CN108960738B

CN108960738B - Laser radar data clustering method under warehouse channel environment

Info

Publication number: CN108960738B
Application number: CN201810784967.1A
Authority: CN
Inventors: 赵敏; 孙棣华; 秦浩
Original assignee: Chongqing University
Current assignee: Chongqing University
Priority date: 2018-07-17
Filing date: 2018-07-17
Publication date: 2022-02-15
Anticipated expiration: 2038-07-17
Also published as: CN108960738A

Abstract

The invention discloses a laser radar data clustering method under a warehouse channel environment, which comprises the following steps: 1. initializing core point number m and neighborhood size sigma; 2. marking all data points in the set of measurement data points as an unvisited state; 3. selecting a data point p with an unvisited state according to the sequence, and marking the state of the point as visited; 4. calculating whether m core points are contained in the neighborhood sigma of the data point p, if so, turning to the next step, otherwise, marking the data point p as a noise point, and turning to the step 2; 5. setting the size of the threshold value sigma to be twice of the return value of the radar divergence model; 6. judging whether the next data point q in the sequence is in the neighborhood of the data point p, if not, turning to the step 2, otherwise, marking the data point q as visited, if the data point q is a core point, namely the neighborhood of the data q contains m data points, updating the threshold, adding the data point q into the cluster C, and turning to the step 5.

Description

Laser radar data clustering method under warehouse channel environment

Technical Field

The invention belongs to the field of environment modeling of mobile robots, and particularly discloses a laser radar data clustering method in a warehouse channel environment.

Background

The warehouse channel environment with the goods bags placed on two sides is a common scene in a warehouse. In the environment, the basis of robot positioning and navigation is to establish an environment model of a channel scene, a goods package is used as a necessary component in the environment model, and the accuracy of the model directly influences the accuracy of the environment model. The clustering algorithm is used as a primary step for extracting the goods package model, and the accuracy of model extraction is directly influenced by the clustering result. Therefore, the method has important significance in accurately clustering the laser radar data in the warehouse channel environment.

There is a fixed drift angle between the measured data points of the lidar, but the actual distance between the data points also depends on the distance between the sensor and the obstacle. In the warehouse passage environment, the laser radar has different measurement distances for different measurement angle distances of the same obstacle, which also causes different dispersion degrees of the laser radar data points. The different degrees of scatter of the data points of the same type cause clustering to become difficult to achieve in the warehouse aisle environment.

In the existing clustering algorithm research, the density clustering represented by K-means clustering and DBSCAN is most widely applied. The K-means clustering is most widely applied, but the incidence relation between adjacent data points of the laser radar is ignored by the K-means clustering algorithm, and the clustering number needs to be specified in advance by the algorithm. The DBCSCAN algorithm partitions data points using a fixed threshold, and this clustering approach is prone to erroneous partitioning of data point partitions due to the characteristics of the warehouse channel environment.

In summary, the current clustering algorithm is difficult to be applied to data clustering of the laser radar in the channel environment due to lack of consideration on the channel environment and the characteristics of the laser radar. Therefore, in order to overcome the defects of the current algorithm in the warehouse channel environment, it is necessary to provide a laser radar data clustering algorithm for the warehouse channel.

Disclosure of Invention

In view of the above, in order to solve the above problems, the present invention provides a laser radar data clustering method in a warehouse channel environment. The method is easy to implement, strong in adaptability and capable of stably and accurately clustering the laser radar data points in the channel environment.

In order to achieve the above and other objects, the present invention provides a method for clustering lidar data in a warehouse channel environment, comprising the steps of:

step 1, initializing core point number m and neighborhood size sigma;

step 2, the measured data point set P is { P ═ P₁,p₂,...,p_nSorting according to the measuring angles from 0 degree to 180 degrees or the order of the angles from small to large, and marking all data points in the measuring data point set as an unaccessed state;

step 3, one data point p with an unvisited state is selected in sequence, the state of the point is marked as visited, if no data point with the unvisited state exists in the measured data point set at the moment, the algorithm is terminated, and clustering is completed;

step 4, calculating whether m core points are contained in the neighborhood sigma of the data point p, if so, turning to the next step, otherwise, marking the data point p as a noise point, and turning to the step 2;

step 5, creating a cluster C of data points p, adding the data points p into the cluster C, calculating the size of a current threshold according to a radar divergence model, and setting the size of the threshold sigma to be twice of a return value of the radar divergence model, wherein sigma is 2 · f (phi, theta);

and 6, judging whether the next data point q in the sequence is in the neighborhood of the data point p, if not, turning to the step 2, otherwise, marking the data point q as visited, if the data point q is a core point, namely m data points are contained in the neighborhood of the data point q, updating a threshold, wherein the new threshold is twice of the return value of the laser radar divergence model at the position q, namely sigma is 2 · f (phi, theta), and in addition, adding the data q into the cluster C, and turning to the step 5.

Preferably, the first and second electrodes are formed of a metal,

φ_iis the angle phi between the ith beam of laser radar and the horizontal direction_i+1The angle between the (i + 1) th beam of laser radar and the horizontal direction is shown, theta is the included angle between the barrier and the vertical direction, and d is the distance between the horizontal line at the position where the laser radar is located and the intersection point of the barrier.

Due to the adoption of the technical scheme, the invention has the following advantages:

the method provides a laser radar measurement divergence model according to a laser radar principle and channel environment characteristics;

the invention provides a clustering method of self-adaptive threshold values by dynamically determining clustering division threshold values according to a measurement divergence model.

The clustering method provided by the invention has higher operation efficiency in a warehouse channel environment, and improves the accuracy of a clustering algorithm result.

Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the means of the instrumentalities and combinations particularly pointed out hereinafter.

Drawings

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings:

FIG. 1 is a divergence model schematic diagram 1;

FIG. 2 is a divergence model diagram of FIG. 2;

FIG. 3 is a flow chart of the present invention.

Detailed Description

The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention.

It should be noted that the drawings provided in the present embodiment are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.

The invention provides a laser radar data clustering method under a warehouse channel environment, which comprises the following specific steps:

firstly, a laser radar divergence measurement model is established, as shown in fig. 1 and 2, XOY represents a current global coordinate system, in the XOY coordinate system, the position of a laser radar is assumed to be C, an obstacle is AB, an included angle between the obstacle and the vertical direction is theta, and an angle between the ith beam of light of the laser radar and the horizontal direction is theta

Angle between the (i + 1) th beam and horizontal directionIs composed of

The horizontal line of the position of the laser radar intersects with the barrier at a point B, the length of BC is recorded as d,

according to fig. 1, the currently measured distances CA and BA may be described as follows.

The distance of adjacent laser beams on an obstacle, i.e., the divergence between laser ranging points, can be described as follows.

Similarly, the divergence model under the scenario shown in fig. 2 can also be described, and by combining the two scenarios, the radar divergence model can be described as follows:

the clustering method based on the measurement divergence variable threshold comprises the following steps:

recording the data set collected by the laser radar as Z ═ Z₁,z₂,...,z_nAnd each datum corresponds to a datum of one direction of the laser radar, and the datum represents the measured distance in the direction. Converting the distance data into a data point set under a rectangular coordinate system, and recording the data point set as P ═ P₁,p₂,...,p_n}. The constant m is set to the minimum number of points around the core point, while σ is defined as the threshold value for each point neighborhood, which is dynamically determined from the divergence model. As shown in fig. 3, the threshold-variable clustering method considering the divergence model includes the following specific steps:

step 1, initializing core point number m and neighborhood size sigma as default values;

step 2, measuring the numberData point set P ═ P₁,p₂,...,p_nSequencing according to the measurement angle from 0 degree to 180 degrees or sequencing the angles from small to large, and marking all data points in the point set as an unaccessed state;

step 3, one data point p with an unvisited state is selected in sequence, the state of the point is marked as visited, if no data point with the unvisited state exists in the time point set, the algorithm is terminated, and clustering is completed;

step 4, calculating whether m points are included in the neighborhood sigma of the measured data point p, if so, turning to the next step, otherwise, marking the data point p as a noise point, and turning to the step 2;

step 5, creating a cluster C of the measurement data points p, adding the measurement data points p into the cluster C, calculating the size of the current threshold according to the measurement divergence model, and setting the size of the threshold sigma to be twice of the model return value, as shown in the following formula:

σ＝2·f(φ,θ)

and 6, judging whether the next data point q in the sequence is in the neighborhood of the data point p, if not, turning to the step 2, otherwise, marking the data point q as visited, if the data point q is a core point, namely the neighborhood of the data point q contains m data points, updating the threshold, adding the data point q into the cluster C, and turning to the step 5.

Finally, the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all of them should be covered in the protection scope of the present invention.

Claims

1. A laser radar data clustering method under a warehouse channel environment is characterized by comprising the following steps:

step 1, initializing core point number m and neighborhood size sigma;

step 6, judging whether the next data point q in the sequence is in the neighborhood of the data point p, if not, turning to the step 2, otherwise, marking the data point q as visited, if the data point q is a core point, namely m data points are contained in the neighborhood of the data point q, updating a threshold, wherein the size of the new threshold is twice of a return value of the laser radar divergence model at the position q, namely sigma is 2 · f (phi, theta), and adding the data point q into a cluster C, and turning to the step 5;