A mobile robot motion tracking control system and control method based on a CLUBS algorithm

Abstract

The application discloses a mobile robot motion tracking control method based on a CLUBS algorithm, which comprises the following steps: S1, acquiring robot gait information, and dividing the whole data space into two types of blocks: one type is that each block contains only one cluster, and the other type is that each block contains only noise; S2, after each division, the Alinski-Harabasz value is recalculated, if the CH value increases, S2 is repeated, if the CH value decreases to below 70%, step S3 is entered; S3, all blocks are incrementally sorted based on density; the blocks containing clusters are separated from all blocks; the separated blocks containing clusters generate an elliptical cluster, S4, if a pair of clusters are merged, a condition that a relatively small increase in the minimum intra-cluster sum of squares is caused; S5, the elliptical cluster is reconstructed, the clustering quality is improved, and the final abnormal point is determined. The application also discloses a mobile robot motion tracking control system based on the CLUBS algorithm, and the method and the system have the functions of robot automatic navigation transportation, human-computer interaction enhancement, and emergency response.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of robot control, and particularly relates to a mobile robot motion tracking control system and method based on a CLUBS algorithm. BACKGROUND

[0002] In recent years, with the continuous development of science and technology, especially the rapid development of robot technology, mobile robots have been applied to more and more fields, greatly facilitating people's life. However, robots may encounter various road conditions and situations when performing tasks, and may easily be stuck or slide and fall, so it is particularly important to track and control the motion process of the mobile robot.

[0003] The mobile robot motion tracking control system based on the CLUB algorithm can accurately track and control the robot motion trajectory, and improve the application performance of the mobile robot. SUMMARY

[0004] In order to overcome the deficiencies in the prior art, the application provides a mobile robot motion tracking control system and method based on a CLUBS algorithm, which aims to solve the technical problems of how to automatically navigate and transport the robot, strengthen human-computer interaction, and respond to emergency events.

[0005] In order to achieve the above purpose, the mobile robot motion tracking control method based on the CLUBS algorithm comprises the following steps,

[0006] S1: acquiring robot gait information, performing binary space partitioning technology, converting the data set into a series of hyper-rectangular blocks, and using a greedy criterion, the blocks are iteratively partitioned into pairs of clusters, and the entire data space is partitioned into two types of blocks: one type contains only one cluster per block, and the other type contains only noise per block;

[0007] S2: after each partitioning, the Calinski-Harabasz value (CH value: a clustering effect index) is recalculated, if the CH value increases, repeat S2), if it decreases to below 70%, go to step S3);

[0008] S3: the main controller detects the density between adjacent blocks, and incrementally sorts all blocks based on the density; based on the principle that the density of the noise-containing block is low, the cluster-containing block is separated from all blocks; the separated cluster-containing block generates an elliptical cluster, if there is an adjacent cluster, go to step S4, if not, go to step S5;

[0009] S4: if merging a pair of clusters results in a small increase in the minimum within-cluster sum of squares (WCSS), and the CH index increases, then merging is implemented; if merging results in a decrease in the CH index, go to S5;

[0010] S5: Reconstructing the ellipse cluster, improving the clustering quality, and determining the final abnormal points.

[0011] Further, in step S1, for a given mobile robot gait dataset, the algorithm starts from a single cluster S corresponding to the entire dataset, S is input to a priority queue Q containing blocks constructed iteratively on the dataset, when Q is not empty, a block B is removed and split into a pair of blocks, if the split is valid, the block pair replaces B in Q, otherwise B becomes the final block of the stage.

[0012] Further, for each dimension i of a d-dimensional block B, we must define distance and functions CiB: R→N and LSi B: R→Rd, the calculation formula is as follows:

[0013]

[0014]

[0015] In the above formula, all values of the point p belonging to the block B are added, and the i-th coordinate value is equal to x. These functions can be represented as mappings, and the minimum WCSS split can be found by linearly scanning these mappings or arrays.

[0016] Further, in step S1, the blocks containing the clusters are separated; because the density of the noise block is low, the CLUBS first sorts all blocks in increasing order based on density, and detects the density jump between adjacent blocks to determine the candidate subset of noise blocks, then, test the candidate subset, generate a hypercube at the center of the block, if the density of the hypercube is less than 2 times the density of the entire block, it indicates that the block belongs to the abnormal block, otherwise it belongs to the cluster block.

[0017] Further, in step S5, the distance formula of each point p=(p1,…,pd) to the cluster center point c=(c1,…,cd) is calculated as:

[0018] where r i is the radius of the i-th dimension cluster, and the commonly used maximum 3σ distance estimation is adopted, then where n is the number of points in the cluster c, and WCSS i (c) is the variance of the i-th dimension;

[0019] Let dist(p,c)=1 define an ellipse, and its center is the cluster center, and the radius of the i-th dimension is r i .

[0020] Also include, a kind of mobile robot motion tracking control system based on CLUBS algorithm, including car body main body, main controller, host computer, drive controller, high-definition 150 It is wide-angle lens, servo steering gear, lithium battery and PC polycarbonate protective shell,

[0021] Host computer includes control panel and operation platform, main controller includes timer, communication interface, control chip, drive controller includes built-in bridge circuit, mobile robot carries out wireless communication with host computer, mobile robot is powered by lithium battery, drive controller is connected servo steering gear, servo steering gear is installed on both sides of car body main body, main controller, drive controller and lithium battery are all placed in the inside of car body main body, high-definition 150 It is wide-angle lens and placed on the upper surface of car body main body, PC polycarbonate protective shell covers on car body main body, main controller, drive controller, high-definition 150 It is wide-angle lens and lithium battery.

[0022] Further, the host computer uses a micro PC and is equipped with an Ubuntu operating system and a ROS operating system.

[0023] Further, the main controller selects an STM32F chip, which has strong performance and meets basic control requirements, and supports external USB expansion.

[0024] Beneficial effects:

[0025] The present application is based on the CLUBS algorithm for robot automatic navigation transportation, strengthens man-machine interaction, and solves the technical problems of sudden events and the like. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 It is one of the structure schematic diagrams of the mobile robot motion tracking control system based on the CLUBS algorithm.

[0027] Figure 2 It is the second structure schematic diagram of the mobile robot motion tracking control system based on the CLUBS algorithm.

[0028] Figure 3 It is a comparison diagram of tracking control accuracy of robot motion tracking control trajectory. DETAILED DESCRIPTION

[0029] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme of the present application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0030] A mobile robot motion tracking control method based on the CLUBS algorithm, comprising the following steps:

[0031] S1: Obtain robot gait information, perform binary space partitioning technology, convert the data set into a series of hyper-rectangular blocks, and use the greedy criterion. The block is iteratively partitioned into pairs of clusters, and the entire data space is partitioned into two types of blocks: one type contains only one cluster per block, and the other type contains only noise per block.

[0032] S2: After each partition, the Calinski-Harabasz value (CH value: a clustering effect index) is recalculated. If the CH value increases, repeat S2), and if it decreases to below 70%, go to step S3).

[0033] S3: The main controller detects the density between adjacent blocks, and sorts all blocks in ascending order based on density. Based on the principle that noise-containing blocks have lower density, separate the cluster-containing blocks from all blocks. Generate an elliptical cluster from the separated cluster-containing blocks. If there are adjacent clusters, go to step S4, and if not, go to step S5.

[0034] S4: If merging a pair of clusters results in a small increase in the minimum within-cluster sum of squares (WCSS), and the CH index increases significantly, then merging is implemented. If merging results in a decrease in the CH index, go to S5.

[0035] S5: Reconstruct the elliptical cluster to improve clustering quality and determine the final outlier points.

[0036] Specifically,

[0037] In step S1: For a given mobile robot gait data set, the algorithm starts from a single cluster S corresponding to the entire data set. S is input into a priority queue Q containing partition blocks iteratively constructed on the data set. When Q is not empty, block B is removed and partitioned into a pair of blocks. If the partition is valid, the block pair replaces B in Q, otherwise B becomes the final block of this stage. To effectively find the optimal partition of the block, edge distribution is used. In particular, for each dimension i of the d-dimensional block B, we must define the distance and function CiB: R→N and LSi B: R→Rd, the calculation formula is as follows:

[0038]

[0039]

[0040] In the above formula, all values of point p belonging to block B are added, and the i-th coordinate value is equal to x. These functions can be represented as mappings, and the minimum WCSS partition is found by performing a linear scan on these mappings or arrays.

[0041] In step S2: the standard effectiveness of the cluster quality and naturalness evaluation using the Calinski-Harabasz index is re-computed after each step. If the CH index increases by splitting, then the splitting is considered effective and the splitting phase continues; otherwise it is stopped.

[0042] 2) The entire robot gait data space is partitioned into two classes of blocks: one class contains only one cluster per block, and the other class contains only noise per block. CLUBS+ seeks to achieve two tasks:

[0043] (a) Separate out the blocks containing clusters; since the density of blocks containing only noise is low, CLUBS first sorts all blocks in increasing order of density and detects density jumps between adjacent blocks to determine a candidate subset of noise blocks. Then, the candidate subset is tested, e.g., a small hypercube is generated at the center of the block, and if the density of the hypercube is less than 2 times the density of the entire block, then the block is considered to belong to the abnormal blocks, otherwise it belongs to the cluster blocks.

[0044] (b) Generate an ellipsoidal cluster from the separated blocks. The distance of each point p = (pi,..., pd) to the cluster center point c = (ci,..., cd) is computed as:

[0045] where r i is the radius of the ith dimension of the cluster, which is estimated using the conventional maximum 3σ distance, then where n is the number of points in the cluster c, and WCSS i (c) is the variance of the ith dimension.

[0046] Let dist(p, c) = 1 define an ellipsoid whose center is then the cluster center and whose ith dimension radius is r i .

[0047] 3) Two clusters are merged whenever their merging improves the overall clustering quality, and the merging process is symmetric to the splitting process. In each step, the pair of clusters whose merging produces the least increase in WCSS is found, and if the merging of these two clusters produces an increase in the CH index, then the agglomerative clustering is completed by replacing the two clusters with their union; if it is determined that the merging of these two clusters produces a decrease in the CH index, then no agglomeration is performed and the entire agglomeration phase is ended, since no other pair of clusters can be merged to improve the CH index.

[0048] 4) Since the approximation and greedy rules used in the previous steps often produce clusters that are slightly irregular in shape, step 2) is often followed by step 4) to improve the quality of the clusters and to identify the final outliers. The algorithm used to perform step 4) is the same as the algorithm used for task (b) of step 2).

[0049] The data clustering quality is improved, outliers formed by uncertain factors such as external environment and self are removed, external disturbance is eliminated, and the purpose of accurately tracking the motion trajectory is finally achieved.

[0050] A mobile robot model is built in Simulink simulation software, and robot parameters are set as shown in the table.

[0051] Table 1 Robot parameters

[0052]

[0053] The motion of the mobile robot is tracked and controlled by using the currently popular control system (a mobile robot trajectory tracking control system based on centroid offset) and the mobile robot motion tracking control system based on the CLUBS+ algorithm of the present application. In the presence of external interference, the tracking control accuracy is tested. The trajectory of the robot motion tracking control is analyzed by using matlab, and the tracking control accuracy is shown in Figure 3

[0054] In this paper, the CLUB+ clustering method is used to optimize and cluster the mobile robot gait big data, improve the data clustering quality, remove outliers formed by uncertain factors such as external environment and self, eliminate external disturbance, and finally achieve the purpose of accurately tracking the motion trajectory.

[0055] It also includes a mobile robot motion tracking control system based on the CLUBS algorithm, which is used in cooperation with the above control method. It includes a vehicle body, a main controller, an upper computer, a drive controller, a high-definition 150° wide-angle lens, a servo steering engine, a lithium battery and a PC polycarbonate protective shell,

[0056] The upper computer includes a control panel and an operation table, and the main controller includes a timer, a communication interface and a control chip. The drive controller includes a built-in bridge circuit. The mobile robot communicates with the upper computer wirelessly, and is powered by a lithium battery. The drive controller is connected to the servo steering engine, which is installed on both sides of the vehicle body. The main controller, the drive controller and the lithium battery are all placed inside the vehicle body. The high-definition 150° wide-angle lens is placed in the middle of the upper surface of the vehicle body, and the PC polycarbonate protective shell covers the vehicle body, the main controller, the drive controller, the high-definition 150° wide-angle lens and the lithium battery.

[0057] The upper computer uses a micro PC and is equipped with Ubuntu and ROS operating systems.

[0058] The main controller uses an STM32F chip, which has strong performance and meets the basic control requirements, and also supports external USB expansion.

[0059] ​The driving controller selects a DBH-01 type double-way driver, the brake function of the driver is stable, and different speeds can be matched, and various control scenes can be used.

[0060] The above is the ideal embodiment according to the application, and the above description can be changed and modified in various ways without departing from the technical idea of the application. The technical scope of the application is not limited to the content of the specification, and must be determined according to the scope of the claims.

Claims

1. A motion tracking control method for a mobile robot based on the CLUBS algorithm, characterized in that, Includes the following steps, S1: Obtain robot gait information, perform binary space segmentation technology to transform the dataset into a series of hyperrectangular blocks, and use a greedy criterion to iteratively segment the blocks into paired clusters. The entire data space is divided into two types of blocks: one type contains only one cluster per block, and the other type contains only noise per block. S2: After each segmentation, the Calinski-Harabasz value is recalculated. If the CH value increases, S2 is repeated. If it decreases to below 70%, proceed to step S3. S3: The main controller detects the density between adjacent blocks and sorts all blocks in ascending order based on the density; based on the principle that the density of noisy blocks is lower, the blocks containing clusters are separated from all blocks; the separated blocks containing clusters are generated into an elliptical cluster. If there are adjacent clusters, proceed to step S4; otherwise, proceed directly to step S5. S4: If merging a pair of clusters results in a small increase in the CH index relative to minimizing the intra-cluster sum of squares, then perform the merge; if merging results in a decrease in the CH index, proceed to S5. S5: Reconstruct the elliptical clusters, improve the clustering quality, and identify the final outliers.

2. The motion tracking control method for a mobile robot based on the CLUBS algorithm according to claim 1, characterized in that, In step S1, for a given mobile robot gait dataset, the algorithm starts with a single cluster S corresponding to the entire dataset. S is input into a priority queue Q containing iteratively constructed partition blocks on the dataset. When Q is not empty, block B is removed and split into a pair of blocks. If the split is valid, the block pair replaces B in Q; otherwise, B becomes the final block of this stage.

3. The motion tracking control method for a mobile robot based on the CLUBS algorithm according to claim 2, characterized in that, For each dimension i of the d-dimensional block B, we must define the distance sum function CiB:R→N and LSiB:R→Rd, calculated as follows: In the above formula, all values ​​of point p belonging to block B are summed, and its i-th coordinate value is equal to x. These functions can be represented as mappings, and the segmentation that minimizes W CSS is found by performing a linear scan of these mappings or arrays.

4. The motion tracking control method for a mobile robot based on the CLUBS algorithm according to claim 1, characterized in that, In step S1, blocks containing clusters are separated. Since the density of blocks containing only noise is low, CLUBS first sorts all blocks in ascending order based on density and detects density jumps between adjacent blocks to determine candidate subsets of noise blocks. Then, the candidate subsets are tested, and a hypercube is generated at the center of the block. If the density of the hypercube is less than twice the density of the entire block, it indicates that the block belongs to an anomalous block; otherwise, it belongs to a cluster block.

5. The motion tracking control method for a mobile robot based on the CLUBS algorithm according to claim 1, characterized in that, In step S5, the formula for calculating the distance from each point p = (p1, ..., pd) to the cluster center point c = (c1, ..., cd) is as follows: Where, r i Let be the radius of the i-th dimensional cluster. Using the commonly used maximum 3σ distance estimation, then... Where n is the number of points in cluster c, WCSS i (c) is the variance of the i-th dimension; Let dist(p,c) = 1 to define an ellipse whose center is the cluster center and whose radius in the i-th dimension is r. i .

6. A motion tracking control system for a mobile robot based on the CLUBS algorithm, used in the control method described in any one of claims 1-5, characterized in that, Includes the vehicle body, main controller, host computer, drive controller, high-definition 150° wide-angle lens, servo motors, lithium battery, and PC polycarbonate protective shell. The upper unit includes a control panel and an operating console. The main controller includes a timer, a communication interface, and a control chip. The drive controller includes a built-in bridge circuit. The mobile robot communicates wirelessly with the host computer. The mobile robot is powered by a lithium battery. The drive controller is connected to servo motors, which are installed on both sides of the main body. The main controller, drive controller, and lithium battery are all located inside the main body. A high-definition 150° wide-angle lens is located in the middle of the upper surface of the main body. A PC polycarbonate protective shell covers the main body, main controller, drive controller, high-definition 150° wide-angle lens, and lithium battery.

7. A motion tracking control system for a mobile robot based on the CLUBS algorithm according to claim 6, characterized in that, The host computer is a micro PC and is equipped with Ubuntu and ROS operating systems.

8. A motion tracking control system for a mobile robot based on the CLUBS algorithm according to claim 6, characterized in that, The main controller uses the STM32F chip, which has strong performance and meets basic control requirements, while also supporting external USB expansion.

Images