Bucket wheel machine anti-collision method and system based on laser radar perception

Abstract

The invention discloses a bucket wheel machine collision prevention method and system based on laser radar perception, and the method comprises the steps: S1, calibrating a laser radar under a stockyard coordinate system; s2, acquiring a three-dimensional point cloud coordinate set around the anti-collision part through a laser radar; s3, judging whether an obstacle exists in the protection area according to the three-dimensional point cloud coordinate set; s4, in response to the existence of the obstacle, calculating the minimum linear distance between the obstacle and the anti-collision part; and S5, converting the minimum linear distance into a switching value signal of the corresponding working mode. According to the scheme, a layer of intelligent anti-collision protection of the bucket wheel machine based on the laser radar is added during operation of the bucket wheel machine, the defect that the anti-collision function cannot be achieved when the distance between any two bucket wheel machines is smaller than a fixed area is overcome, and the autonomous safety of equipment is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of automatic operation collision avoidance of bucket wheel machines, and in particular, relates to a collision avoidance method and system for bucket wheel machines based on laser radar perception. Background Art

[0002] Bucket wheel excavator is a mechanical equipment used for continuous bulk material loading and unloading. It is the core equipment for stacking and reclaiming operations in large bulk cargo yards. In large yards, in order to improve the working efficiency of the yard, multiple bucket wheel excavators are often arranged on the same track. In a material yard where multiple bucket wheel excavators are operating at the same time, there is a certain risk of collision between the two machines. Once a collision occurs, it will cause economic losses and casualties. Therefore, it is of great significance to study the anti-collision method of bucket wheel excavators in large yards.

[0003] Figure 1-3 Three operating states of the two bucket wheel machines 1 are given. It can be seen that the anti-collision of the two bucket wheel machines 1 in the yard mainly needs to solve the following two collision risk problems: the first point is that the front and rear two equipments are prone to collision during operation, and the second point is to prevent the rear machine from colliding with the tail of the front machine when the cantilever rotates.

[0004] The traditional anti-collision method of bucket wheel excavators mainly relies on manual operation and sensors such as trolley encoders and infrared ranging to achieve anti-collision between two machines in a fixed area. However, when the two machines are smaller than the fixed area, they will not be able to operate normally. If the two machines are to work normally when the distance is smaller than the fixed area, the anti-collision function cannot be achieved. Therefore, there are problems such as small protection range of the anti-collision mechanism, low accuracy, and inability to update the anti-collision distance in real time. In addition, the anti-collision method of infrared ranging sensors is to install infrared ranging sensors at the anti-collision position, and take anti-collision measures when obstacles are detected within the distance range. However, since the risk of collision in the stockyard is mainly caused by two or more bucket wheel excavators in operation at the same time, and bucket wheel excavators operating at the same time may have multiple collision points, the use of infrared ranging sensors requires the installation of sensors at all these points where collisions may occur, which is difficult to achieve and prone to blind spots. In addition, infrared ranging sensors cannot distinguish between materials and obstacles, so they cannot achieve anti-collision for the parts of the bucket wheel excavator that need to contact materials. With the development of technology, a more efficient and intelligent anti-collision method for bucket wheel excavators is urgently needed to improve the safety and efficiency of operations. Summary of the invention

[0005] The purpose of the present invention is to provide a bucket wheel excavator collision avoidance method and system based on laser radar perception in order to solve the above-mentioned problems.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A bucket wheel machine collision avoidance method based on laser radar perception, the method comprising:

[0008] S1. Calibrate the laser radar in the stockpile yard coordinate system;

[0009] S2. Obtaining a three-dimensional point cloud coordinate set around the anti-collision part through a laser radar;

[0010] S3. Determine whether there are obstacles in the protection area based on the three-dimensional point cloud coordinate set;

[0011] S4. In response to the presence of an obstacle, calculating the minimum straight-line distance between the obstacle and the anti-collision part;

[0012] S5. Convert the minimum straight-line distance into a switch signal of the corresponding working mode.

[0013] In the above-mentioned bucket wheel machine collision avoidance method based on laser radar perception, in step S1, the laser radar is calibrated in the following manner:

[0014] Establish the stockpile yard coordinate system;

[0015] Obtain the attitude and position data of the bucket wheel machine from the bucket wheel machine PLC to determine the coordinate position of the laser radar in the stockyard coordinate system;

[0016] Obtaining a laser radar coordinate system referenced by the laser radar for collecting three-dimensional point cloud data;

[0017] According to the coordinate position of the laser radar in the stockpile yard coordinate system, the laser radar coordinate system and the stockpile yard coordinate system, the mapping relationship from the three-dimensional point cloud data in the laser radar coordinate system to the three-dimensional point cloud coordinate set in the stockpile yard coordinate system is determined to complete the calibration of the laser radar.

[0018] In the above-mentioned bucket wheel machine collision avoidance method based on laser radar perception, step S3 specifically includes:

[0019] S31. Obtain the coordinate position of the current anti-collision part in the stockpile coordinate system according to the bucket wheel machine posture and position data;

[0020] S32. Delimit the protection area according to the pre-set protection range and the current coordinate position of the anti-collision part;

[0021] S33. Determine whether the number of coordinate points in the three-dimensional point cloud coordinate set within the protection area exceeds a quantity threshold. If so, determine that there is an obstacle.

[0022] In the above-mentioned bucket wheel machine anti-collision method based on laser radar perception, in step S2, the anti-collision parts include the bucket wheel machine cantilever and counterweight, and the laser radar obtains the three-dimensional point cloud data by scanning the bucket wheel machine cantilever and the surrounding configuration, and converts the three-dimensional point cloud data obtained by the laser radar into the stockpile yard coordinate system through the calibration of step S1.

[0023] In the above-mentioned bucket wheel machine collision avoidance method based on laser radar perception, in step S2, after the three-dimensional point cloud data acquired by the laser radar is converted into the stockpile yard coordinate system, the relevant point cloud data located in the working area is screened out from the three-dimensional point cloud data;

[0024] Then, candidate point clouds of non-materials are screened out from relevant point cloud data according to reflectivity information of the three-dimensional point cloud data to obtain the three-dimensional point cloud coordinate set.

[0025] In the above-mentioned bucket wheel machine collision avoidance method based on lidar perception, candidate point clouds are screened out from relevant point cloud data in the following way:

[0026] The point cloud data is divided into two categories: high intensity and low intensity according to the reflection intensity of the reflectivity information;

[0027] Divide the low-intensity point cloud data into a first aggregation group according to the position, and the remaining point clouds constitute a first discrete group;

[0028] Divide the high-intensity point cloud data into a second aggregation group according to the position, and the remaining point clouds constitute a second discrete group;

[0029] Defining a first boundary for the first aggregation group and a second boundary for the second aggregation group;

[0030] The point cloud in the second discrete group that is within the first boundary is classified into the first aggregated group, and the point cloud in the first discrete group that is within the second boundary is classified into the second aggregated group;

[0031] The remaining point clouds of the second aggregated group and the second discrete group that are not classified into the first aggregated group are screened as the candidate point clouds.

[0032] In the above-mentioned bucket wheel machine collision avoidance method based on laser radar perception, the first aggregation group / the second aggregation group are divided from the low-intensity / high-intensity point cloud data in the following manner:

[0033] S21. Select any point from the low-intensity / high-intensity point cloud dataset as the center point;

[0034] S22. The points close to the center point and less than the set distance are classified into the pre-selected aggregation group;

[0035] S23. Taking any point in the pre-selected aggregation group as the center point, continue to classify points close to any center point and less than the set distance into the pre-selected aggregation group;

[0036] S24. Expand outward from the first center point, and repeat step S23 until there is no point in the remaining points of the point cloud data set that is less than the set distance from any center point in the pre-selected aggregation group;

[0037] S25. Repeat steps S21 to S24 for the remaining point cloud data sets of the low-intensity / high-intensity point cloud data sets to obtain more pre-selected aggregation groups until all the point cloud data of the low-intensity / high-intensity are divided;

[0038] S26. The pre-selected aggregation group with the largest number of point clouds at this time is used as the first aggregation group / the second aggregation group.

[0039] In the above-mentioned bucket wheel machine anti-collision method based on laser radar perception, step S4 calculates the minimum straight-line distance between the obstacle and the anti-collision part as follows:

[0040] According to the bucket wheel machine posture and position data, the current coordinates P2 (p 2-1 , p 2-2 …… 2-n ), n represents the number of anti-collision part coordinates;

[0041] Get the three-dimensional point cloud coordinate set P1(p 1-1 , p 1-2 …… 1-m ), m represents the number of three-dimensional point cloud coordinate points in the protection area;

[0042] Get the nearest p by nearest neighbor search 1-i -p 2-j Group;

[0043] Calculate p 1-i -p 2-j The straight-line distance is taken as the minimum straight-line distance between the obstacle and the anti-collision part;

[0044] In step S5, the working mode includes any one or more combinations of a deceleration mode, a stop mode and a normal mode;

[0045] When the minimum straight-line distance is less than the first set distance and is in a continuously decreasing state, the switch signal is switched to the deceleration mode. When it continues to decrease to less than the second set distance, the switch signal is switched to the stop mode.

[0046] When the minimum straight-line distance is in a state of continuous increase, the switch signal switches to the normal mode.

[0047] A bucket wheel machine anti-collision system based on laser radar perception includes a bucket wheel machine PLC and a bucket wheel machine. The bucket wheel machine is equipped with at least two 3D laser radars with a field of view covering the anti-collision area, and the 3D laser radars are connected to an industrial computer. The industrial computer is used to execute the method described in any one of claims 1 to 8 and is connected to the bucket wheel machine PLC so that the bucket wheel machine PLC can switch the working mode of the bucket wheel machine according to a switch signal.

[0048] In the above-mentioned bucket wheel machine anti-collision system based on laser radar perception, the system also includes a self-cleaning component installed at the laser radar for cleaning the laser radar lens, and the self-cleaning component is connected to the industrial computer.

[0049] The advantages of the present invention are:

[0050] 1) This solution adds a layer of intelligent anti-collision protection for bucket wheel machines based on laser radar when bucket wheel machines are operating, which makes up for the deficiency that the anti-collision function cannot be realized when the distance between any two bucket wheel machines is less than a fixed area, and effectively improves the autonomous safety of the equipment;

[0051] 2) This solution first sets the protection area, determines whether there are obstacles in the protection area, and then performs anti-collision calculations to reduce the resource occupation for the normal operation of the bucket wheel excavator;

[0052] 3) This scheme uses reflectivity intensity to screen candidate point clouds that are not materials, and then obtains a three-dimensional point cloud coordinate set for judging whether there is an obstacle. By distinguishing between coal and obstacles, even the parts that need to contact the material can be prevented from collision, and the anti-collision mechanism has a large protection range;

[0053] 4) This scheme proposes a specific strategy to screen out candidate point clouds, and divides the aggregation groups required for screening candidate point clouds in a specific way, which can effectively improve the accuracy of the anti-collision protection mechanism and reduce the impact of the intensity threshold setting on the implementation of the anti-collision protection mechanism;

[0054] 5) This solution determines the minimum distance between the bucket wheel machine and the obstacle, and switches the switch signal according to the minimum distance. The switch signal enables the bucket wheel machine PLC 2 to switch to the corresponding working mode, and can flexibly adjust the mode according to the latest distance to achieve the effect of the electronic limit switch. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is one of the states when two bucket wheel machines are operating;

[0056] Figure 2 This is the second state of two bucket wheel machines in operation;

[0057] Figure 3 This is the third state when two bucket wheel machines are operating;

[0058] Figure 4 The system structure block diagram of the bucket wheel machine anti-collision system based on laser radar perception of the present invention;

[0059] Figure 5 It is a flow chart of the bucket wheel machine anti-collision method based on laser radar perception of the present invention;

[0060] Figure 6The present invention is a flow chart of a laser radar calibration method in a bucket wheel machine collision avoidance method based on laser radar perception;

[0061] Figure 7 A flow chart of a method for determining whether there is an obstacle in a bucket wheel machine anti-collision method based on laser radar perception according to the present invention;

[0062] Figure 8 A flow chart of a method for screening candidate point clouds in a bucket wheel machine anti-collision method based on laser radar perception according to the present invention;

[0063] Fig. 9 The present invention is a flow chart of a method for dividing aggregation groups in the bucket wheel excavator collision avoidance method based on laser radar perception.

[0064] Figure numerals: bucket wheel machine 1; bucket wheel machine PLC 2; industrial computer 3; laser radar 4; self-cleaning component 5; stockpile yard 6. DETAILED DESCRIPTION

[0065] The present invention provides a bucket wheel machine 1 anti-collision system based on laser radar 4 perception, which can avoid the failure of the anti-collision function when the distance between the two bucket wheel machines 1 is less than a fixed area, improve the anti-collision protection range and accuracy, and realize the anti-collision protection of the bucket wheel machine 1 more efficiently. Figure 4 As shown, the system includes a bucket wheel machine PLC 2 and a bucket wheel machine 1, and the bucket wheel machine 1 is controlled by the bucket wheel machine PLC 2. In particular, in this solution, at least two 3D laser radars 4 with a field of view covering the anti-collision part are installed on the bucket wheel machine 1, and the 3D laser radars 4 are connected to the industrial computer 3, so that the industrial computer 3 can obtain and process the three-dimensional point cloud data scanned by the 3D laser radar 4 in real time. The industrial computer 3 is connected to the bucket wheel machine PLC 2 so that the bucket wheel machine PLC 2 can switch the working mode of the bucket wheel machine 1 according to the switch signal. The 3D laser radar 4 is preferably installed on the platforms on the left and right sides of the cantilever head of the bucket wheel machine 1, and the field of view (FOV) of the laser radar 4 can cover the key areas on both sides of the cantilever and counterweight of the bucket wheel machine 1.

[0066] Furthermore, the system also includes a self-cleaning component 5 installed at the laser radar 4 for cleaning the lens of the laser radar 4. The number of self-cleaning components 5 is consistent with the number of laser radars 4. Each laser radar 4 is equipped with a self-cleaning component 5. Each self-cleaning component 5 is connected to the industrial computer 3 and is used to regularly clean the lens of the laser radar 4. The self-cleaning component 5 can use an existing structure, such as a wiper that can remove dust. The specific structure is not limited here. It only needs to be able to wipe the lens of the laser radar 4 regularly or irregularly as needed to ensure the reliability of the anti-collision data. In addition, the lens of the laser radar cover can be made of coated glass. The coated glass increases the light transmittance by adding a layer of anti-reflection film on the glass surface.

[0067] Furthermore, if Figure 5 As shown, the laser collision avoidance method of bucket wheel machine 1 implemented based on the above system includes the following steps:

[0068] S1. Calibrate the laser radar 4 in the stockyard coordinate system;

[0069] S2. Obtain a three-dimensional point cloud coordinate set around the anti-collision part through the laser radar 4;

[0070] S3. Determine whether there are obstacles in the protection area based on the three-dimensional point cloud coordinate set;

[0071] S4. In response to the presence of an obstacle, calculating the minimum straight-line distance between the obstacle and the anti-collision part;

[0072] S5. Convert the minimum straight-line distance into a switch signal of the corresponding working mode.

[0073] Specifically, Figure 6 As shown, in step S1, the laser radar 4 is calibrated in the following manner:

[0074] Establish the stockpile yard coordinate system;

[0075] The attitude and position data of bucket wheel 1 (mainly the position of the trolley, the rotation angle, and the pitch angle) are obtained from bucket wheel PLC 2 to determine the coordinate position of laser radar 4 in the stockyard coordinate system. That is, the position of laser radar 4 is obtained. The installation position of laser radar 4 on bucket wheel 1 is known. According to the position of the trolley, the rotation angle and the pitch angle and the parameters of bucket wheel 1 itself, the position of the key parts of bucket wheel 1 (including the installation part of laser radar 4) in the stockyard coordinate system can be known, and then the installation position of laser radar 4 in the stockyard coordinate system can be known, and even the position of laser radar 4 in the stockyard coordinate system can be known.

[0076] A laser radar 4 coordinate system with the laser radar 4 as a reference is obtained for collecting three-dimensional point cloud data. That is to say, in this coordinate system, for the same point in space, the position of the laser radar 4 changes, and the point cloud coordinate value also changes, while the laser radar 4 and the bucket wheel machine 1 whose relative position to the laser radar 4 remains unchanged remain the same at all times.

[0077] According to the coordinate position of the laser radar 4 in the stockpile coordinate system, the laser radar coordinate system and the stockpile coordinate system, the mapping relationship from the three-dimensional point cloud data in the laser radar coordinate system to the three-dimensional point cloud coordinate set in the stockpile coordinate system is determined to complete the calibration of the laser radar 4. In this way, the point cloud coordinates can be converted to a unified stockpile coordinate system for subsequent obstacle judgment and collision risk judgment, etc.

[0078] Specifically, in step S2, the anti-collision parts include the bucket wheel cantilever and the counterweight. The laser radar 4 obtains the three-dimensional point cloud data by scanning the bucket wheel cantilever and the surrounding configuration, and converts the three-dimensional point cloud data obtained by the laser radar into the stockpile coordinate system through the calibration of step S1 to obtain a three-dimensional point cloud coordinate set.

[0079] like Figure 7 As shown, step S3 specifically includes:

[0080] S31. Obtain the coordinate position of the current anti-collision part in the stockpile coordinate system according to the bucket wheel machine 1 posture and position data;

[0081] S32. Delimit the protection area according to the pre-set protection range and the current coordinate position of the anti-collision part;

[0082] S33. Determine whether the number of coordinate points of the three-dimensional point cloud coordinate set in the protection area exceeds the number threshold. If so, it is determined that there is an obstacle. The number threshold is determined by those skilled in the art based on experience. If it is less than a certain number, it means that either it is a detection error, and it is not considered that there is an obstacle, or only a corner or a small part of the obstacle has entered the protection area, and it is more likely that it just passes through the protection area without a real collision. Therefore, it is not considered that there is an obstacle.

[0083] Only when the number of coordinate points exceeds a certain amount, that is, exceeds the number threshold, it means that a large part of the obstacle is located in the protection area, then it is determined that there is an obstacle, and steps S4 and S5 are continued.

[0084] Specifically, step S4 calculates the minimum straight-line distance between the obstacle and the anti-collision part as follows:

[0085] According to the attitude and position data of the bucket wheel machine 1, the current coordinate P2 (p 2-1 , p 2-2 …… 2-n ), n represents the number of anti-collision part coordinates. There may be multiple anti-collision parts, each of which may have one coordinate or multiple coordinates. n is determined according to the set point cloud density, the size of the anti-collision part, etc.

[0086] Get the three-dimensional point cloud coordinate set P1(p 1-1 , p 1-2 …… 1-m ), m represents the number of three-dimensional point cloud coordinate points in the protection area;

[0087] Get the nearest p by nearest neighbor search 1-i -p 2-j Group, p 1-i represents the i-th coordinate of the anti-collision part, p 2-jRepresents the jth coordinate of the 3D point cloud within the protection area.

[0088] Calculate p 1-i -p 2-j The straight-line distance is taken as the minimum straight-line distance between the obstacle and the anti-collision part;

[0089] In step S5, the working modes include deceleration mode, stop mode and normal mode;

[0090] When the minimum straight-line distance is less than the first set distance (such as 1.5 meters) and is in a continuously decreasing state, such as being in a continuously decreasing state for 30 consecutive seconds, the switch signal is switched to the deceleration mode. After the switch signal is given to the bucket wheel machine PLC 2, the PLC will control the bucket wheel machine 1 to decelerate.

[0091] In this stage, the minimum straight-line distance continues to be monitored. If it continues to decrease to less than the second set distance (1 meter), it will switch to the switch signal of the stop mode. After this switch signal is given to the bucket wheel machine PLC 2, the PLC will control the bucket wheel machine 1 to stop working until the obstacle is removed and the movement in this direction is resumed. Generally, if the minimum straight-line distance is greater than the first set distance, it is considered that the obstacle has been removed. If the minimum straight-line distance turns to continuous increase before reaching the second set distance, the minimum straight-line distance will continue to be monitored. When it is greater than the first set distance, the normal mode will be restored.

[0092] When the minimum straight-line distance is in a state of continuous increase, indicating that there is no risk of collision, the switch signal is switched to the normal mode. After this switch signal is given to the bucket wheel machine PLC 2, the PLC will control the bucket wheel machine 1 to work normally.

[0093] Note that the coordinates mentioned above refer to the coordinates in the stockpile yard coordinate system.

[0094] Furthermore, in step S2, after the three-dimensional point cloud data acquired by the laser radar is converted to the stockyard coordinate system, relevant point cloud data is screened out from the three-dimensional point cloud data according to the working area; then, candidate point clouds of non-materials are screened out from the relevant point cloud data according to the reflectivity information of the three-dimensional point cloud data to obtain the three-dimensional point cloud coordinate set. In other words, the stockyard will have a working area, which is demarcated by the staff according to the actual situation. This working area is an area in the stockyard coordinate system. This process can exclude point cloud data that is not in the working area, so as to facilitate the subsequent screening of candidate point clouds of non-materials from the point clouds in the working area by means of aggregation groups and discrete groups.

[0095] For example, a reflection intensity threshold is set based on experience. When the reflection intensity of the point cloud is greater than the reflection intensity threshold, it is considered to be material - coal. Otherwise, it is considered to be an obstacle. The coal point and the obstacle point are distinguished by the reflectivity intensity, which can ensure the normal operation of the bucket wheel excavator 1 while playing an anti-collision effect.

[0096] Preferably, if Figure 8 As shown, based on the set intensity threshold, this solution further filters out candidate point clouds from the relevant point cloud data in the following manner. The point cloud data mentioned below are all relevant point cloud data filtered out based on the working area in the previous step.

[0097] According to the reflection intensity of the reflectivity information, the relevant point cloud data is divided into high intensity and low intensity; those above the intensity threshold are classified as high intensity, and those below or equal to the intensity threshold are classified as low intensity. Due to the loose and porous characteristics of coal, the intensity of the laser reflected back is usually low. Instead of coal pile obstacles, this scheme is mainly to avoid collisions between two or more bucket wheel machines 1 working at the same time, and bucket wheel machines 1 as man-made mechanical equipment generally have the characteristics of high reflectivity.

[0098] Divide the low-intensity related point cloud data into a first aggregation group according to the position, and the remaining related point cloud data constitute a first discrete group;

[0099] Divide the high-intensity related point cloud data into a second aggregation group according to the position, and the remaining related point cloud data constitute a second discrete group;

[0100] Defining a first boundary for the first aggregation group and a second boundary for the second aggregation group;

[0101] The point cloud in the second discrete group that is within the first boundary is classified into the first aggregated group, and the point cloud in the first discrete group that is within the second boundary is classified into the second aggregated group;

[0102] The remaining point clouds of the second aggregated group and the second discrete group that are not classified into the first aggregated group are screened as the candidate point clouds. If there is no first discrete group and no second discrete group, the candidate point cloud can be directly screened out. This candidate point cloud is the three-dimensional point cloud coordinate set obtained in step S2.

[0103] In this way, a more accurate distinction between coal and obstacles can be achieved, thereby achieving a more accurate anti-collision effect of the bucket wheel excavator 1.

[0104] Furthermore, if Fig. 9 As shown, this scheme divides the first aggregation group from the low-intensity point cloud data in the following way:

[0105] S21. Select any point from the low-intensity point cloud data set as the center point;

[0106] S22. The points close to the center point and less than the set distance are classified into the pre-selected aggregation group;

[0107] S23. Taking any point in the pre-selected aggregation group as the center point, continue to classify points close to any center point and less than the set distance into the pre-selected aggregation group;

[0108] S24. In this way, expanding outward from the first center point, repeating step S23 until there is no point in the remaining points of the point cloud data set that is less than the set distance from any center point in the pre-selected aggregation group, indicating that the point cloud of this pre-selected aggregation group has been included;

[0109] S25. Repeat steps S21 to S24 for the remaining point cloud data sets of the low-intensity point cloud data set to obtain more pre-selected aggregation groups until all the point cloud data of the low-intensity are divided;

[0110] S26. The pre-selected aggregation group with the largest number of point clouds at this time is used as the first aggregation group.

[0111] The method of dividing the second aggregation group from the high-intensity point cloud data is similar to the above method and will not be described here.

[0112] Most coal piles have low-intensity reflectivity, and most obstacles have high-intensity reflectivity. In the actual stockpile, coal piles and obstacles (the remaining bucket wheel machine 1) have obvious partition characteristics. In the low-intensity reflectivity data set, most of them are coal points, while in the high-intensity reflectivity data set, most of them are obstacle points. The above-mentioned division of aggregation groups and the screening of candidate point clouds can overcome the problem of large coal obstacle recognition errors caused by improper intensity threshold setting.

[0113] In another embodiment, the first aggregation group / the second aggregation group may be divided from the low-intensity / high-intensity point cloud data in the following manner. This embodiment takes the division of the second aggregation group from the high-intensity point cloud data as an example:

[0114] If the number of point clouds of the first pre-selected aggregation group obtained exceeds half of the high-intensity point cloud data set, then this pre-selected aggregation group can be directly used as the second aggregation group for the final division without further division and comparison. On the contrary, continue to divide the second pre-selected aggregation group, starting from the second pre-selected aggregation group, compare with the pre-selected aggregation group with the largest number of point clouds previously determined to determine the latest pre-selected aggregation group with the largest number of point clouds, and directly compare the second pre-selected aggregation group with the first pre-selected aggregation group to select the larger number of pre-selected aggregation groups. And each time a pre-selected aggregation group is divided, the pre-selected aggregation group with the largest number of point clouds at this time is compared with the latest remaining number of point clouds. If it is more than the remaining number of point clouds, there is no need to continue to divide and compare, and it is used as the second aggregation group for the final division. In this way, the efficiency of the division of the aggregation group can be effectively improved, especially when the point clouds of the discrete groups are more dispersed, the efficiency can be significantly improved.

[0115] Similarly, dividing the first aggregation group from the low-intensity point cloud data is similar and will not be described here.

[0116] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

[0117] Although the terms such as bucket wheel machine 11, bucket wheel machine PLC 2, industrial computer 3, 3D laser radar 4, and self-cleaning component 5 are used more frequently in this article, the possibility of using other terms is not excluded. The use of these terms is only for more convenient description and explanation of the essence of the present invention; interpreting them as any additional limitation is contrary to the spirit of the present invention.

Claims

1. A bucket wheel machine collision avoidance method based on laser radar perception, characterized in that: The method includes: S1. Calibrate the laser radar in the stockpile yard coordinate system; S2. Obtaining a three-dimensional point cloud coordinate set around the anti-collision part through a laser radar; S3. Determine whether there are obstacles in the protection area based on the three-dimensional point cloud coordinate set; S4. In response to the presence of an obstacle, calculating the minimum straight-line distance between the obstacle and the anti-collision part; S5. Convert the minimum straight-line distance into a switch signal of the corresponding working mode.

2. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 1 is characterized in that: In step S1, the laser radar is calibrated in the following manner: Establish the stockpile yard coordinate system; Obtain the attitude and position data of the bucket wheel machine from the bucket wheel machine PLC to determine the coordinate position of the laser radar in the stockyard coordinate system; Obtaining a laser radar coordinate system referenced by the laser radar for collecting three-dimensional point cloud data; According to the coordinate position of the laser radar in the stockpile yard coordinate system, the laser radar coordinate system and the stockpile yard coordinate system, the mapping relationship from the three-dimensional point cloud data in the laser radar coordinate system to the three-dimensional point cloud coordinate set in the stockpile yard coordinate system is determined to complete the calibration of the laser radar.

3. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 2 is characterized in that: Step S3 specifically includes: S31. Obtain the coordinate position of the current anti-collision part in the stockpile coordinate system according to the bucket wheel machine posture and position data; S32. Delineate the protection area according to the current coordinate position of the protection range and the anti-collision part; S33. Determine whether the number of coordinate points in the three-dimensional point cloud coordinate set within the protection area exceeds a quantity threshold. If so, determine that there is an obstacle.

4. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 3 is characterized in that: In step S2, the anti-collision parts include the bucket wheel machine cantilever and counterweight. The laser radar obtains the three-dimensional point cloud data by scanning the bucket wheel machine cantilever and the surrounding configuration, and converts the three-dimensional point cloud data obtained by the laser radar into the stockpile yard coordinate system through the calibration of step S1.

5. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 4 is characterized in that: In step S2, after converting the three-dimensional point cloud data acquired by the laser radar into the stockpile yard coordinate system, relevant point cloud data located in the working area are screened out from the three-dimensional point cloud data; Then, candidate point clouds of non-materials are screened out from relevant point cloud data according to reflectivity information of the three-dimensional point cloud data to obtain the three-dimensional point cloud coordinate set.

6. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 5 is characterized in that: Candidate point clouds are selected from relevant point cloud data in the following ways: The point cloud data is divided into two categories: high intensity and low intensity according to the reflection intensity of the reflectivity information; Divide the low-intensity point cloud data into a first aggregation group according to the position, and the remaining point clouds constitute a first discrete group; Divide the high-intensity point cloud data into a second aggregation group according to the position, and the remaining point clouds constitute a second discrete group; Defining a first boundary for the first aggregation group and a second boundary for the second aggregation group; The point cloud in the second discrete group that is within the first boundary is classified into the first aggregated group, and the point cloud in the first discrete group that is within the second boundary is classified into the second aggregated group; The remaining point clouds of the second aggregated group and the second discrete group that are not classified into the first aggregated group are screened as the candidate point clouds.

7. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 6 is characterized in that: The first aggregation group / second aggregation group is divided from the low-intensity / high-intensity point cloud data in the following manner: S21. Select any point from the low-intensity / high-intensity point cloud dataset as the center point; S22. The points close to the center point and less than the set distance are classified into the pre-selected aggregation group; S23. Taking any point in the pre-selected aggregation group as the center point, continue to classify points close to any center point and less than the set distance into the pre-selected aggregation group; S24. Expand outward from the first center point, and repeat step S23 until there is no point in the remaining points of the point cloud data set that is less than the set distance from any center point in the pre-selected aggregation group; S25. Repeat steps S21 to S24 for the remaining point cloud data sets of the low-intensity / high-intensity point cloud data sets to obtain more pre-selected aggregation groups until all the point cloud data of the low-intensity / high-intensity are divided; S26. The pre-selected aggregation group with the largest number of point clouds at this time is used as the first aggregation group / the second aggregation group.

8. The bucket wheel machine collision avoidance method based on laser radar perception according to claim 5 is characterized in that: Step S4 calculates the minimum straight-line distance between the obstacle and the anti-collision part as follows: According to the bucket wheel machine posture and position data, the current coordinates P2 (p 2-1 , p 2-2 …… 2-n ), n represents the number of anti-collision part coordinates; Get the three-dimensional point cloud coordinate set P1(p 1-1 , p 1-2 …… 1-m ), m represents the number of three-dimensional point cloud coordinate points in the protection area; Get the nearest p by nearest neighbor search 1-i -p 2-j Group; Calculate p 1-i -p 2-j The straight-line distance is taken as the minimum straight-line distance between the obstacle and the anti-collision part; In step S5, the working mode includes any one or more combinations of a deceleration mode, a stop mode and a normal mode; When the minimum straight-line distance is less than the first set distance and is in a continuously decreasing state, the switch signal is switched to the deceleration mode. When it continues to decrease to less than the second set distance, the switch signal is switched to the stop mode. When the minimum straight-line distance is in a state of continuous increase, the switch signal switches to the normal mode.

9. A bucket wheel machine anti-collision system based on laser radar perception, comprising a bucket wheel machine PLC, a bucket wheel machine, characterized in that: The bucket wheel machine is equipped with at least two 3D laser radars with a field of view covering the anti-collision area, and the 3D laser radars are connected to an industrial computer. The industrial computer is used to execute the method described in any one of claims 1 to 8, and is connected to the bucket wheel machine PLC so that the bucket wheel machine PLC can switch the working mode of the bucket wheel machine according to the switch signal.

10. The bucket wheel machine collision avoidance system based on laser radar perception according to claim 9 is characterized in that: The system also includes a self-cleaning component installed at the laser radar for cleaning the laser radar lens, and the self-cleaning component is connected to the industrial computer.

Images