Working Machine Operation System and Working Machine with Working Machine Operation System

Abstract

To highly accurately plan a load loading operation by a working machine so that a load after the load loading operation has a target shape. A working machine operation system is a working machine operation system with a bucket, including: an operation recording unit which records a loading operation of a hydraulic excavator when a load is loaded from the bucket onto a carrier; a shape acquiring unit which acquires a shape of the load loaded on the carrier after the loading operation; a correlation calculation unit which calculates a correlation of the loading operation and the shape of the load; and a loading operation calculation unit which calculates the loading operation based on the correlation and the target shape of the load.

Description

TECHNICAL FIELD

The present invention relates to a working machine operation system and a working machine with the working machine operation system.

BACKGROUND ART

PTL 1 discloses an example of a technique for estimating a shape of a load in a container as below. A load map indicating an ideal loading configuration inside the container is divided into grid-shaped parts and a value indicating an ideal level height of a substance inside the grid-shaped parts is calculated.

CITATION LIST

Patent Literature

PTL 1: Japanese Unexamined Patent Application Publication No. 11-310389

SUMMARY OF INVENTION

Technical Problem

When a working machine such as a hydraulic excavator loads a load onto a loading object of a transportation machine such as a dump truck, a shape or a gravity center position of the load loaded thereon may cause the overloading of the transportation machine and decrease the durability thereof. For this reason, there is a need to keep an ideal gravity center position or an ideal load shape on the loading object in the loading operation of the working machine. In order to realize this, an operation system automatically performing the loading operation by the working machine with high accuracy is considered.

In order to automatically perform the load loading operation by the working machine with high accuracy, there is a need to plan the load loading operation to have a target load shape on the loading object. Since a shape in which the load falls onto the loading object may be changed depending on the environment of the load such as the type, viscosity, and size of the load, it is necessary to recognize the shape of the load at a certain time point and to estimate how the load falls from the container used to carry the load in the working machine such as a hydraulic excavator with a bucket in order to realize the target load shape.

In PTL 1, since the environment of the load is not considered, it is difficult to highly accurately plan the load loading operation by the working machine so that the load after the load loading operation has the target shape.

An object of the invention is to highly accurately plan the load loading operation by the working machine so that the load after the load loading operation has the target shape.

Solution to Problem

One of features of the invention for solving the aforementioned problems is, for example, as follows.

A working machine operation system 100 is a working machine operation system 100 with a bucket 15, including: an operation recording unit 51 which records a loading operation of a hydraulic excavator 1 when a load 42 is loaded from the bucket 15 onto a carrier 41; a shape acquiring unit 52 which acquires a shape of the load 42 loaded on the carrier 41 after the loading operation; a correlation calculation unit 53 which calculates a correlation of the loading operation and the shape of the load; and a loading operation calculation unit 54 which calculates the loading operation based on the correlation and the target shape of the load 42.

Advantageous Effects of Invention

According to the invention, it is possible to highly accurately plan the load loading operation by the working machine so that the load after the load loading operation has the target shape. The objects, configurations, and effects other than those described above will be clarified by the description of the embodiments below.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a side view illustrating a hydraulic excavator and a dump truck according to an embodiment of the invention.

FIG. 2 is a block diagram illustrating a configuration in the vicinity of a calculation device according to the embodiment of the invention.

FIG. 3 is a block diagram illustrating a configuration of the calculation device according to the embodiment of the invention.

FIG. 4 is a side view illustrating a loading operation of a hydraulic excavator according to the embodiment of the invention.

FIG. 5 is a diagram illustrating a parameter for defining a loading operation according to the embodiment of the invention.

FIG. 6 illustrates a method of learning a correlation of parameters according to the embodiment of the invention.

DESCRIPTION OF EMBODIMENTS

Hereinafter, an embodiment of the invention will be described with reference to the drawings or the like. A description below indicates a detailed example of the contents of the invention. Then, the invention is not limited to the description and can be modified and corrected in various forms within the scope of the technical spirit disclosed in the specification. Further, in all drawings for describing the invention, those having the same function are indicated by the same reference numerals and the repetitive description thereof may be omitted in some cases.

An operation system of the invention is described according to a plurality of steps. However, the order of description does not limit the order in which a plurality of steps are performed. For this reason, the order of the plurality of steps can be modified within a range that does not hinder the contents at the time of operating the operation system of the invention.

Further, the plurality of steps of the operation system of the invention are not limited to being performed at individually different timings. For this reason, another step may be performed during a certain step or a certain step performing timing and another step performing timing may partially or entirely overlap.

FIG. 1 is a side view illustrating a dump truck and a hydraulic excavator including an operation system according to the embodiment of the invention. Hereinafter, the operation system will be described by exemplifying a hydraulic excavator having a bucket as a working machine having a container with reference to FIGS. 1 to 6. Additionally, the working machine of the invention is not limited to the hydraulic excavator and may be also applied to, for example, other working machines such as a wheel loader.

Similarly to a general hydraulic excavator, a hydraulic excavator 1 includes an upper turning body 11, a lower traveling body 12 which includes a crawler, a boom 13, an arm 14, and a bucket 15 which constitute a front part used for a work such as an excavation operation, a boom cylinder 16 which drives the boom 13, an arm cylinder 17 which drives the arm 14, a bucket cylinder 18 which drives the bucket 15, and the like. The upper turning body 11 is rotatably supported by the lower traveling body 12 and the upper turning body 11 is driven relative to the lower traveling body 12 by a turning motor (not illustrated). One end of the boom 13 is rotatably supported by the upper turning body 11 and the boom 13 is rotatably driven relative to the upper turning body 11 in response to the telescopic movement of the boom cylinder 16. One end of the arm 14 is rotatably supported by the boom 13 and the arm 14 is driven relative to the boom 13 in response to the telescopic movement of the arm cylinder 17. The bucket 15 is rotatably supported by the arm 14 and the bucket 15 is rotatably driven relative to the arm 14 in response to the telescopic movement of the bucket cylinder 18. The hydraulic excavator 1 with such a configuration can perform a desired work by controlling the bucket 15 at an arbitrary position and in an arbitrary posture while appropriately driving the boom cylinder 16, the arm cylinder 17, and the bucket cylinder 18.

In addition to these configurations, the hydraulic excavator 1 includes a boom inclination sensor 21 which acquires a rotation posture of the boom 13, an arm inclination sensor 22 which acquires a rotation posture of the arm 14, a bucket inclination sensor 23 which acquires a rotation posture of the bucket 15, a stereo camera 25 which acquires shapes of a loading object disposed in the upper turning body 11 and a load 42 loaded on the loading object, and a calculation device 26. The boom cylinder 16, the arm cylinder 17, and the bucket cylinder 18 are controlled by the calculation device 26. The stereo camera 25 is a device which includes two or more cameras and measures a distance from a subject to the stereo camera 25 based on an image captured by the plurality of cameras. Instead of the stereo camera 25, one or more sensors exhibiting the same effect as the stereo camera 25 may be provided. For example, the stereo camera 25 may be replaced by a laser sensor or a time of flight (TOF) type distance image camera.

In the embodiment of the invention, the loading object is set as a carrier 41 of the dump truck 4 and the load 42 is set as an excavated substance loaded on the carrier 41. Additionally, the loading object is not limited to the carrier 41 of the dump truck 4 and may be, for example, a ground or the like. In this case, the load 42 is the excavated substance loaded on the ground.

Referring to FIGS. 2 and 3, a configuration of the operation system of the embodiment of the invention will be described. FIG. 2 is a block diagram illustrating a configuration in the periphery of the calculation device. The calculation device 26 acquires the rotation postures of the boom 13, the arm 14, and the bucket 15 from the boom inclination sensor 21, the arm inclination sensor 22, and the bucket inclination sensor 23. In addition, the calculation device 26 acquires a shape of the carrier 41 or the load 42 from the stereo camera 25. Then, the calculation device 26 performs a calculation for obtaining a correlation of the acquired rotation posture or shape, plans an excavated substance loading operation based on the correlation, and generates an instruction for each cylinder 20.

FIG. 3 is a block diagram illustrating a configuration of the calculation device 26. The calculation device 26 includes a vehicle body controller 19 and an automatic control controller 24 generating operation signals for automatically operating the hydraulic excavator 1. The automatic control controller 24 includes an operation system 100 and a loading operation instruction unit 55. The operation system 100 includes an operation recording unit 51, a shape acquiring unit 52, a correlation calculation unit 53, and a loading operation calculation unit 54. Referring to FIG. 3, a process of the calculation device 26 in the case where the carrier 41 is a plane, that is, the load 42 is not loaded on the carrier 41 will be described.

The operation recording unit 51 acquires the above-described rotation posture. Then, a horizontal position of a bucket claw, a horizontal speed of the bucket claw, and a bucket rotation posture corresponding to a rotation posture of the bucket 15 are obtained based on the acquired rotation posture and are recorded. That is, the operation recording unit 51 records the loading operation of the hydraulic excavator 1 at the time of loading the excavated substance from the bucket 15 to the carrier 41. The shape acquiring unit 52 acquires a shape of the load 42 loaded on the carrier 41 after the loading operation. The correlation calculation unit 53 calculates a correlation of the loading operation and the load shape based on the horizontal position of the bucket claw, the horizontal speed of the bucket claw, and the bucket rotation posture during the loading operation along with the shape of the load 42 acquired by the shape acquiring unit 52.

With the above-described configuration, it is possible to obtain information for planning an optimal loading operation even in an environment in which the type or viscosity of the excavated substance changes. Then, it is possible to plan the optimal loading operation by using the correlation. Additionally, when the carrier 41 does not have a flat shape or the load 42 is loaded on the carrier 41, two load shapes including the load shape before the loading operation and the load shape after the loading operation may be acquired by the shape acquiring unit 52 and the correlation may be obtained by the correlation calculation unit 53 based on the load shape before the loading operation and the load shape after the loading operation. By using the correlation, a more optimal loading operation can be planned. Additionally, even when the carrier 41 is a plane, two load shapes including the load shape before the loading operation and the load shape after the loading operation may be acquired by the shape acquiring unit 52 and the correlation may be obtained based on the load shape. In the following description, a case in which two load shapes including the load shape before the loading operation and the load shape after the loading operation are obtained by the shape acquiring unit 52 will be described.

The loading operation calculation unit 54 calculates the loading operation of the hydraulic excavator 1 based on the correlation calculated by the correlation calculation unit 53 and the target shape of the load 42 loaded on the carrier 41. Based on the target shape in addition to the correlation calculated by the correlation calculation unit 53, the loading operation for the target shape can be calculated. Then, the loading operation instruction unit 55 transmits an operation signal for the loading operation calculated by the loading operation calculation unit 54 to the vehicle body controller 19.

The vehicle body controller 19 generates an instruction for the boom cylinder 16, the arm cylinder 17, and the bucket cylinder 18 based on the operation signal transmitted from the loading operation instruction unit 55. Accordingly, the boom 13, the arm 14, and the bucket 15 can be controlled at arbitrary rotation postures. Further, the boom 13, the arm 14, and the bucket 15 can perform arbitrary operations changing with time by sequentially changing the rotation postures of the boom 13, the arm 14, and the bucket 15. When the excavated substance is discharged to the carrier 41 by the same path, that is, the same loading operation, there is a possibility that the load 42 is loaded only on a part in the carrier 41 and the gravity center balance of the dump truck 4 is deteriorated. Thus, it is desirable to plan a loading operation having a uniform load shape by controlling the rotation posture of the bucket 15 while depicting a path in which the bucket 15 follows the carrier 41.

With the above-described configuration, it is possible to perform the loading operation of the hydraulic excavator 1 such that the load shape after the loading operation becomes the target shape. In addition, the calculation device 26 includes a central processing unit (CPU), a random access memory (RAM), a read only memory (ROM), and other peripheral circuits. For example, a method is considered in which respective units including the operation recording unit 51 or the correlation calculation unit 54 are stored in the ROM and are performed by the CPU using the RAM.

FIG. 4 is a side view illustrating the bucket 15 by focusing on the operation of the bucket 15 in the loading operation of the hydraulic excavator 1. Only the outline of the carrier 41 of the dump truck 4 is illustrated. Further, the bucket 15 during the loading operation is illustrated as a bucket posture 15a, a bucket posture 15b, a bucket posture 15c, and a bucket posture 15d in time series order. The bucket claw moves along the bucket claw movement path 31 and discharges the excavated substance to the carrier 41. For the description below, a plane in which the boom 13, the arm 14, and the bucket 15 of the hydraulic excavator 1 can be operated is set as an XZ plane, a horizontal direction illustrated in FIG. 4 is set as an X axis, and a vertical direction in the drawing is set as a Z axis.

The hydraulic excavator 1 performs a loading operation of sequentially changing the bucket rotation posture while the bucket 15 advances on the carrier 41 in the X-axis direction as illustrated from the bucket posture 15a to the bucket posture 15d. That is, an operation of sequentially changing the bucket rotation posture is performed while sequentially changing the rotation posture of the boom 13 or the arm 14.

The bucket rotation postures from the bucket posture 15a to the bucket posture 15d acquired by the operation recording unit 51 are respectively set as a bucket rotation posture 33a to a bucket rotation posture 33d. The bucket rotation posture when the opening surface of the bucket 15 is parallel to the X axis is set to 0°. In FIG. 4, the bucket rotation posture 33a is set to 0°. When the clockwise rotation in FIG. 4 is defined as a positive direction, the bucket rotation posture changes in the positive direction from the vicinity of 0° during the loading operation and the excavated substance inside the bucket 15 is discharged into the carrier 41.

An outline 42a and an outline 42b of the load 42 loaded on the carrier 41 are depicted inside the carrier 41. The outline 42a is a load shape on the XZ plane of the load 42 during a certain loading operation and the outline 42b is a load shape on the XZ plane of the load 42 during the next loading operation. Additionally, the outline on the XZ plane of the load 42 may be an average of those obtained by projecting a three-dimensional shape onto the XZ plane or an outline on an arbitrary plane parallel to the XZ plane. In general, since the loading operation of the hydraulic excavator 1 is performed a plurality of times for the same dump truck 4, there are a plurality of load shapes of the load 42.

Next, a method of acquiring the horizontal position of the bucket claw, the horizontal speed of the bucket claw, the bucket rotation posture, the load shape before the loading operation, and the load shape after the loading operation used to obtain the correlation by the correlation calculation unit 53 will be described.

The operation recording unit 51 acquires the three-dimensional shape of the load 42 loaded on the carrier 41 before the loading operation from the stereo camera 25. Next, the outline (for example, 42a) on the XZ plane of the load 42 is extracted from the acquired three-dimensional shape. Further, the three-dimensional shape of the load 42 is acquired by the stereo camera 25 similarly to the case before the loading operation after the loading operation and the outline (for example, 42b) on the XZ plane is extracted.

A speed vector of the bucket claw from the bucket posture 15a to the bucket posture 15d is set as a speed vector 32a to a speed vector 32d. The shape acquiring unit 52 records the rotation posture (for example, the bucket rotation posture 33b) of the bucket 15 during the loading operation, the horizontal speed (for example, the X-axis direction element of the speed vector 32b) of the bucket claw, and the horizontal position of the bucket claw. The bucket rotation posture can be obtained from the bucket inclination sensor 23. The horizontal position of the bucket claw can be obtained by the rotation postures of the boom 13, the arm 14, and the bucket 15 obtained from the boom inclination sensor 21, the arm inclination sensor 22, and the bucket inclination sensor 23 and the geometric relation among the boom 13, the arm 14, and the bucket 15 stored in advance. The horizontal speed of the bucket claw can be obtained based on the horizontal position of the bucket claw at different timings.

As described above, the automatic control controller 24 acquires five information items including the load shape before the loading operation, the time-series bucket rotation posture, the time-series horizontal position of the bucket claw, the time-series horizontal speed of the bucket claw, and the load shape after the loading operation every loading operation. When the correlation between the loading operation and the load shape is calculated based on the information, it is possible to plan the optimal loading operation in consideration of the environment of the load obtaining the target load shape. As the environment of the load, for example, the type or viscosity of the load can be exemplified. For example, when the excavated substance contains or does not contain a large amount of moisture, different load shapes are formed when the same loading operation is performed in both cases. Here, when the correlation of the load shape and the loading operation is obtained and the loading operation is planned based on the correlation, the target shape can be realized with high accuracy.

Referring to FIG. 5, a method of calculating a parameter for defining the loading operation will be described. The bucket rotation posture for the horizontal position is calculated from the time-series bucket rotation posture and the time-series horizontal position of the bucket claw and the bucket rotation posture is approximated as the cubic function of the horizontal position. In this case, the horizontal position of the bucket claw is set as X and the bucket rotation posture is set as θ_k. Similarly, the horizontal speed of the bucket claw is approximated as the cubic function of the horizontal position X. In this case, the horizontal speed of the bucket claw is set as V_k. The load shape is approximated as the cubic function of the horizontal position X, the load shape by the K-th loading operation is set as Z_k, and the load shape by the K+1-th loading operation is set as Z_k+1. These four cubic functions are expressed by the following equation (1).

$\begin{array}{cc}\left[\mathrm{Mathematical}\mathrm{Formula}1\right]& \\ \left[\begin{array}{c}{\theta }_k\\ V_k\\ Z_{k+1}\\ Z_k\end{array}\right]=\left[\begin{array}{cccc}{a}_{11}& a_{12}& a_{13}& a_{14}\\ a_{21}& a_{22}& a_{23}& a_{24}\\ a_{31}& a_{32}& a_{33}& a_{34}\\ a_{41}& a_{42}& a_{43}& a_{44}\end{array}\right]\left[\begin{array}{c}{x}^3\\ x^2\\ x\\ 1\end{array}\right]& \mathrm{Equation}\left(1\right)\end{array}$

In Equation (1), a₁₁ to a₁₄ are parameters indicating the bucket rotation posture θ_k, a₂₁ to a₂₄ are parameters indicating the horizontal speed V_k of the bucket claw, a₃₁ to a₃₄ are parameters indicating the load shape Z_k+1, and a₄₁ to a₄₄ are parameters indicating the load shape Z_k. In this way, when the parameters like the bucket rotation posture and the like are obtained, the information can be defined as four parameters at one time. Then, it is possible to highly accurately plan an optimal loading operation having the target load shape by learning the correlation of sixteen parameters in total.

Referring to FIG. 6, a method of learning the correlation of sixteen parameters will be described. In the embodiment of the invention, learning is performed by using a three-layered neural network. Additionally, various methods of learning the correlation exist and any method may be used.

Twelve parameters of i₁ to i₁₂ are set as input signals and four parameters from o₁ to o₄ are set as output signals. Then, the three-layered neural network outputting the output signal is constructed. Regarding the intermediate-layer neuron, the learning accuracy is improved as the number of neurons increases. However, since the calculation time increases as the number of neurons increases, it is desirable to determine the number of neurons in the intermediate layer in consideration of the necessary accuracy and the calculation capability of the automatic control controller 24. In the embodiment of the invention, the number of neurons at the intermediate layer is set as N. Similarly to a neural network of a general error back propagation method, each neuron has weight and threshold and performs learning by the following calculation.

[Step 1] Twelve parameters of the parameter a₂₁ to the parameter a₄₄ are input to the input signal i₁ to the input signal i₁₂.
[Step 2] An output of each neuron is calculated from an input layer toward an output layer.
[Step 3] The parameter a₁₁ to the parameter a₁₄ are given to the teaching signal t₁ to the teaching signal t₄ and an error signal of each neuron is calculated a difference of the output signals o₁ to o₄ and the teaching signals t₁ to t₄.
[Step 4] The weight and the threshold of each neuron are updated by using the error signal.
[Step 5] Steps 2 to 4 are repeated until a difference of the output signals o₁ to o₄ and the teaching signals t₁ to t₄ decreases enough.

By the above-described learning, it is possible to highly accurately estimate the parameter a₁₁ to the parameter a₁₄ from twelve parameters of the parameter a₂₁ to the parameter a₄₄. This means that the bucket rotation posture θ_k can be obtained by the input of the load shape Z_k before the loading operation, the target load shape Z_k+1, and the horizontal speed V_k of the bucket claw. That is, when the horizontal speed of the bucket claw is determined at the time of determining the load shape before the loading operation and the target shape, it is possible to obtain the bucket rotation posture having the target load shape. Additionally, when the initial position of the loading operation is given at the time of determining the horizontal speed of the bucket claw and the bucket rotation posture, the loading operations of the boom 13, the arm 14, and the bucket 15 are determined at the same time.

With such a configuration, the operation system 100 can highly accurately plan the loading operation having the target load shape even in a different environment by learning the environment while repeating the loading operation. In addition, it is desirable to increase the learning accuracy by performing learning using the neural network mentioned in the embodiment a plurality of times in the environment which is the same as the actual environment before the actual work.

Additionally, control lines and information lines are those which are considered to be desirable in the description and all control lines and information lines are not necessarily illustrated in the drawings.

In the embodiment of the invention, the operation recording unit 51, the shape acquiring unit 52, the correlation calculation unit 53, and the loading operation calculation unit 54 have been described as a part of the operation system 100. However, regarding respective components constituting the operation system 100, a position provided with respective components or a position for performing the processes of the respective components is not limited to one and, for example, the process of the correlation calculation unit 53 may be performed outside the calculation device 26. For example, an apparatus for centrally managing the working machine may be provided separately from the working machine and the apparatus may be provided with the operation system 100.

REFERENCE SIGNS LIST

• 1 hydraulic excavator
• 11 upper turning body
• 12 lower traveling body
• 13 boom
• 14 arm
• 15 bucket
• 15a to 15d bucket posture
• 16 boom cylinder
• 17 arm cylinder
• 18 bucket cylinder
• 19 vehicle body controller
• 21 boom inclination sensor
• 22 arm inclination sensor
• 23 bucket inclination sensor
• 24 automatic control controller
• 25 stereo camera
• 26 calculation device
• 31 bucket claw movement path
• 32a to 32d bucket claw speed vector
• 33a to 33d bucket rotation posture
• 4 dump truck
• 41 carrier
• 42 load
• 51 operation recording unit
• 52 shape acquiring unit
• 53 correlation calculation unit
• 54 loading operation calculation unit
• 55 loading operation instruction unit
• 100 operation system

Claims

1. A working machine operation system for a working machine with a container, comprising:

an operation recording unit which records a loading operation of the working machine when a load is loaded from the container onto a loading object;

a shape acquiring unit which acquires a shape of the load loaded on the loading object after the loading operation;

a correlation calculation unit which calculates a correlation of the loading operation and the shape of the load; and

a loading operation calculation unit which calculates the loading operation based on the correlation and a target shape of the load.

2. The working machine operation system according to claim 1,

wherein the correlation calculation unit calculates a correlation of the loading operation, the shape of the load before the loading operation, and the shape of the load after the loading operation.

3. A working machine comprising:

the working machine operation system according to claim 1.