Multi-robot 3D vision cooperation space attitude sensing calibration method and system

Abstract

The invention provides a multi-robot 3D vision cooperation space attitude sensing calibration method and system, and relates to the technical field of multi-robot cooperation work. And collecting three-dimensional visual data in real time in the cooperative working area by using at least three robots. Setting a calibration precision factor to quantify an attitude estimation error, and preprocessing the collected data, including noise reduction and normalization; the weight of each data source is calculated to reflect the importance of the data source in the fusion process, a fused attitude estimation model is generated through dynamic Kalman filtering, and historical information is introduced. A motion estimation error is set and monitored in real time, when the error exceeds a threshold value, a calibration re-correction mechanism is triggered, the robot is reconfigured for data collection and weight updating, and it is ensured that the calibration precision continuously meets the requirement. According to the method, the attitude sensing precision of the multiple robots in a complex environment is improved, the data processing complexity is reduced, the real-time performance of the system is enhanced, and a foundation is laid for wide application of cooperative work of the multiple robots.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-robot collaborative work, and in particular to a multi-robot 3D visual collaborative spatial posture perception calibration method and system. Background Art

[0002] With the rapid development of robotics technology, multi-robot systems have been widely used in industrial automation, warehousing logistics, intelligent manufacturing and other fields. The collaborative work of multiple robots can effectively improve work efficiency and flexibility. However, how to accurately perceive and calibrate the spatial posture of each robot is still one of the key challenges to achieve effective collaboration.

[0003] Traditional posture perception methods usually rely on a single robot to locate itself through an inertial measurement unit (IMU) or visual sensor. However, in complex environments, a single sensor is easily affected by noise and occlusion, resulting in inaccurate posture estimation. In addition, as the number of robots increases, the complexity of data processing also increases. How to effectively fuse visual data from different robots to improve the accuracy and robustness of the overall system has become an urgent problem to be solved.

[0004] In recent years, the progress of 3D vision technology has provided a new solution for multi-robot posture perception. Using multiple robots to collect 3D visual data in real time in a collaborative work area can more comprehensively reflect environmental information. However, how to process and fuse this data to obtain reliable posture estimation still requires effective calibration methods and algorithm support.

[0005] Existing research on multi-robot systems in data fusion and posture estimation focuses on the establishment of theoretical models, lacking effective algorithms and system designs for practical application scenarios. Therefore, an innovative multi-robot 3D visual collaborative spatial posture perception calibration method is proposed, which can not only improve the accuracy of posture estimation, but also significantly enhance the adaptability and reliability of the system, which has important practical application value and research significance. Summary of the invention

[0006] In order to solve the technical problems in the prior art of insufficient posture perception accuracy of multiple robots in complex environments, high data processing complexity, poor real-time performance, incomplete data collection, and difficulty in monitoring and correcting calibration errors, the present invention provides a multi-robot 3D vision collaborative spatial posture perception calibration method and system.

[0007] The technical solution provided by the present invention is as follows:

[0008] First aspect:

[0009] The present invention provides a multi-robot 3D vision collaborative spatial posture perception calibration method, comprising:

[0010] S1. In the collaborative work area, use at least three robots to collect a 3D visual data set D = {d 1 ,d 2 ,...,d n}, each data point d i Represents three-dimensional coordinate information;

[0011] S2. Set the calibration accuracy factor C Align , used to quantify the error range of attitude estimation, the calibration accuracy factor C Align The calculation formula is:

[0012]

[0013] Where n represents the number of collected 3D data points, m represents the dimension of each data point, and d ij represents the measured value of the i-th data point in the j-th dimension, d' ij Represents the true coordinates of the i-th data point in the j-th dimension;

[0014] S3, preprocessing the collected three-dimensional data, including noise reduction and data normalization, to obtain a unified data format D' = {d' 1 ,d' 2 ,...,d' n};

[0015] S4. Data fusion is performed through the following formula:

[0016]

[0017] Among them, F Blend is the fused pose estimation result, w i represents the weight of the i-th data source, reflecting its importance in fusion, d' i represents the i-th data point after preprocessing;

[0018] S5, generating a fused attitude estimation model and applying dynamic Kalman filtering to optimize the data fusion process;

[0019] S6. Set motion estimation error E Shift , the motion estimation error E Shift The calculation method is:

[0020]

[0021] Among them, E Shift represents the estimated motion error, represents the estimated motion value of the jth dimension, represents the true motion value of the jth dimension, α represents the weight factor, reflecting the impact of calibration accuracy on the error;

[0022] S7, when the motion estimation error E Shift When the set threshold is exceeded, the calibration recalibration mechanism is triggered.

[0023] Second aspect:

[0024] The present invention provides a multi-robot 3D vision collaborative space posture perception calibration system, comprising:

[0025] Multi-robot configuration, with at least three autonomous mobile robots equipped with high-precision 3D vision sensors that can collect three-dimensional visual data in real time within the collaborative work area;

[0026] The data processing unit is equipped with a centralized processor, which is responsible for receiving the three-dimensional data sets collected from each robot and performing data fusion and motion estimation;

[0027] Wireless communication module, configuring wireless communication network to realize real-time data transmission between robots and between robots and data processing units, so as to ensure the timeliness and accuracy of data sharing;

[0028] The calibration precision factor calculation module is used to dynamically calculate the calibration precision factor to quantify the error range of attitude estimation and to optimize the data fusion process;

[0029] The weight calculation module calculates the weight factor based on the confidence of each data source to reflect the importance of each data source in the fusion process and improve the accuracy of the fusion result;

[0030] The calibration recalibration module monitors and analyzes the motion estimation error. When the error exceeds the set threshold, the recalibration mechanism is automatically triggered to update the robot calibration status.

[0031] The user interaction interface provides a visual operation interface, allowing users to monitor data acquisition, fusion and calibration status in real time and adjust related parameters.

[0032] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0033] (1) In the present invention, through the multi-robot collaborative working mechanism, at least three robots are used to collect three-dimensional visual data in real time, which effectively improves the accuracy of posture perception. This collaborative mode not only enhances the comprehensiveness of the data, but also optimizes the diversity of data sources, making posture estimation in complex environments more reliable;

[0034] (2) In the present invention, the state transfer matrix is ​​dynamically adjusted by combining the relative position changes of each robot through the dynamic Kalman filter algorithm, which significantly reduces the complexity of data processing and enhances the real-time performance of the system. This technical means enables the system to respond to environmental changes in real time and quickly adjust the motion strategy, thereby improving the efficiency of data collection;

[0035] (3) In the present invention, through the calibration recalibration mechanism, the system can monitor the motion estimation error in real time and automatically trigger the recalibration process when the error exceeds the set threshold. This mechanism ensures that the robot can continue to maintain high-efficiency calibration accuracy during long-term operation, avoiding performance degradation due to environmental changes or equipment aging, thereby improving the stability and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 A schematic diagram of a flow chart of a multi-robot 3D vision collaborative spatial posture perception calibration method provided by an embodiment of the present invention;

[0038] Figure 2 A schematic diagram of the structure of a multi-robot 3D vision collaborative spatial posture perception and calibration system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0040] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0041] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0042] In the embodiments of the present invention, sometimes the subscripts such as W 1 It may be mistakenly written as a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0043] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0044] Reference Manual Attached Figure 1 , which shows a flow chart of a multi-robot 3D vision collaborative spatial posture perception calibration method provided by an embodiment of the present invention.

[0045] The embodiment of the present invention provides a multi-robot 3D vision collaborative space posture perception calibration method, which can be implemented by a multi-robot 3D vision collaborative space posture perception calibration device, which can be a terminal or a server. The processing flow of the multi-robot 3D vision collaborative space posture perception calibration method may include the following steps:

[0046] S1. In the collaborative work area, use at least three robots to collect a 3D visual data set D = {d 1 ,d 2 ,...,d n}, each data point d i Represents three-dimensional coordinate information.

[0047] It should be noted that by collecting 3D visual data sets in real time by at least three robots, the collaborative working area can be fully covered, blind spots can be reduced, and rich and accurate environmental information can be obtained. This multi-source data collection provides a more reliable basis for subsequent posture estimation and improves the overall data quality of the system.

[0048] S2. Set the calibration accuracy factor C Align , used to quantify the error range of attitude estimation and calibrate the precision factor C Align The calculation formula is:

[0049]

[0050] Where n represents the number of collected 3D data points, m represents the dimension of each data point, and d ij represents the measured value of the i-th data point in the j-th dimension, d' ij Represents the true coordinates of the i-th data point in the j-th dimension.

[0051] It should be noted that the calibration precision factor is set to quantify the error range in attitude estimation. The calculation of this factor helps to clarify the accuracy requirements in data processing and enhance confidence in the estimation results, thereby improving the accuracy of data fusion and motion estimation in subsequent steps.

[0052] S3, preprocessing the collected three-dimensional data, including noise reduction and data normalization, to obtain a unified data format D' = {d' 1 ,d' 2 ,...,d' n}.

[0053] It should be noted that the preprocessing of 3D data (noise reduction and data normalization) is to improve data quality and consistency. By eliminating noise and standardizing the data format, the comparability of data in subsequent analysis can be ensured, thereby optimizing the effect of data fusion.

[0054] S4. Data fusion is performed through the following formula:

[0055]

[0056] Among them, F Blend is the fused pose estimation result, w i represents the weight of the i-th data source, reflecting its importance in fusion, d' i represents the i-th data point after preprocessing.

[0057] It should be noted that data fusion aims to integrate multiple data sources by setting weights to reflect the importance of different data sources and generate more accurate posture estimates. This step ensures that the system can dynamically adapt to changes in data quality and enhance its responsiveness to environmental changes.

[0058] S5. Generate a fused attitude estimation model and apply dynamic Kalman filtering to optimize the data fusion process.

[0059] It should be noted that the purpose of generating a fused attitude estimation model and applying dynamic Kalman filtering is to update and optimize the data fusion process in real time. This step improves the accuracy of attitude estimation and the real-time performance of the system by reducing the impact of noise on the estimation results.

[0060] S6. Set motion estimation error E Shift , motion estimation error E Shift The calculation method is:

[0061]

[0062] Among them, E Shift represents the estimated motion error, represents the estimated motion value of the jth dimension, represents the true motion value of the jth dimension, α represents the weight factor, and reflects the influence of calibration accuracy on the error.

[0063] It should be noted that the motion estimation error and its calculation method are set to monitor the robot's motion accuracy. By quantifying the error, the effectiveness of the model can be evaluated, and a basis for optimization can be provided to ensure the accuracy of the final output.

[0064] S7, when the motion estimation error E Shift When the set threshold is exceeded, the calibration recalibration mechanism is triggered.

[0065] It should be noted that when the motion estimation error exceeds the set threshold, the calibration recalibration mechanism is triggered to ensure that the system continues to maintain high efficiency and accuracy. This step ensures that the system can self-optimize in real time, adjust the calibration status in time, and improve overall performance.

[0066] In a possible implementation, collecting the three-dimensional visual data set specifically includes:

[0067] Each robot navigates in its coverage area in a staggered motion, ensuring coverage of non-overlapping areas to maximize comprehensiveness of data collection;

[0068] The robot collects point cloud data synchronously through its 3D vision sensor and uses the following formula to calculate the effective data volume of each robot within its field of view:

[0069]

[0070] Among them, D i is the amount of valid data collected by the ith robot, in units of point clouds, N i is the number of targets detected by the ith robot during data collection, F is the average number of points of each target in the point cloud data, set according to the resolution of the robot sensor, and A i is the area covered by the ith robot in a specific time, and T is the duration of data collection;

[0071] The robot transmits the collected data to the processor in real time via a wireless communication network.

[0072] It should be noted that when collecting three-dimensional visual data sets, an interlaced navigation method is used to ensure that the robot coverage area does not overlap, which can maximize the comprehensiveness of data collection, avoid data blind spots, and improve the richness and diversity of data. The 3D visual sensor synchronously collects point cloud data and calculates the amount of effective data to ensure the accuracy and reliability of the data obtained. At the same time, real-time transmission of data to the processor can speed up data processing and improve system responsiveness, thereby more effectively supporting subsequent posture estimation and fusion processes.

[0073] In a possible implementation, the weight w of the i-th data source is i The calculation method is:

[0074] Collect more than five measurement results of the i-th data source and calculate the mean μ i and standard deviation σ i ;

[0075] Define data confidence C i for:

[0076]

[0077] Among them, ∈ is a constant to avoid the denominator being zero, C i is the confidence of the i-th data source;

[0078] Combined data confidence C i Calculate the weight w i :

[0079]

[0080] Where N is the total number of data sources, C j is the confidence of the j-th data source.

[0081] It should be noted that the weight calculation method measures the i-th data source multiple times and calculates the mean and standard deviation to ensure the stability and accuracy of the collected data. Defining data confidence helps to quantify the reliability of each data source and avoid errors introduced by data fluctuations. Combining confidence calculation weights can effectively reflect the importance of each data source in the data fusion process, ensure the accuracy and credibility of the final attitude estimation results, and thus improve the overall performance and robustness of the system.

[0082] In one possible implementation, the posture estimation model is constructed by the following formula:

[0083] M Pose =β·F Blend +γ·Pred Prev

[0084] Among them, M Pose is the generated pose estimation model, F Blend is the fused pose estimation result, Pred Prev is the result of the last estimation, which is used to introduce historical information. β and γ are adjustment factors to control the impact of historical information on the current estimation.

[0085] In the dynamic Kalman filter, the following state update formula is introduced:

[0086] Xt =F·X t-1 +B·U+K·(ZH·X t-1 )

[0087] Among them, X t is the state vector at the current moment, F is the state transfer matrix, which is dynamically updated in combination with the actual motion model, B is the control input matrix, U is the control vector, and K is the Kalman gain;

[0088] The Kalman gain is calculated as follows:

[0089] K=P t-1 ·H T ·(H·P t-1 ·H T +R) -1

[0090] Among them, P t-1 is the estimated error covariance matrix of the previous moment, reflecting the reliability of state estimation, H T is the transpose of the observation matrix, R is the observation noise covariance matrix, which describes the uncertainty in the observation process;

[0091] In the Kalman filtering process, the state transfer matrix F is dynamically adjusted in combination with the relative positions of each robot to optimize the accuracy of posture estimation. Specifically, an adaptive state transfer model is formed by obtaining feedback data on the position changes of each robot in real time.

[0092] It should be noted that the construction of the posture estimation model integrates historical and current data and uses adjustment factors to control the influence of historical information to enhance the accuracy and robustness of the estimation. The introduction of dynamic Kalman filtering allows the system to update the state in real time according to the actual motion model, thereby improving the responsiveness to motion dynamics. The calculation of the Kalman gain combines the estimation error and observation noise to ensure that the system can still provide reliable state estimation in an uncertain environment. At the same time, by combining the relative positions between robots and dynamically adjusting the state transfer matrix, the posture estimation is made more accurate, which helps to achieve the collaborative operation and efficiency of the multi-robot system.

[0093] In a possible implementation, the regulating factors specifically include:

[0094] The adjustment factors β and γ are used to control the weights of historical information and current data in the fused posture estimation. The calculation method is as follows:

[0095]

[0096] Among them, D CurrentIt is the reliability measure of the current data set, which is defined as the ratio of the number of valid data points in the current data set to the noise level, indicating the signal-to-noise ratio of the current data. Previous It is a reliability measure of the historical data set, which is defined as the ratio of the number of valid data points in the historical data set to the noise level, reflecting the validity of the historical data;

[0097] By dynamically calculating β and γ, the influence of current data is enhanced when the data quality is high, and the weight of historical data is enhanced when the current data quality is low, thereby improving the robustness and accuracy of posture estimation;

[0098] To ensure the validity of β and γ, during the calculation process, both β and γ should satisfy the conditions 0≤β, γ≤1 and β+γ=1.

[0099] It should be noted that the design of the adjustment factor is to optimize the fused posture estimation, and adjust the influence of historical information and current data through dynamic weights. When the quality of the current data is high, its weight in the estimation is increased so that the system can respond quickly to the latest information; when the quality of the current data is low, the weight of the historical data is strengthened to maintain the stability and reliability of the estimation. This flexible weight distribution mechanism not only improves the robustness and accuracy of the posture estimation, but also effectively copes with the fluctuation of data in a dynamic environment, thereby enhancing the collaborative ability of the multi-robot system. At the same time, the effectiveness of the weight calculation is ensured to avoid unreasonable fusion results, providing a guarantee for the overall performance of the system.

[0100] In a possible implementation, a method for determining whether a data point in a data set is valid is as follows:

[0101] Set the signal-to-noise ratio threshold to SNR threshold =15dB, ensuring that only data points with low noise levels are included in the valid data set;

[0102] Set the distance range of valid data points to [d min ,d max ], where d min =0.3m,d max =8.0m, ensuring that the data points are within the effective working distance;

[0103] Data point P i The effective judgment formula is:

[0104]

[0105] Among them, Validity (P i ) is the data point P i The validity of the signal, 1 means valid, 0 means invalid, SNR(P i) is the data point P i The signal-to-noise ratio, d(P i ) is the data point P i The distance to the robot;

[0106] Calculate the mean μ and standard deviation σ of all valid data points in the data set, and set the outlier threshold to Threshold outlier =μ±1.5σ, compare the value of data point Pi:

[0107] If P i <μ-1.5σor P i >+1.5σ, then Validity(P i )=0

[0108] A data point is considered valid only if it meets the validity criteria, otherwise it is marked as an invalid data point.

[0109] It should be noted that the validity judgment method of data points is designed to ensure that the selected data can truly reflect the state of the environment, thereby improving the accuracy of posture estimation. By setting the signal-to-noise ratio threshold, the system ensures that data is only adopted when the noise level is low, thereby filtering out noise data that may affect the results. At the same time, the distance range of valid data points is set to ensure that the collected data is within the operating distance of the robot, further improving the relevance of the data. The validity judgment formula combines the signal-to-noise ratio and distance to form a comprehensive standard to ensure that only qualified data points are included in the valid data set. Finally, by calculating the mean and standard deviation and setting the outlier threshold, the outliers in the data set can be effectively identified, ensuring the final data quality and providing a solid foundation for subsequent posture estimation and fusion. This method not only improves the reliability of the data, but also lays the foundation for the overall performance optimization of the system.

[0110] In a possible implementation, the weight factor further includes:

[0111] The weight factor α is used to quantify the contribution of each valid data source to the fusion result, reflecting its importance in the calibration process. The calculation steps are as follows:

[0112] For each valid data source, calculate the measurement error ∈ i , calculate the weight factor α based on the measurement error i :

[0113]

[0114] Among them, α i is the measurement error of the ith data source, reflecting the accuracy of the data source when collecting data, ∈ iis the weight factor of the i-th data source. The smaller the measurement error, the larger the weight factor.

[0115] Normalize all weight factors:

[0116]

[0117] Among them, α' i is the normalized weight factor, It is the sum of weight factors of all valid data sources. Ensure that the sum of all weight factors is 1 to facilitate relative comparison.

[0118] It should be noted that the weight factor is introduced to quantify the contribution of each valid data source in the fusion result, thereby improving the accuracy and robustness of the system. First, by calculating the measurement error of each data source, its reliability in the data collection process can be directly evaluated. The smaller the measurement error, the higher the accuracy of the data source, so the data source should be given a greater weight in the fusion process. This calculation method ensures that high-precision data sources are given priority during fusion, thereby improving the accuracy of the overall posture estimation.

[0119] Secondly, all weight factors are normalized to make the contributions of different data sources comparable. The sum of the normalized weight factors is 1, so that the weights of each data source can be compared within a uniform range, which can effectively avoid instability caused by large weight differences. In this way, the system can dynamically adjust the importance of each data source in the fusion process according to its actual performance, thereby enhancing the system's adaptability to different environments and conditions, and ultimately achieving more accurate posture estimation.

[0120] In a possible implementation, the calibration recalibration mechanism specifically includes:

[0121] Reconfigure the robot to collect data in the collaborative work area to ensure that the latest 3D visual data set is obtained. During the collection process, the number of data points should be increased to improve the representativeness and accuracy of the data;

[0122] After new data is collected, the weight factor of each data source is recalculated and normalized to ensure that the updated weight reflects the validity of the current data;

[0123] Re-execute data fusion using the updated weight factors and the newly collected data to generate new fusion results, and optimize the motion estimation model based on the new fusion results to ensure that it more accurately reflects the robot posture;

[0124] The new estimation result is compared with the actual motion value to verify the effectiveness of the calibration recalibration. If the error still exceeds the threshold, the above steps can be repeated until the preset accuracy requirement is met.

[0125] It should be noted that the calibration recalibration mechanism is designed to ensure that the robot's posture estimation in complex environments remains highly accurate and reliable. First, the robot is reconfigured for data collection in order to obtain the latest 3D visual data set. By increasing the number of data points, the representativeness of the data can be improved and the estimation error caused by insufficient samples can be reduced. This step ensures that the system can reflect changes in the environment in a timely manner, thereby improving the accuracy of the overall data.

[0126] Next, the weight factors of each data source are recalculated and normalized to ensure that the updated weights can truly reflect the validity of the new data. This process helps to dynamically adjust the importance of different data sources in the fusion process and further improve the reliability of the fusion results. Using these updated weight factors to re-fuse with the newly collected data to generate new posture estimation results is to optimize the motion estimation model so that it can more accurately reflect the current posture of the robot.

[0127] Finally, the effectiveness of the calibration recalibration is verified by comparing the new estimated results with the actual motion values. This step not only confirms whether the updated model meets the accuracy requirements, but also provides a feedback mechanism. If the error exceeds the preset threshold, the above steps can be repeated to ensure that the system is always in the best performance state. This dynamic calibration recalibration mechanism greatly enhances the adaptability and accuracy of the system.

[0128] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0129] (1) In the present invention, through the multi-robot collaborative working mechanism, at least three robots are used to collect three-dimensional visual data in real time, which effectively improves the accuracy of posture perception. This collaborative mode not only enhances the comprehensiveness of the data, but also optimizes the diversity of data sources, making posture estimation in complex environments more reliable;

[0130] (2) In the present invention, the state transfer matrix is ​​dynamically adjusted by combining the relative position changes of each robot through the dynamic Kalman filter algorithm, which significantly reduces the complexity of data processing and enhances the real-time performance of the system. This technical means enables the system to respond to environmental changes in real time and quickly adjust the motion strategy, thereby improving the efficiency of data collection;

[0131] (3) In the present invention, through the calibration recalibration mechanism, the system can monitor the motion estimation error in real time and automatically trigger the recalibration process when the error exceeds the set threshold. This mechanism ensures that the robot can continue to maintain high-efficiency calibration accuracy during long-term operation, avoiding performance degradation due to environmental changes or equipment aging, thereby improving the stability and reliability of the system.

[0132] Reference Manual Attached Figure 2 , showing a structural schematic diagram of a multi-robot 3D vision collaborative spatial posture perception and calibration system provided by an embodiment of the present invention.

[0133] The present invention also provides a multi-robot 3D vision collaborative space posture perception calibration system, which is applied to the multi-robot 3D vision collaborative space posture perception calibration method, comprising:

[0134] Multi-robot configuration, with at least three autonomous mobile robots equipped with high-precision 3D vision sensors that can collect three-dimensional visual data in real time within the collaborative work area;

[0135] The data processing unit is equipped with a centralized processor, which is responsible for receiving the three-dimensional data sets collected from each robot and performing data fusion and motion estimation;

[0136] Wireless communication module, configuring wireless communication network to realize real-time data transmission between robots and between robots and data processing units, so as to ensure the timeliness and accuracy of data sharing;

[0137] The calibration precision factor calculation module is used to dynamically calculate the calibration precision factor to quantify the error range of attitude estimation and to optimize the data fusion process;

[0138] The weight calculation module calculates the weight factor based on the confidence of each data source to reflect the importance of each data source in the fusion process and improve the accuracy of the fusion result;

[0139] The calibration recalibration module monitors and analyzes the motion estimation error. When the error exceeds the set threshold, the recalibration mechanism is automatically triggered to update the robot calibration status.

[0140] The user interaction interface provides a visual operation interface, allowing users to monitor data acquisition, fusion and calibration status in real time and adjust related parameters.

[0141] The working mechanism of the system includes:

[0142] Dynamic data collection: each robot navigates in its coverage area in a staggered motion to ensure comprehensive and non-overlapping data collection. Each robot synchronously collects point cloud data through the 3D vision sensor it carries and transmits it to the data processing unit in real time via a wireless network.

[0143] Real-time data fusion: Through the dynamic Kalman filter algorithm, the data collected from each robot is integrated to generate a comprehensive posture estimation model, and the motion strategy of each robot is adjusted in real time to optimize the data collection quality;

[0144] Adaptive adjustment of weight factors: During the fusion process, the weight factors of each data source are dynamically calculated and adjusted according to the quality and confidence of the data collected in real time to improve the robustness and accuracy of posture estimation;

[0145] Calibration and recalibration process,In the calibration and recalibration mechanism, the system will re-evaluate the weight of each data source based on the latest collected 3D visual data, and re-execute data fusion to ensure the accuracy of the robot's posture.

[0146] The multi-robot 3D vision collaborative space posture perception calibration system provided by the present invention can execute the multi-robot 3D vision collaborative space posture perception calibration method and achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0147] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0148] (1) In the present invention, through the multi-robot collaborative working mechanism, at least three robots are used to collect three-dimensional visual data in real time, which effectively improves the accuracy of posture perception. This collaborative mode not only enhances the comprehensiveness of the data, but also optimizes the diversity of data sources, making posture estimation in complex environments more reliable;

[0149] (2) In the present invention, the state transfer matrix is ​​dynamically adjusted by combining the relative position changes of each robot through the dynamic Kalman filter algorithm, which significantly reduces the complexity of data processing and enhances the real-time performance of the system. This technical means enables the system to respond to environmental changes in real time and quickly adjust the motion strategy, thereby improving the efficiency of data collection;

[0150] (3) In the present invention, through the calibration recalibration mechanism, the system can monitor the motion estimation error in real time and automatically trigger the recalibration process when the error exceeds the set threshold. This mechanism ensures that the robot can continue to maintain high-efficiency calibration accuracy during long-term operation, avoiding performance degradation due to environmental changes or equipment aging, thereby improving the stability and reliability of the system.

[0151] The above contents are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0152] There are a few points to note:

[0153] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention, and other structures may refer to the general design.

[0154] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0155] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0156] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A multi-robot 3D vision collaborative spatial posture perception calibration method, characterized in that: include: S1. In the collaborative work area, use at least three robots to collect a 3D visual data set D = {d1, d2, ..., d n }, each data point d i Represents three-dimensional coordinate information; S2. Set the calibration accuracy factor C Align , used to quantify the error range of attitude estimation, the calibration accuracy factor C Align The calculation formula is: Where n represents the number of collected 3D data points, m represents the dimension of each data point, and d ij represents the measurement value of the i-th data point in the j-th dimension, d′ ij Represents the true coordinates of the i-th data point in the j-th dimension; S3, preprocessing the collected three-dimensional data, including noise reduction and data normalization, to obtain a unified data format D′={d′1, d′2, ..., d′ n }; S4. Data fusion is performed through the following formula: Among them, F Blend is the fused pose estimation result, w i represents the weight of the i-th data source, reflecting its importance in fusion, d′ i represents the i-th data point after preprocessing; S5, generating a fused attitude estimation model and applying dynamic Kalman filtering to optimize the data fusion process; S6. Set motion estimation error E Shift , the motion estimation error E Shift The calculation method is: Among them, E Shift represents the estimated motion error, represents the estimated motion value of the jth dimension, represents the true motion value of the jth dimension, α represents the weight factor, reflecting the impact of calibration accuracy on the error; S7, when the motion estimation error E Shift When the set threshold is exceeded, the calibration recalibration mechanism is triggered.

2. A multi-robot 3D vision collaborative spatial posture perception calibration method according to claim 1, characterized in that: The collecting of the three-dimensional visual data set specifically includes: Each robot navigates in its coverage area in a staggered motion, ensuring coverage of non-overlapping areas to maximize comprehensiveness of data collection; The robot collects point cloud data synchronously through its 3D vision sensor and uses the following formula to calculate the effective data volume of each robot within its field of view: Among them, D i is the amount of valid data collected by the ith robot, in units of point clouds, N i is the number of targets detected by the ith robot during data collection, F is the average number of points of each target in the point cloud data, and is set according to the resolution of the robot sensor. i is the area covered by the ith robot in a specific time, and T is the duration of data collection; The robot transmits the collected data to the processor in real time via a wireless communication network.

3. A multi-robot 3D vision collaborative spatial posture perception calibration method according to claim 2, characterized in that: Further including: The weight w of the i-th data source i The calculation method is: Collect more than five measurement results of the i-th data source and calculate the mean μ i and standard deviation σ i ; Define data confidence C i for: Among them, ∈ is a constant to avoid the denominator being zero, C i is the confidence of the i-th data source; Combined with the data confidence C i Calculate the weight w i : Where N is the total number of data sources, C j is the confidence of the j-th data source.

4. The method for multi-robot 3D vision collaborative spatial posture perception calibration according to claim 1, characterized in that: include: The posture estimation model is constructed by the following formula: M Pose =β·F Blend +γ·Pred Prev Among them, M Pose is the generated pose estimation model, F Blend is the fused pose estimation result, Pred Prev is the result of the last estimation, which is used to introduce historical information. β and γ are adjustment factors to control the impact of historical information on the current estimation. In the dynamic Kalman filter, the following state update formula is introduced: X t =F·X t-1 +B·U+K·(Z-H·X t-1 ) Among them, X t is the state vector at the current moment, F is the state transfer matrix, which is dynamically updated in combination with the actual motion model, B is the control input matrix, U is the control vector, and K is the Kalman gain; The Kalman gain is calculated by the following formula: K=P t-1 ·H T ·(H·P t-1 ·H T +R) -1 Among them, P t-1 is the estimated error covariance matrix of the previous moment, reflecting the reliability of state estimation, H T is the transpose of the observation matrix, R is the observation noise covariance matrix, which describes the uncertainty in the observation process; In the Kalman filtering process, the state transfer matrix F is dynamically adjusted in combination with the relative positions of each robot to optimize the accuracy of posture estimation. Specifically, an adaptive state transfer model is formed by obtaining feedback data on the position changes of each robot in real time.

5. A multi-robot 3D vision collaborative spatial posture perception calibration method according to claim 4, characterized in that: The regulating factors specifically include: The adjustment factors β and γ are used to control the weights of historical information and current data in the fused posture estimation, and the calculation method is as follows: Among them, D Current It is the reliability measure of the current data set, which is defined as the ratio of the number of valid data points in the current data set to the noise level, indicating the signal-to-noise ratio of the current data. Previous It is a reliability measure of the historical data set, which is defined as the ratio of the number of valid data points in the historical data set to the noise level, reflecting the validity of the historical data; By dynamically calculating β and γ, the influence of current data is enhanced when the data quality is high, and the weight of historical data is enhanced when the current data quality is low, thereby improving the robustness and accuracy of posture estimation; To ensure the validity of β and γ, during the calculation process, both β and γ should satisfy the conditions 0≤β, γ≤1 and β+γ=1.

6. A multi-robot 3D vision collaborative spatial posture perception calibration method according to claim 5, characterized in that: Further including: The method to determine whether the data points in the data set are valid is as follows: Set the signal-to-noise ratio threshold to SNR threshold =15dB, ensuring that only data points with low noise levels are included in the valid data set; Set the distance range of valid data points to [d min , d max ], where d min =0.3m,d max =8.0m, ensuring that the data points are within the effective working distance; Data point P i The effective judgment formula is: Among them, Validity (P i ) is the data point P i The validity of the signal, 1 means valid, 0 means invalid, SNR(P i ) is the data point P i The signal-to-noise ratio, d(P i ) is the data point P i The distance to the robot; Calculate the mean μ and standard deviation σ of all valid data points in the data set, and set the outlier threshold to Threshold outlier =μ±1.5σ, compared with data point P i Values: If P i <μ-1.5σor P i >+1.5σ,then Validity(P i )=0 A data point is considered valid only if it meets the validity criteria, otherwise it is marked as an invalid data point.

7. The method for multi-robot 3D vision collaborative spatial posture perception calibration according to claim 1, characterized in that: The weight factors further include: The weight factor α is used to quantify the contribution of each valid data source to the fusion result, reflecting its importance in the calibration process. The calculation steps are as follows: For each valid data source, calculate the measurement error ∈ i , calculate the weight factor α based on the measurement error i : Among them, α i is the measurement error of the ith data source, reflecting the accuracy of the data source when collecting data, ∈ i is the weight factor of the i-th data source. The smaller the measurement error, the larger the weight factor. Normalize all weight factors: Among them, α′ i is the normalized weight factor, It is the sum of weight factors of all valid data sources. Ensure that the sum of all weight factors is 1 to facilitate relative comparison.

8. The method for multi-robot 3D vision collaborative spatial posture perception calibration according to claim 1, characterized in that: The calibration recalibration mechanism specifically includes: Reconfigure the robot to collect data in the collaborative work area to ensure that the latest 3D visual data set is obtained. During the collection process, the number of data points should be increased to improve the representativeness and accuracy of the data; After new data is collected, the weight factor of each data source is recalculated and normalized to ensure that the updated weight reflects the validity of the current data; Re-execute data fusion using the updated weight factors and the newly collected data to generate new fusion results, and optimize the motion estimation model based on the new fusion results to ensure that it more accurately reflects the robot posture; The new estimation result is compared with the actual motion value to verify the effectiveness of the calibration recalibration. If the error still exceeds the threshold, the above steps can be repeated until the preset accuracy requirement is met.

9. A multi-robot 3D vision collaborative spatial posture perception calibration system, characterized in that: include: Multi-robot configuration, with at least three autonomous mobile robots equipped with high-precision 3D vision sensors that can collect three-dimensional visual data in real time within the collaborative work area; The data processing unit is equipped with a centralized processor, which is responsible for receiving the three-dimensional data sets collected from each robot and performing data fusion and motion estimation; Wireless communication module, configuring wireless communication network to realize real-time data transmission between robots and between robots and data processing units, so as to ensure the timeliness and accuracy of data sharing; The calibration precision factor calculation module is used to dynamically calculate the calibration precision factor to quantify the error range of attitude estimation and to optimize the data fusion process; The weight calculation module calculates the weight factor based on the confidence of each data source to reflect the importance of each data source in the fusion process and improve the accuracy of the fusion result; The calibration recalibration module monitors and analyzes the motion estimation error. When the error exceeds the set threshold, the recalibration mechanism is automatically triggered to update the robot calibration status. The user interaction interface provides a visual operation interface, allowing users to monitor data acquisition, fusion and calibration status in real time and adjust related parameters.

10. The multi-robot 3D vision collaborative spatial posture perception calibration system according to claim 9, characterized in that: The working mechanism of the system includes: Dynamic data collection: each robot navigates in its coverage area in a staggered motion to ensure comprehensive and non-overlapping data collection. Each robot synchronously collects point cloud data through the 3D vision sensor it carries and transmits it to the data processing unit in real time via a wireless network. Real-time data fusion: Through the dynamic Kalman filter algorithm, the data collected from each robot is integrated to generate a comprehensive posture estimation model, and the motion strategy of each robot is adjusted in real time to optimize the data collection quality; Adaptive adjustment of weight factors: During the fusion process, the weight factors of each data source are dynamically calculated and adjusted according to the quality and confidence of the data collected in real time to improve the robustness and accuracy of posture estimation; Calibration and recalibration process,In the calibration and recalibration mechanism, the system will re-evaluate the weight of each data source based on the latest collected 3D visual data, and re-execute data fusion to ensure the accuracy of the robot's posture.

Images