device for humanoid robot data acquisition provided in the embodiments of the present application.

[0057] Through the above-mentioned accompanying drawings, specific embodiments of the present application have been shown, and more detailed descriptions will be provided hereinafter. These drawings and textual descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. Detailed Description of the Embodiments

[0058] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without making creative efforts fall within the scope of protection of the present application.

[0059] It should be noted that the quality detection method and device for data collection of humanoid robots provided in the present application can be used in the technical field of humanoid robots, and can also be used in any field other than the technical field of humanoid robots. The application field of the quality detection method and device for data collection of humanoid robots provided in the present application is not limited.

[0060] In order to enable humanoid robots to achieve more efficient, intelligent, and safe operations and lay a technical foundation for future widespread applications, it is necessary to collect data from humanoid robots. Taking the data collection when a humanoid robot performs tasks through its upper limbs (such as grasping an object on a table, washing dishes, etc.) as an example, the upper limb data of the humanoid robot can be collected through teleoperation. The upper limb data can include the body data of the robotic arm (the actual value and the expected value of the robotic arm). In the specific implementation process, the operator remotely operates the upper limb of the humanoid robot to collect data. The quality of the collected data will be affected by problems such as the operator's operation problems, temporary failures of the joint motors of the robotic arm, and delays in the robotic arm receiving communication signals from the teleoperation device. These problems are likely to cause abnormalities in the body data of the upper limb of the humanoid robot.

[0061] In the related art, the quality of the collected data can be detected by manual quality inspection. However, the method of manual quality inspection has many deficiencies. First of all, manual quality inspection often relies on the experience and subjective judgment of the operator, which is likely to lead to judgment differences among different personnel and form a certain experience deviation. Secondly, during the manual quality inspection process, the operator needs to concentrate on processing a large amount of data for a long time, which is likely to cause visual fatigue and further affect the accuracy and efficiency of quality inspection. Facing a large amount of collected data, manual quality inspection is not only extremely time-consuming and laborious, but also difficult to ensure full coverage of every detail. Especially when dealing with collected data with high requirements for continuity and low latency, omissions are likely to occur.

[0062] To solve the above technical problems, the inventors of the present application have found through research that the anomalies in data collection are reflected in the lack of continuity and the increase in latency. Therefore, the continuity and low latency of data collection are the key aspects that need to be concerned about in quality inspection. Moreover, with the continuous progress of robot technology, especially in scenarios where humanoid robots need to perform complex tasks, the requirements for real-time and high-precision data are getting higher and higher. Therefore, it is crucial to achieve automatic quality inspection as soon as possible. By formulating a reasonable strategy to analyze the continuity and low latency of the collected data, so as to achieve automatic quality inspection, it can not only save a large amount of labor costs, but also improve the efficiency and consistency of quality inspection while ensuring the accuracy of data processing. In addition, automated quality inspection can also monitor the changes in data in real time, quickly discover potential problems, and give early warnings in a timely manner, thereby enhancing the reliability and safety of the entire robot system. To sum up, the automated quality inspection method is of great significance in ensuring data quality, optimizing the quality inspection process, and improving production efficiency, and can provide a solid data guarantee for the application of humanoid robots in complex environments. Based on this, the embodiments of the present application provide a quality inspection method for data collection of humanoid robots.

[0063] Figure 1 It is a schematic diagram of the scenario of the quality inspection for data collection of humanoid robots provided by the embodiments of the present application. As Figure 1 shown, the teleoperation device 101 is communicatively connected to the humanoid robot 102, and the teleoperation device 101 is used to remotely control the humanoid robot 102. Both the teleoperation device 101 and the humanoid robot 102 are connected to the quality inspection device 103. Among them, the teleoperation device 101 is further used to generate an expected value and transmit the expected value to the quality inspection device 103. The humanoid robot 102 is used to generate a motion control instruction according to the expected value generated by the teleoperation device 101, generate a corresponding actual value based on the motion control instruction, and transmit the actual value to the quality inspection device 103. The quality inspection device 103 is used to perform quality inspection on the collected data (the sequence of expected values and the sequence of actual values for a certain task). The quality inspection device 103 can be a terminal device or a server. It can be seen that both the expected value and the actual value are data recorded by the humanoid robot during the task execution process. Therefore, both need to be reviewed and verified to obtain a more accurate quality inspection result.

[0064] In the specific implementation process, during the execution of a specific task (such as putting the apple on the table into the plate, or pouring a glass of water into a water cup, etc.), the teleoperation device 101 generates an expected value, obtains an expected value sequence and transmits it to the quality inspection device 103. The humanoid robot 102 receives the expected value sent by the teleoperation device 101, generates a motion control instruction based on the expected value, controls the upper limb to move based on the motion control instruction, generates a corresponding actual value, and obtains an actual value sequence. For example, assuming that it takes 20 seconds to complete the task and the data acquisition speed is 30 frames per second, 600 frames of data can be obtained. The expected value sequence includes the expected values corresponding to the 600 frames of data, and the actual value sequence includes the actual values corresponding to the 600 frames of data. The quality inspection device 103 acquires the acquired data, and the acquired data includes the expected value sequence and the actual value sequence of the target parameters (such as torque, angle, position, speed, acceleration, etc. of joints or ends) of the humanoid robot 102. According to the expected value sequence and the actual value sequence, a latency evaluation value and a continuity evaluation value of the acquired data are determined; according to the latency evaluation value and the continuity evaluation value, the quality inspection result of the acquired data is determined. The quality inspection method for humanoid robot data acquisition provided by the embodiments of the present application analyzes two time series, namely the expected value sequence and the actual value sequence, quantifies the latency and continuity errors of the two sequences, obtains the latency evaluation value and the continuity evaluation value, and then determines whether the acquired data is qualified based on the evaluation value, which can realize automatic quality inspection, replace manual visual inspection, reduce costs, and improve the detection efficiency.

[0065] It should be noted that Figure 1 The shown scene schematic diagram is only an example. The quality inspection method and the scene described in the embodiments of the present application for humanoid robot data acquisition are used to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation to the technical solutions provided by the embodiments of the present application. Those of ordinary skill in the art know that with the evolution of the system and the emergence of new business scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0066] The following uses specific embodiments to elaborate on the technical solutions of the present application in detail. The following several specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0067] Figure 2 It is a flow schematic of the quality inspection method for humanoid robot data acquisition provided by the embodiments of the present application Figure 1 As Figure 2 shown, the method includes:

[0068] 201. Acquire the acquired data of the humanoid robot; the acquired data includes the expected value sequence and the actual value sequence of the target parameters of the humanoid robot.

[0069] The execution subject of this embodiment is a terminal device or a server, such as Figure 1 the quality inspection device shown.

[0070] In this embodiment, the collected data refers to the motion data of each joint and / or the execution end of the upper limb, lower limb, torso, head, etc. of the robot collected when the humanoid robot performs a task. Exemplarily, taking the upper limb of the humanoid robot as an example, it can be the motion data of each joint of the left and right robotic arms of the humanoid robot. The target parameter can be at least one of angle, position, torque, and speed.

[0071] Among them, the data acquisition method can be a teleoperation method. The expected values in the expected value sequence can be generated by the teleoperation device under the operation of the operator, and the corresponding actual values in the actual value sequence are generated after the humanoid robot generates a motion control instruction based on the corresponding expected value and executes the motion control instruction. The expected value sequence and the corresponding actual value sequence are usually for a certain task, such as putting the apple on the table into the plate, or pouring water into the water cup, etc.

[0072] 202. Determine the latency evaluation value and the continuity evaluation value of the collected data according to the expected value sequence and the actual value sequence.

[0073] Specifically, after obtaining the expected value sequence and the actual value sequence in the collected data, the two sequences can be analyzed from two aspects of latency and continuity to obtain the latency evaluation value and the continuity evaluation value of the collected data.

[0074] Among them, the latency evaluation value refers to the quantitative evaluation result of the degree of delayed response of the actual value sequence relative to the expected value sequence, and the continuity evaluation value refers to the quantitative evaluation result that the adjacent data points of the actual value sequence change smoothly and without abnormal interruption.

[0075] In this embodiment, there can be multiple ways to calculate the delay evaluation value. In one implementable way, the error value between two sequences and the similarity value between the two sequences can be calculated first, and then the delay evaluation value can be determined based on the error value and the similarity value. Specifically, determining the delay evaluation value of the collected data according to the expected value sequence and the actual value sequence may include: determining the error value and the similarity value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence; determining the delay evaluation value of the collected data according to the error value and the similarity value between the expected value sequence and the actual value sequence. The quality detection method for humanoid robot data collection provided in this embodiment can comprehensively consider the error value and the similarity value between the expected value and the actual value sequence to determine the delay evaluation value, can comprehensively evaluate the delay characteristics of the collected data, and give a quantitative result that is more in line with the actual delay situation of the data.

[0076] Exemplarily, as Figure 3 shown, after obtaining the actual value sequence and the expected value sequence in the collected data, the similarity value and the error value between the two sequences are calculated respectively, and then the comprehensive delay evaluation value (i.e., the delay evaluation value) between the two sequences is determined based on the similarity value and the error value.

[0077] In another implementable way, based on the stage characteristics of the tasks corresponding to the two sequences, the expected value sequence can be segmented to obtain multiple expected value subsequences, and the actual value sequence can be segmented to obtain multiple actual value subsequences. Then, the delay evaluation value of each subsequence is calculated, and the delay evaluation value of the collected data is determined based on the delay evaluation value of each subsequence. Specifically, determining the delay evaluation value of the collected data according to the expected value sequence and the actual value sequence may include: determining multiple task stages according to the task characteristics corresponding to the collected data; segmenting the expected value sequence and the actual value sequence respectively according to the multiple task stages to obtain multiple expected value subsequences and multiple actual value subsequences corresponding to the multiple expected value subsequences respectively; determining the delay evaluation value of the collected data according to the multiple expected value subsequences and the multiple actual value subsequences. The quality detection method for humanoid robot data collection provided in this embodiment can segment the data sequence by combining the task characteristics and determine the delay evaluation value based on the segmented subsequences, can refine the analysis from the task dimension, more accurately measure the delay characteristics of the collected data, and reflect the data delay situation in actual applications.

[0078] There can be multiple ways to calculate the continuity evaluation value. In one implementable way, to simplify the operation and improve the calculation efficiency, the differences between adjacent elements of the two sequences can be calculated respectively, and then the continuity evaluation value of the collected data can be determined based on the differences between adjacent frame elements of the two sequences. In another implementable way, to improve the accuracy, the first-order derivatives of adjacent frames in the two sequences can be calculated respectively, and then the continuity evaluation value of the collected data can be determined based on the first-order derivatives of the two sequences.

[0079] In yet another implementable way, to further improve the accuracy, the differences and first-order derivatives of adjacent elements (i.e., adjacent frames) can be comprehensively considered to determine the continuity evaluation value of the collected data. Specifically, determining the continuity evaluation value of the collected data according to the expected value sequence and the actual value sequence may include: for each element in multiple expected value elements in the expected value sequence, determining the difference between adjacent elements and the first-order derivative corresponding to the element according to the element, the adjacent element of the element, and the time interval between the element and the adjacent element of the element, and determining the first continuity evaluation value corresponding to the element according to the difference between adjacent elements and the first-order derivative; for each element in multiple actual value elements in the actual value sequence, determining the difference between adjacent elements and the first-order derivative corresponding to the element according to the element, the adjacent element of the element, and the time interval between the element and the adjacent element of the element, and determining the second continuity evaluation value corresponding to the element according to the difference between adjacent elements and the first-order derivative; determining the continuity evaluation value of the collected data according to multiple first continuity evaluation values corresponding to the expected value sequence and multiple second continuity evaluation values corresponding to the actual value sequence. The quality detection method for humanoid robot data collection provided in this embodiment can automatically quantify the continuity and stability of data by analyzing the dynamic change trend (adjacent differences and first-order derivatives) of time-series data, accurately identify anomalies such as transient jumps and response lags, and then obtain more accurate quality inspection results, improving the efficiency compared to manual visual inspection and supporting early fault warning.

[0080] In this embodiment, determining the continuity evaluation value of the collected data according to multiple first continuity evaluation values corresponding to the expected value sequence and multiple second continuity evaluation values corresponding to the actual value sequence may include: using multiple first continuity evaluation values and multiple second continuity evaluation values as the continuity evaluation value of the collected data.

[0081] Exemplarily, as Figure 4a and Figure 4b shown, the continuity evaluation values of the two sequences can be calculated respectively. As Figure 4a shown, for the actual value sequence, taking the element as an example, first calculate The difference between corresponding adjacent elements and the first derivative, and then determine according to the difference between adjacent elements and the first derivative The corresponding second continuity evaluation value. For example Figure 4b As shown, for the expected value sequence, taking the element as an example, first calculate The difference between corresponding adjacent elements and the first derivative, and then determine according to the difference between adjacent elements and the first derivative The corresponding first continuity evaluation value. In the specific implementation process, assume that the actual value sequence includes and other n elements, and the corresponding expected value sequence includes and other n elements. In the actual value sequence The difference between adjacent elements of

[0082] (1)

[0083] Among them, Is the i-th element (i.e., the actual value) in the actual value sequence, Is the (i + 1)-th element in the actual value sequence, y i-1 Is the (i - 1)-th element in the actual value sequence, y i+1 And y i-1 Are both adjacent elements of y i .

[0084] The corresponding first derivative is:

[0085] (2)

[0086] Among them, h is And The time interval between, y i Is the i-th element in the actual value sequence, y i+1 Is the (i + 1)-th element in the actual value sequence, and is y i Adjacent elements.

[0087] Then y i The corresponding second continuity evaluation value is:

[0088] (3)

[0089] Among them, y i Is the i-th element (i.e., the actual value) in the actual value sequence, y i+1 Is the (i + 1)-th element in the actual value sequence, y i-1 Is the (i - 1)-th element in the actual value sequence, and β is the weighting coefficient.

[0090] Similarly, in the expected value sequence The first continuity evaluation value is:

[0091] (4)

[0092] where is the i-th element (i.e., the expected value) in the expected value sequence, is the (i + 1)-th element in the expected value sequence, is the (i - 1)-th element in the expected value sequence, is the weighting coefficient, which can be determined according to the prior initial value.

[0093] 203. Determine the quality inspection result of the collected data according to the delay evaluation value and the continuity evaluation value.

[0094] Specifically, after determining the delay evaluation value and the continuity evaluation value of the collected data, thresholds can be set for the delay evaluation value and the continuity evaluation value respectively, and then the final quality inspection result can be determined through the threshold comparison result.

[0095] In this embodiment, a qualified quality inspection result means that both the expected value sequence and the actual value sequence are qualified. Conversely, an unqualified quality inspection result means that both the expected value sequence and the actual value sequence are unqualified.

[0096] Among them, there are various ways to determine the quality inspection result. In one implementable way, when the amount of data of the collected data is relatively small, in order to obtain more qualified collected data, relatively loose quality inspection conditions can be set. For example, when at least one of the delay evaluation value and the continuity evaluation value is qualified, the quality inspection result can be determined as qualified, and when both evaluation values are unqualified, the quality inspection result is determined as unqualified. If the quality inspection result is qualified, it means that both the expected value sequence and the actual value sequence are qualified and can be normally applied, such as being used as training data for a humanoid robot model. Conversely, if it means that both the expected value sequence and the actual value sequence are unqualified, for example, these collected data can be discarded.

[0097] In another implementable manner, in order to obtain more accurate data, relatively strict quality inspection conditions can be set. When one of the latency evaluation value or the continuity evaluation value is unqualified, the quality inspection result is determined to be unqualified. Only when both the latency evaluation value and the continuity evaluation value are qualified can the quality inspection result be determined to be qualified. Specifically, determining the quality inspection result of the collected data according to the latency evaluation value and the continuity evaluation value may include: if the latency evaluation value is greater than a first preset value, or the continuity evaluation value is greater than a second preset value, it is determined that the quality inspection result of the collected data is unqualified. The quality inspection method for humanoid robot data collection provided in this embodiment can effectively screen out low-quality data and retain high-quality data by setting strict quality inspection conditions and judging based on the latency evaluation value and the continuity evaluation value, ensuring the accuracy of the collected data.

[0098] Exemplarily, as Figure 5 shown, for the actual value sequence and the expected value sequence in the collected data, the comprehensive latency evaluation value (latency evaluation value) between the two sequences, the comprehensive continuity evaluation value of the actual value sequence (multiple second continuity evaluation values), and the comprehensive continuity evaluation value of the expected value sequence (multiple first continuity evaluation values) can be calculated respectively. Then, based on the comparison results between the comprehensive continuity evaluation value of the actual value sequence and the comprehensive continuity evaluation value of the expected value sequence and the continuity threshold, it is determined whether the continuity of the two sequences is qualified. Based on the comparison result between the latency evaluation value and the latency threshold, it is determined whether the latency is qualified. If both the latency evaluation value and the continuity evaluation value are qualified, the collected data is determined to be qualified. If there is one unqualified, the collected data is determined to be unqualified.

[0099] The quality inspection method for humanoid robot data collection provided in this embodiment analyzes two time series, namely the expected value sequence and the actual value sequence, quantifies the latency and continuity errors of the two sequences, obtains the latency evaluation value and the continuity evaluation value, and then determines whether the collected data is qualified based on this evaluation value, which can realize automatic quality inspection, replace manual visual inspection, reduce costs, and improve the detection efficiency.

[0100] The following provides a detailed example description of two implementable manners of the latency evaluation value.

[0101] For the first implementable manner of determining the latency evaluation value in the above embodiment, there are various ways to calculate the error value. In one implementable manner, the error value of the entire sequence can be calculated based on a preset error algorithm.

[0102] In another implementable manner, the sequence can be segmented first, and then based on the subsequences obtained by segmentation, the error values of the subsequences can be calculated respectively, and then the error value between the expected value sequence and the actual value can be determined based on the error values of the subsequences. Specifically, determining the error value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence may include: determining a plurality of task stages according to the task characteristics corresponding to the collected data; segmenting the expected value sequence and the actual value sequence respectively according to the plurality of task stages to obtain a plurality of expected value subsequences and the actual value subsequences respectively corresponding to the plurality of expected value subsequences; for each of the expected value subsequences, determining a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; and determining the error value between the expected value sequence and the actual value sequence according to the first error values respectively corresponding to the plurality of expected value subsequences. The quality detection method for humanoid robot data acquisition provided in this embodiment can refine the delay characteristic analysis in different stages, accurately locate the data delay problem, improve the accuracy of the overall error evaluation, and better meet the quality detection requirements of the actual task scenario by calculating the error values of the subsequences in segments in combination with the task characteristics and comprehensively evaluating.

[0103] In yet another implementable manner, different error algorithms can be adopted for the segmented subsequences to calculate the errors of the corresponding subsequences, and then the error value between the expected value sequence and the actual value sequence can be determined based on the error values of each subsequence. Specifically, determining the first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence may include: determining the first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence based on a preset error algorithm according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; and there are at least two preset error algorithms among the preset error algorithms respectively corresponding to the plurality of expected value subsequences. The quality detection method for humanoid robot data acquisition provided in this embodiment can optimize the error evaluation according to the data characteristics in different stages, improve the accuracy of the overall error calculation, and more effectively reflect the data quality problems in actual applications by segmenting in combination with the task characteristics and using a differential error algorithm to calculate the subsequence errors.

[0104] In this embodiment, the preset error algorithm may be an algorithm such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), or symmetric Mean Absolute Percentage Error (sMAPE).

[0105] Exemplarily, different error algorithms may be adopted according to the characteristics of different stages of the task to improve the accuracy and rationality of error value calculation. Suppose that in the starting time period, there is a lot of noise and many outliers, the MAE algorithm can be selected, which is robust to outliers and reduces the influence of transient interference. If the middle time period is stable and is a critical business stage, the MSE or RMSE algorithm can be selected to emphasize accurate tracking and punish large errors. In the ending time period, attenuation or mutation may occur, and the sMAPE algorithm can be selected to avoid the zero-value problem and focus on trend consistency.

[0106] In some embodiments, determining the error value between the expected value sequence and the actual value sequence according to the first error values respectively corresponding to multiple expected value subsequences may include: allocating corresponding weights to the first error values respectively corresponding to multiple expected value subsequences according to the number of elements of multiple expected value subsequences; performing weighted sum on multiple first error values according to the weights respectively corresponding to multiple first error values to obtain a first result; and determining the error value between the expected value sequence and the actual value sequence according to the first result. The quality detection method for humanoid robot data acquisition provided in this embodiment can objectively reflect the contribution of the data volume of each stage to the overall error, improve the objectivity and accuracy of error evaluation, and more comprehensively measure the data quality by allocating weights based on the number of subsequence elements and performing weighted sum.

[0107] Exemplarily, suppose the actual value sequence includes and other n elements, and the corresponding expected value sequence includes and other n elements. The calculation formula for the error value between the two sequences may be:[[]]

[0108] (5)

[0109] where is the i-th element in the actual value sequence, is the i-th element in the expected value sequence, n is the number of elements in the sequence, k and m are the k-th and m-th elements in the sequence respectively, and 1 < k < m < n. The values of k and m are determined according to the length n of the current sequence data and the task characteristics of the corresponding task. Generally, the larger n is, the larger k and m are. The first k elements represent the data in the starting time period of the trajectory, the elements from k + 1 to m in the middle represent the middle segment data, and the elements after m represent the data in the ending time period.

[0110] In this embodiment, the first k elements are used as the subsequence of the starting time period, the elements from k + 1 to m are used as the subsequence of the middle time period, and the elements after m are used as the subsequence of the ending time period. The RMSE algorithm is adopted, and the weight values of each subsequence are determined according to the number of elements, and then the error value between the actual value sequence and the expected value sequence is determined.

[0111] For the first realizable manner of determining the delay evaluation value in the above embodiment, there are multiple ways to calculate the similarity value. In one realizable manner, the similarity value of the entire sequence can be calculated based on a preset similarity algorithm.

[0112] In another realizable manner, the sequence can be segmented first, and then based on the subsequences obtained by segmentation, the similarity values of the subsequences are calculated respectively, and then the similarity value between the expected value sequence and the actual value is determined based on the similarity values of the subsequences. Specifically, according to the expected value sequence and the actual value sequence, determining the similarity value between the expected value sequence and the actual value sequence may include: determining multiple task stages according to the task characteristics corresponding to the collected data; segmenting the expected value sequence and the actual value sequence respectively according to the multiple task stages to obtain multiple expected value subsequences and the actual value subsequences corresponding to the multiple expected value subsequences respectively; for each expected value subsequence, determining the first distance between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; determining the similarity between the expected value sequence and the actual value sequence according to the first distances corresponding to the multiple expected value subsequences. The quality detection method for humanoid robot data collection provided in this embodiment can perform refined analysis on the data characteristics of different stages by calculating the similarity of subsequences in combination with task characteristics, accurately reflect the similarity levels of each stage, thereby improving the accuracy and comprehensiveness of the overall similarity evaluation, and better meeting the quality detection requirements for humanoid robot data collection in actual application scenarios.

[0113] Exemplarily, for each time step i and k, define the distance between the i-th element in the actual value sequence and the k-th element in the expected value sequence as:

[0114] (6)

[0115] Furthermore, a matrix D is constructed, where each element D(i,k) in the matrix represents the cumulative minimum distance from the i-th element of the actual value sequence to the k-th element of the expected value sequence. This matrix can be calculated through the following recurrence relation:

[0116] (7)

[0117] Then, for D(k,m) and D(m,n), similarly, we can obtain:

[0118] (8)

[0119] (9)

[0120] Finally, the similarity between the actual value sequence and the expected value sequence can be expressed as:

[0121] (10)

[0122] Therefore, the latency evaluation value of the collected data is:

[0123] (11)

[0124] Among them, is the weighting coefficient, and the initial value can be determined according to the prior.

[0125] In some embodiments, if the latency evaluation value is greater than the set threshold , it is determined that the currently collected data is unqualified.

[0126] In one implementable manner, different similarity algorithms can be adopted for the segmented subsequences to calculate the similarity of the corresponding subsequences, and then the similarity value between the expected value sequence and the actual value sequence can be determined based on the similarity values of each subsequence. Specifically, determining the first similarity value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence may include: based on a preset similarity algorithm, determining the first similarity value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; there are at least two preset similarity algorithms among the preset similarity algorithms respectively corresponding to multiple expected value subsequences. The quality detection method for humanoid robot data acquisition provided in this embodiment can select the most suitable algorithm according to the characteristics of data in each task stage by calculating the similarity value using different similarity algorithms for different subsequences, and measure the subsequence similarity more accurately. Furthermore, the overall similarity value obtained comprehensively will be more accurate and comprehensive, can better reflect the similarity degree between the expected value sequence and the actual value sequence, and improve the reliability of data quality evaluation.

[0127] In some embodiments, determining the similarity between the expected value sequence and the actual value sequence according to the first distances respectively corresponding to multiple expected value subsequences may include: allocating corresponding weights to the first distance values respectively corresponding to multiple expected value subsequences according to the number of elements of multiple expected value subsequences; performing a weighted sum on multiple first distance values according to the weights respectively corresponding to multiple first distance values to obtain a second result; determining the similarity between the expected value sequence and the actual value sequence according to the second result. The quality detection method for humanoid robot data acquisition provided in this embodiment can fully consider the influence of the data volume in different stages on the overall similarity by allocating weights to the first distance values of each subsequence according to the number of elements of the expected value subsequence and performing a weighted sum, making the finally determined similarity more in line with the actual situation and improving the evaluation accuracy and reliability.

[0128] For the second realizable manner of determining the delay evaluation value in the above embodiments, according to the multiple expected value subsequences and the multiple actual value subsequences, determining the delay evaluation value of the collected data may include: for each expected value subsequence, according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence, determining an error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; based on a preset distance algorithm corresponding to the expected value subsequence, according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence, determining a similarity value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; according to the error values and similarity values respectively corresponding to the multiple expected value subsequences, determining the delay evaluation value of the collected data. The quality detection method for humanoid robot data collection provided in this embodiment can accurately measure the delay characteristics from two dimensions of numerical difference and trend consistency by comprehensively analyzing the error value and similarity value for subsequence delay, and combining the differential algorithm to evaluate the data matching degree at each stage, improving the comprehensiveness and flexibility of the evaluation, and ensuring that the delay evaluation value better meets the data quality requirements of the actual application scenario.

[0129] Among them, according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence, determining the error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence may include: based on a preset error algorithm corresponding to the expected value subsequence, according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence, determining the error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; the preset error algorithms corresponding to different expected value subsequences are not completely the same; and / or, according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence, determining the similarity value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence may include: based on a preset distance algorithm corresponding to the expected value subsequence, according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence, determining the similarity value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; the preset distance algorithms corresponding to different expected value subsequences are not completely the same. The quality detection method for humanoid robot data collection provided in this embodiment can accurately analyze according to the characteristics of the subsequences by using different preset error algorithms and distance algorithms to calculate the error and similarity between each expected value subsequence and its corresponding actual value subsequence, making the delay evaluation value of the collected data more accurately reflect the actual delay situation and improving the reliability of data quality evaluation.

[0130] Figure 6 This is a schematic structural diagram of a quality detection device for humanoid robot data collection provided by an embodiment of the present application. As Figure 6As shown in the figure, the quality inspection device 60 for humanoid robot data acquisition includes: an acquisition module 601, a first determination module 602, and a second determination module 603.

[0131] The acquisition module 601 is configured to acquire the acquisition data of the humanoid robot; the acquisition data includes an expected value sequence and an actual value sequence of the target parameters of the humanoid robot.

[0132] The first determination module 602 is configured to determine a latency evaluation value and a continuity evaluation value of the acquisition data according to the expected value sequence and the actual value sequence.

[0133] The second determination module 603 is configured to determine the quality inspection result of the acquisition data according to the latency evaluation value and the continuity evaluation value.

[0134] The quality inspection device for humanoid robot data acquisition provided by the embodiments of the present application analyzes two time series, namely the expected value sequence and the actual value sequence, quantifies the latency and continuity errors of the two sequences, obtains the latency evaluation value and the continuity evaluation value, and then determines whether the acquisition data is qualified based on the evaluation value, which can realize automatic quality inspection, replace manual visual inspection, reduce costs, and improve the detection efficiency.

[0135] In some embodiments, the first determination module 602 is specifically configured to determine an error value and a similarity value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence; and determine the latency evaluation value of the acquisition data according to the error value and the similarity value between the expected value sequence and the actual value sequence.

[0136] In some embodiments, the first determination module 602 is specifically configured to determine multiple task stages according to the task characteristics corresponding to the acquisition data; segment the expected value sequence and the actual value sequence respectively according to the multiple task stages to obtain multiple expected value subsequences and multiple actual value subsequences corresponding to the multiple expected value subsequences respectively; for each expected value subsequence, determine a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; and determine the error value between the expected value sequence and the actual value sequence according to the first error values corresponding to the multiple expected value subsequences.

[0137] In some embodiments, the first determination module 602 is specifically configured to determine a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence based on a preset error algorithm according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; there are at least two preset error algorithms among the preset error algorithms corresponding to the multiple expected value subsequences respectively.

[0138] In some embodiments, the first determination module 602 is specifically configured to assign corresponding weights to the first error values corresponding to the multiple expected value subsequences according to the number of elements of the multiple expected value subsequences; perform a weighted sum on the multiple first error values according to the weights corresponding to the multiple first error values respectively to obtain a first result; determine an error value between the expected value sequence and the actual value sequence according to the first result.

[0139] In some embodiments, the first determination module 602 is specifically configured to determine multiple task stages according to the task characteristics corresponding to the collected data; segment the expected value sequence and the actual value sequence respectively according to the multiple task stages to obtain multiple expected value subsequences and the actual value subsequences corresponding to the multiple expected value subsequences respectively; for each expected value subsequence, determine a first distance between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; determine the similarity between the expected value sequence and the actual value sequence according to the first distances corresponding to the multiple expected value subsequences respectively.

[0140] In some embodiments, the first determination module 602 is specifically configured to assign corresponding weights to the first distance values corresponding to the multiple expected value subsequences according to the number of elements of the multiple expected value subsequences; perform a weighted sum on the multiple first distance values according to the weights corresponding to the multiple first distance values respectively to obtain a second result; determine the similarity between the expected value sequence and the actual value sequence according to the second result.

[0141] In some embodiments, the first determination module 602 is specifically configured to, for each element in a plurality of expected value elements in the expected value sequence, determine an adjacent element difference and a first-order derivative corresponding to the element according to the element, an adjacent element of the element, and a time interval between the element and the adjacent element of the element, and determine a first continuity evaluation value corresponding to the element according to the adjacent element difference and the first-order derivative; for each element in a plurality of actual value elements in the actual value sequence, determine an adjacent element difference and a first-order derivative corresponding to the element according to the element, an adjacent element of the element, and a time interval between the element and the adjacent element of the element, and determine a second continuity evaluation value corresponding to the element according to the adjacent element difference and the first-order derivative; and determine a continuity evaluation value of the collected data according to the plurality of first continuity evaluation values corresponding to the expected value sequence and the plurality of second continuity evaluation values corresponding to the actual value sequence.

[0142] In some embodiments, the second determination module 603 is specifically configured to determine that the quality inspection result of the collected data is unqualified if the delay evaluation value is less than a first preset value or the continuity evaluation value is less than a second preset value.

[0143] The quality inspection device for humanoid robot data collection provided in the embodiments of the present application can be used to execute the above method embodiments, and its implementation principle and technical effects are similar, which will not be elaborated herein.

[0144] Figure 7 It is a schematic hardware structure diagram of the quality inspection device for humanoid robot data collection provided in the embodiments of the present application. As Figure 7 shown, the electronic device 70 provided in this embodiment includes at least one processor 701 and a memory 702. Optionally, the device 70 further includes a communication component 703. Among them, the processor 701, the memory 702, and the communication component 703 are connected through a bus.

[0145] In a specific implementation process, at least one processor 701 executes computer execution instructions stored in the memory 702, so that at least one processor 701 executes the above method.

[0146] The specific implementation process of the processor 701 can refer to the above method embodiments, and its implementation principle and technical effects are similar, which will not be elaborated herein.

[0147] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU for short), or other general-purpose processors, digital signal processors (DSP for short), application specific integrated circuits (ASIC for short), etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the method disclosed in combination with the invention can be directly implemented by the execution of the hardware processor, or by the combination of hardware and software modules in the processor.

[0148] The memory may include a random access memory (RAM), and may also include a non-volatile memory (NVM), such as at least one disk memory.

[0149] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0150] This application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0151] This application also provides a computer-readable storage medium, in which computer-executable instructions are stored, and when the processor executes the computer-executable instructions, the above method is implemented.

[0152] The above-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk. The readable storage medium can be any available medium accessible by a general-purpose or special-purpose computer.

[0153] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an Application Specific Integrated Circuits (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in a device.

[0154] The division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Additionally, the couplings or direct couplings or communication connections shown or discussed among each other can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0155] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0156] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can physically exist alone, or two or more units can be integrated in one unit.

[0157] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0158] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including those of the above method embodiments; and the foregoing storage medium includes: various media such as ROM, RAM, magnetic disk, or optical disc that can store program codes.

[0159] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include known common general knowledge or conventional technical means in the technical field not disclosed by the present invention. It is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A quality inspection method for data collection of humanoid robots, characterized in that, Including: Obtain the acquisition data of the humanoid robot; The acquisition data includes the expected value sequence and the actual value sequence of the target parameters of the humanoid robot; According to the expected value sequence and the actual value sequence, determine the latency evaluation value and the continuity evaluation value of the acquisition data; Determine the quality inspection result of the acquisition data according to the latency evaluation value and the continuity evaluation value; The determining the latency evaluation value of the acquisition data according to the expected value sequence and the actual value sequence includes: According to the expected value sequence and the actual value sequence, determine the error value and the similarity value between the expected value sequence and the actual value sequence; Determine the latency evaluation value of the acquisition data according to the error value and the similarity value between the expected value sequence and the actual value sequence; The determining the continuity evaluation value of the acquisition data according to the expected value sequence and the actual value sequence includes: For each of multiple expected value elements in the expected value sequence, according to the element, its adjacent element, and the time interval between the element and its adjacent element, determine the adjacent element difference and the first derivative corresponding to the element, and determine the first continuity evaluation value corresponding to the element according to the adjacent element difference and the first derivative; For each of multiple actual value elements in the actual value sequence, according to the element, its adjacent element, and the time interval between the element and its adjacent element, determine the adjacent element difference and the first derivative corresponding to the element, and determine the second continuity evaluation value corresponding to the element according to the adjacent element difference and the first derivative; Determine the continuity evaluation value of the acquisition data according to the multiple first continuity evaluation values corresponding to the expected value sequence and the multiple second continuity evaluation values corresponding to the actual value sequence.

2. The method according to claim 1, characterized in that, The determining the error value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence includes: Determine multiple task stages according to the task characteristics corresponding to the acquisition data; According to the multiple task stages, segment the expected value sequence and the actual value sequence respectively to obtain multiple expected value subsequences and the actual value subsequences corresponding to the multiple expected value subsequences respectively; For each expected value subsequence, determine the first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; Determine the error value between the expected value sequence and the actual value sequence according to the first error values corresponding to the multiple expected value subsequences respectively.

3. The method according to claim 2, wherein The determining the first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence includes: Based on a preset error algorithm, determine a first error value between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; there are at least two preset error algorithms among the preset error algorithms corresponding to multiple expected value subsequences.

4. The method according to claim 2, wherein The determining of the error value between the expected value sequence and the actual value sequence according to the first error values corresponding to multiple expected value subsequences respectively includes: According to the number of elements of multiple expected value subsequences, assign corresponding weights to the first error values corresponding to multiple expected value subsequences respectively; Perform a weighted sum on multiple first error values according to the weights corresponding to multiple first error values respectively to obtain a first result; Determine the error value between the expected value sequence and the actual value sequence according to the first result.

5. The method according to claim 1, wherein The determining of the similarity value between the expected value sequence and the actual value sequence according to the expected value sequence and the actual value sequence includes: Determine multiple task stages according to the task characteristics corresponding to the collected data; Segment the expected value sequence and the actual value sequence respectively according to multiple task stages to obtain multiple expected value subsequences and the actual value subsequences corresponding to multiple expected value subsequences respectively; For each expected value subsequence, determine a first distance between the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence according to the expected value subsequence and the actual value subsequence corresponding to the expected value subsequence; Determine the similarity between the expected value sequence and the actual value sequence according to the first distances corresponding to multiple expected value subsequences respectively.

6. The method according to claim 5, wherein The determining of the similarity between the expected value sequence and the actual value sequence according to the first distances corresponding to multiple expected value subsequences respectively includes: According to the number of elements of multiple expected value subsequences, assign corresponding weights to the first distance values corresponding to multiple expected value subsequences respectively; Perform a weighted sum on multiple first distance values according to the weights corresponding to multiple first distance values respectively to obtain a second result; Determine the similarity between the expected value sequence and the actual value sequence according to the second result.

7. The method according to any one of claims 1 to 6, characterized in that, The determining of the quality inspection result of the collected data according to the latency evaluation value and the continuity evaluation value includes: If the latency evaluation value is less than a first preset value, or the continuity evaluation value is less than a second preset value, then determine that the quality inspection result of the collected data is unqualified.

8. A quality inspection device for data collection of humanoid robots, characterized in that, Includes: At least one processor and a memory; The memory stores computer execution instructions; The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the quality inspection method for humanoid robot data collection according to any one of claims 1 to 7.

Images