SYSTEM FOR ESTIMATING HIDDEN OBJECTS AND EXCAVATORS

Abstract

Technology for appropriately estimating the presence of a hidden object in the excavation target during excavation work performed by an excavator is provided. The excavator comprises: a lower drive body; an upper rotating body, which is rotatably mounted on the lower drive body; a head, which is mounted on the upper rotating body and configured to excavate the target; and a control system comprising a processor and memory, configured to control the movement of the head. The hidden object estimation system comprises an information processing computer comprising a processor and memory, and is configured to estimate the probability of the presence of a hidden object in the excavation target, train a learning model to detect the hidden object by acquiring information regarding excavation reaction force during the excavation work performed by the head, and transmit the learning model to the control system.The control system estimates the presence or absence of the hidden object in the excavation target based on an excavation reaction force recorded during actual excavation work and the learning model.

Description

BACKGROUND OF THE INVENTION Field of invention The present disclosure relates to a system for estimating hidden objects and an excavator. Description of the state of the art During excavation work, an excavator may damage a hidden object within the excavation target, such as soil or the like. Therefore, work is carried out at a construction site to confirm the presence of the hidden object within the excavation target. For example, a disclosed excavation system acquires measurement data of a soil using a subsurface surveying device and estimates the location of a hidden object based on the measurement data using a trained model that has been trained on the location of a subsurface hidden object according to measurement data for training purposes. When estimating the presence of a hidden object using the subsurface investigation device during excavation work carried out by the excavator as described above, the excavation is frequently interrupted, and work efficiency is reduced. Therefore, it is necessary to be able to estimate the presence of any object potentially hidden within the excavation target simultaneously with the excavation work being carried out by the excavator at the construction site. SUMMARY OF THE INVENTION The present disclosure provides a technology with which the probability of the presence of a hidden object in an excavation target can be appropriately estimated during excavation work carried out by an excavator. One embodiment of the present disclosure provides a system for estimating hidden objects for an excavator, configured to estimate the probability of the presence of a hidden object in an excavation target. The excavator comprises a lower travel body, an upper rotating body rotatably mounted on the lower travel body, a head mounted on the upper rotating body and configured to excavate the target, and a controller comprising a processor and memory configured to control movement of the head.The hidden object estimation system comprises an information processing computer including a processor and memory, configured to train a learning model for detecting the hidden object by acquiring information regarding the excavation reaction force during excavation work performed by the target piece and transmitting the learning model to the controller. The controller estimates the presence or absence of the hidden object in the excavation target based on an excavation reaction force recorded during actual excavation work and the learning model. According to one embodiment, the probability of the presence of a hidden object in an excavation target can be appropriately estimated during excavation work carried out by an excavator. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a side view showing an excavator according to one embodiment; Fig. 2 is a side view showing various physical quantities in relation to excavation attachments; Fig. 3 is an explanatory diagram showing a basic excavator system; Fig. 4 is a diagram showing an example configuration of an excavation control system mounted on the excavator of Fig. 1; Fig. 5 is a diagram showing a cross-section of soil in which a water pipe is concealed; Fig. 6 is a graph showing the relationship between excavation reaction force and approach distance; Fig. 7 is a block diagram showing a detection system; Fig. 8 is a block diagram showing functional parts of an information processing device for generating a learning model; Fig. 9 is an explanatory diagram showing the acquisition of a learning model by an excavator, tuning of the learning model, and use of the learning model during excavation work; Fig.Figure 10 is a diagram illustrating image information displayed on a display device when a hidden object estimation function is performed; Figure 11A is a flowchart showing a processing sequence prior to actual excavation work; Figure 11B is a flowchart showing a hidden object estimation procedure when actual excavation work is performed; and Figure 12 is a schematic view showing an operating system configuration example. DETAILED DESCRIPTION OF THE REVELATION An embodiment for carrying out the present disclosure is described below with reference to the drawings. In the drawings, the same components are identified by the same reference numerals, and redundant descriptions may be omitted. Fig. 1 is a side view showing an excavator 100 according to one embodiment. The excavator 100 comprises a lower chassis 1 and an upper rotating body 3, which is rotatably mounted on the lower chassis 1 via a rotary mechanism 2. The excavator 100 includes an attachment AT, which is an example of an attachment as a working element. The attachment AT comprises a boom 4, which is attached to the upper rotating body 3, an arm 5, which is attached to the tip of the boom 4, and a bucket 6, which is attached to the tip of the arm 5. For the sake of simplicity, in this description, one side of the upper rotating body 3, to which the boom 4 is attached, is referred to as the front, and the side to which a counterweight is attached is referred to as the rear. The boom 4 is driven by a boom cylinder 7. The arm 5 is driven by an arm cylinder 8. The bucket 6 is driven by a bucket cylinder 9. The upper rotating body 3 is equipped with a cabin 10 and a power source, such as a motor 11, an electric motor, or the like. The cabin 10 contains an operating device 26, a control unit 30, a display device 40, a sound output device 45, and the like. The excavator 100 includes position detection devices M1 for detecting the position of the attachment AT. The position detection devices M1 also serve as detection devices for information about an excavation reaction force. In particular, the position detection devices M1 include a boom angle sensor M1a, an arm angle sensor M1b, and a bucket angle sensor M1c. For example, a rotation angle sensor for detecting the rotation angle of a boom foot bolt, a stroke sensor for detecting the stroke amount of the boom cylinder 7, and a tilt (acceleration) sensor for detecting the tilt angle of the boom 4 can be used as the boom angle sensor M1a. Similar sensors can also be used as the arm angle sensor M1b and as the bucket angle sensor M1c. The excavator 100 includes an object detection device 70 and the like on the upper rotating body 3. The object detection device 70 detects an object located around the excavator 100. The object could be, for example, a person, an animal, a vehicle, a construction machine, a building, a hole, and the like. The object detection device 70 is composed, for example, of an ultrasonic sensor, a millimeter-wave radar, an imaging device, an infrared sensor, or the like, or a combination thereof. The imaging device could be, for example, a monocular camera, a stereo camera, a LiDAR, a distance imaging sensor, or the like.In the illustrated example, the object detection device 70 comprises a rear camera 70B attached to the rear end of the top of the upper rotating body 3, a front camera 70F attached to the front end of the top of the cabin 10, a left camera 70L attached to the left end of the top of the upper rotating body 3, and a right camera 70R attached to the right end of the top of the upper rotating body 3. The object detection device 70 can be configured to detect an object (for example, a person) within an area set around the excavator 100. For example, the object detection device 70 can be configured to detect objects while distinguishing between a person and a non-human object. Fig. 2 is a side view showing various physical quantities related to the attachment AT. The boom angle sensor M1a detects a boom angle θ1. The boom angle θ1 is an angle, with respect to a horizontal line, of a line segment P1-P2 connecting a boom foot bolt position P1 and a boom connecting bolt position P2 in the XZ plane. The boom angle sensor M1b detects a boom angle θ2. The boom angle θ2 is an angle, with respect to a horizontal line, of a line segment P2-P3 connecting the boom connecting bolt position P2 and a blade connecting bolt position P3 in the XZ plane. The blade angle sensor M1c detects a blade angle θ3. The blade angle θ3 is an angle, with respect to a horizontal line, of a line segment P3-P4 that connects the blade connection bolt position P3 and a blade claw tip position P4 in the XZ plane.The blade angle θ3 can be calculated based on the output of an operation at the operating device 26. For example, the blade angle θ3 can be calculated based on outputs from pilot pressure sensors 15a and 15b (Fig. 3). In this case, the blade angle sensor M1c can be omitted. Fig. 3 is an explanatory diagram showing a basic system of the excavator 100. The basic system of the excavator 100 comprises the motor 11, a main pump 14, a pilot pump 15, a control valve unit 17, the operating device 26, the control unit 30, a display device 40, a tone output device 45, a motor control device 74, an operating mode switch 75, a switch 76 for estimating hidden objects, the position detection devices M1, excavation pressure sensors S1, and the like. The engine 11 is a power source for the excavator 100 and is, for example, a diesel engine that operates to maintain a predetermined speed. The output shaft of the engine 11 is connected to the input shafts of the main pump 14 and the pilot pump 15. The main pump 14 is a hydraulic pump for supplying hydraulic oil via a hydraulic oil line 16 to the control valve unit 17 and is, for example, a variable displacement swashplate hydraulic pump. The variable displacement swashplate hydraulic pump changes the piston stroke length, which defines a displacement volume, according to a change in the swashplate tilt angle in order to change the delivery flow rate per revolution. The swashplate tilt angle is controlled by a controller 14a. The controller 14a changes the swashplate tilt angle according to a change in a control current from the controller 30. For example, the controller 14a increases the swashplate tilt angle according to an increase in the control current, thereby increasing the delivery flow rate of the main pump 14.Conversely, the controller 14a reduces the tilt angle of the swashplate in accordance with a reduction in the control current, thereby decreasing the delivery flow rate of the main pump 14. A delivery pressure sensor 14b detects the delivery pressure of the main pump 14. An oil temperature sensor 14c detects the temperature of the hydraulic oil drawn in by the main pump 14. The pilot pump 15 is a hydraulic pump for supplying hydraulic oil to various hydraulic control devices, such as the operating device 26 and the like, via a pilot line 25. For example, a fixed displacement hydraulic pump can be used as the pilot pump 15. The control valve unit 17 controls the flow of hydraulic oil to the hydraulic actuators. In the illustrated example, the control valve unit 17 comprises several flow rate control valves. The control valve unit 17 selectively supplies one or more hydraulic actuators with hydraulic oil received from the main pump 14 via the hydraulic oil line 16 according to a change in pressure (pilot pressure) corresponding to the direction in which the control device 26 is actuated and the amount by which the control device 26 is actuated. The hydraulic actuators include, for example, the boom cylinder 7, the arm cylinder 8, the bucket cylinder 9, a left-hand travel hydraulic motor 1A, a right-hand travel hydraulic motor 1B, a rotary hydraulic motor 2A, and the like.In the illustrated example, the hydraulic motors (left-hand drive hydraulic motor 1A, right-hand drive hydraulic motor 1B, and rotary hydraulic motor 2A) are swashplate piston motors. However, the hydraulic motors can be replaced by electric motors. The control device 26 is a device used by an operator to actuate the hydraulic actuators and comprises a lever 26A, a lever 26B, a pedal 26C, and the like. The control device 26 receives a supply of hydraulic oil from the pilot pump 15 via the pilot line 25 to generate a pilot pressure. The control device 26 applies the pilot pressure via a pilot line 25a to a pilot port of a corresponding flow rate control valve. The pilot pressure changes according to both the direction in which the control device 26 is actuated and the amount with which the control device 26 is actuated. The control device 26 can be operated remotely. In remote operation, the control device 26 generates a pilot pressure based on information about the direction and amount of actuating received via wireless communication. The operating device 26 can be an electric operating device instead of a hydraulic operating device as described above. In this case, a solenoid valve for regulating the pilot pressure can be arranged between the flow rate control valves in the control valve unit 17 and the pilot pump 15. Information about the direction in which the electric operating device is actuated and the amount by which it is actuated is transmitted by the electric operating device to the controller 30 as an electrical signal. The controller 30 can regulate the magnitude of the pilot pressure applied to the flow rate control valves by adjusting the opening area of ​​the solenoid valve according to the electrical signal received from the electric operating device. The control unit 30 functions as a control element for driving and controlling the excavator 100. The functions of the control unit 30 can be implemented by any hardware or a combination of hardware and software. For example, the control unit 30 is primarily composed of a microcomputer comprising a processor, such as a central processing unit (CPU), memory, such as random access memory (RAM), read-only memory (ROM), and an interface device for various inputs and outputs. The control unit 30 performs various functions by executing different programs stored in the ROM and other memory on the CPU.For example, the controller 30 changes the size of the control current to the regulator 14a according to the pressure of the hydraulic oil in a negative control valve and controls the delivery flow rate of the main pump 14 via the regulator 14a. The display device 40 is a device for displaying various information and is located in the cab 10 near the operator's seat. In the illustrated example, the display device 40 comprises an image display unit 41 and an input unit 42. The image display unit 41 is a liquid crystal display. The input unit 42 is a membrane switch. The operator can input information and instructions into the controller 30 using the input unit 42. The operator can recognize the operating status and control information of the excavator 100 by viewing the image display unit 41. The display device 40 is connected to the controller 30 via a communication network, such as CAN or the like. However, the display device 40 can also be connected to the controller 30 via a dedicated line. The display device 40 is powered by receiving energy from a storage battery 90. The storage battery 90 is charged with energy generated by a generator 11a. The electrical energy of the storage battery 90 is also supplied to devices other than the control unit 30 and the display device 40, such as electrical components 72 and the like of the excavator 100. A starter 11b of the engine 11 can be driven by the electrical energy from the storage battery 90 to start the engine 11. The sound output device 45 is a device for outputting sound information. In the illustrated example, the sound output device 45 is a loudspeaker located in cabin 10 near the driver's seat. The sound output device 45 can be a buzzer. The engine control device 74 is a device for controlling the engine 11. For example, the engine control device 74 controls a fuel injection quantity and the like in order to achieve an engine speed that is set via the input device. The engine control device 74 transmits various data indicating the state of the engine 11 (for example, data relating to physical quantities, such as the coolant temperature detected by a water temperature sensor 11c, and the like) to the controller 30. The controller 30 stores various data in a memory 30a and transmits them to the display device 40 or the like as needed. The same applies to data indicating the swashplate tilt angle, output by the controller 14a; data indicating the main pump 14 discharge pressure, output by the discharge pressure sensor 14b; data indicating the hydraulic oil temperature, output by the oil temperature sensor 14c; data indicating the pilot pressure, output by the pilot pressure sensors 15a and 15b; and the like. The operating mode switch 75 is a switch for changing the operating modes of the excavator 100 and is located in the cab 10. The operating modes include, for example, an M (manual) mode and an SA (semi-automatic) mode. The control unit 30 switches the operating mode of the excavator 100 according to an output of the operating mode switch 75. The M (manual) mode is a mode in which the excavator 100 is operated according to an input from the operator into the control device 26. For example, it is a mode in which the boom cylinder 7, arm cylinder 8, bucket cylinder 9, and the like are operated according to the input from the operator into the control device 26. The SA (semi-automatic) mode is a mode in which the excavator 100 operates automatically when a predetermined condition is met, regardless of the input from the control device 26. For example, when the predetermined condition is met, at least one of the boom cylinder 7, arm cylinder 8, or bucket cylinder 9 is automatically actuated, regardless of the input from the control device 26.The operating modes can include a fully automatic mode in which the entire lower carriage 1, the slewing mechanism 2, the boom cylinder 7, the arm cylinder 8 and the bucket cylinder 9 and the like operate autonomously. Switch 76, for the hidden object estimation mode, is located in cab 10 and is used to start this mode. In this mode, a process is performed to estimate the presence or absence of a hidden object within the excavated soil. Specifically, the hidden object estimation mode of the excavator 100 estimates the presence or absence of a hidden object based on the excavation reaction force. The operator toggles between starting and stopping the hidden object estimation mode by pressing switch 76. The controller 30 executes the hidden object estimation function mode according to a start command from switch 76 for hidden object estimation mode and stops the hidden object estimation function mode according to a stop command from switch 76 for hidden object estimation mode. However, the controller 30 can activate the hidden object estimation function mode if it determines, based on the position of the attachment piece AT and the like, that an excavation operation is being carried out, regardless of whether switch 76 for hidden object estimation mode is actuated. For example, the controller 30 can continuously execute the hidden object estimation function mode from the time the excavation operation is started until the time a boom lifting operation is performed. The excavation pressure sensors S1 are examples of detection devices that detect information about an excavation reaction force. These devices detect the pressure of the hydraulic oil in the hydraulic cylinders, such as the boom cylinder 7, and output the detected data to the controller 30. The excavation pressure sensors S1 according to the embodiment are composed of a combination of excavation pressure sensors S11 to S18. Excavation pressure sensor S11 detects a boom bottom pressure, which is the pressure of the hydraulic oil in a bottom-side oil chamber of the boom cylinder 7. Excavation pressure sensor S12 detects a boom rod pressure, which is the pressure of the hydraulic oil in a rod-side oil chamber of the boom cylinder 7.Similarly, the S13 lifting pressure sensor detects arm bottom pressure, the S14 lifting pressure sensor detects arm rod pressure, the S15 lifting pressure sensor detects bucket bottom pressure, and the S16 lifting pressure sensor detects bucket rod pressure. The S17 lifting pressure sensor detects counterclockwise rotation pressure, which is the hydraulic oil pressure at the first port (left-hand port) of the rotary hydraulic motor 2A. The S18 lifting pressure sensor detects clockwise rotation pressure, which is the hydraulic oil pressure at the second port (right-hand port) of the rotary hydraulic motor 2A. A control valve E1 is a valve that operates according to a command from the controller 30. In the illustrated example, the control valve E1 is used to forcefully actuate the flow rate control valves relating to predetermined hydraulic cylinders, regardless of the content of any input to the control device 26. The control valve E1 is equivalent to the solenoid valve located between the flow rate control valves and the pilot pump 15 in a case where the electrical control device described above is used. Fig. 4 is a diagram showing a configuration example of an excavation control system mounted on the excavator 100 shown in Fig. 1. The excavation control system comprises the position detection devices M1, the excavation pressure sensors S1, the operating mode switch 75, the switch 76 for the hidden object estimation mode, the control unit 30, the control valve E1, the display device 40, the tone output device 45, and the like. The control unit 30 forms an excavation reaction force calculation section 31 and a hidden object estimation section 32 within the control unit 30, by means of a processor that reads and executes a program stored in memory. The excavation reaction force calculation unit 31 is a functional element for calculating an excavation reaction force. The excavation reaction force calculation unit 31 is configured to calculate an excavation reaction force based on at least one output from the excavation pressure sensors S1. According to the embodiment, the excavation reaction force calculation unit 31 calculates an excavation reaction force based on an output from the excavation pressure sensors S1 and the position of the attachment piece AT detected by the position detection devices M1. The excavation reaction force calculation unit 31 can additionally use an output from a vehicle body tilt sensor. The vehicle body tilt sensor can, for example, be based on an accelerometer or a gyroscope. Examples of output from the S1 excavation pressure sensors include, for example, at least one of the boom ground pressure, boom rod pressure, arm ground pressure, arm rod pressure, bucket ground pressure and bucket rod pressure detected by the S11 to S16 excavation pressure sensors. The excavation reaction force calculation part 31 can calculate a cylinder thrust force based on an output from the excavation pressure sensors S1. A cylinder thrust force is calculated, for example, based on an excavation pressure and the pressure-bearing area of ​​a piston sliding in a cylinder. Examples of cylinder thrust forces include a boom cylinder thrust force (f1), a boom cylinder thrust force (f2), and a bucket cylinder thrust force (f3). Specifically, the boom cylinder force (f1) is expressed as the difference between a cylinder extension force, which is the product of the boom bottom pressure and the pressure-bearing area of ​​the piston in the boom bottom-side oil chamber, and a cylinder contraction force, which is the product of the boom rod pressure and the pressure-bearing area of ​​the piston in the boom rod-side oil chamber. The same applies to the boom cylinder thrust force (f2) and the bucket cylinder thrust force (f3). The excavation reaction force calculation part 31 can calculate an excavator torque based on the position of the attachment piece AT and a cylinder thrust force. As shown in Fig. 2, the magnitude of a bucket excavation torque (τ3) is expressed as a value obtained by multiplying the magnitude of the bucket cylinder thrust force (f3) by a distance G3 between a line of action of the bucket cylinder thrust force (f3) and the bucket connection bolt position P3. The distance G3 is a function of the bucket angle θ3 and is an example of a connection reinforcement. The same applies to a boom excavation torque (τ1) and a boom excavation torque (τ2). A distance G1 is the distance between a line of action of the boom cylinder thrust force (f1) and the boom foot bolt position P1, and a distance G2 is the distance between the line of action of the boom cylinder thrust force (f2) and the boom connection bolt position P2. The excavation reaction force is calculated, for example, as the product of a mechanism function where the boom angle θ1, arm angle θ2, and bucket angle θ3, shown in Fig. 2, are arguments, and a function where the boom excavation torque (τ1), arm excavation torque (τ2), and bucket excavation torque (τ3) are arguments. The function where the boom excavation torque (τ1), arm excavation torque (τ2), and bucket excavation torque (τ3) are arguments can be a function where the boom cylinder thrust force (f1), arm cylinder thrust force (f2), and bucket cylinder thrust force (f3) are arguments. The function, where the boom angle θ1, arm angle θ2 and blade angle θ3 are arguments, can be a function based on a force equilibrium equation, can be a function based on Jacobi, or can be a function based on the principle of virtual work. As described above, the excavation reaction force can be derived from the currently detected values ​​of the various sensors. The detected value from the excavation pressure sensors S1 can be used unchanged as the information about the excavation reaction force. Alternatively, the cylinder thrust force values ​​calculated based on the detected values ​​from the excavation pressure sensors S1 can be used as the information about the excavation reaction force. The excavation reaction force can also be derived from the values ​​of the excavation torques, calculated from the cylinder thrust force values ​​(calculated based on the detected values ​​from the excavation pressure sensors S1), and from values ​​relating to the position of the attachment piece AT, derived from the detected values ​​of the position detection devices M1. The excavation reaction calculation part 31 can calculate an excavation reaction force acting in the direction of rotation based on outputs from excavation pressure sensors S17 and S18. If the counterclockwise rotational pressure (P17) detected by excavation pressure sensor S17 is higher than the clockwise rotational pressure (P18) detected by excavation pressure sensor S18, the upper rotating body 3 will rotate counterclockwise. Conversely, the upper rotating body 3 will rotate clockwise if the clockwise rotational pressure (P18) detected by excavation pressure sensor S18 is higher than the counterclockwise rotational pressure (P17) detected by excavation pressure sensor S17. For example, the excavation reaction force calculation part 31 can calculate the counterclockwise rotational reaction force as the counterclockwise rotational pressure in a case where the counterclockwise rotational pressure (P17) is higher than the clockwise rotational pressure (P18).The excavation reaction force calculation part 31 can, for example, calculate the clockwise rotational pressure (P18) in a case where the clockwise rotational pressure (P18) is higher than the counterclockwise rotational pressure (P17), as the excavation reaction force acting in the clockwise direction. If an electric rotary motor is installed instead of the rotary hydraulic motor 2A, the excavation reaction force calculation part 31 can calculate the excavation reaction force acting in the direction of rotation based on information relating to electrical energy, such as the direction and magnitude of a current supplied to the electric rotary motor. The hidden object estimator 32 is configured to detect a hidden object based on information relating to the excavation reaction force. According to the embodiment, the hidden object estimator 32 estimates the presence or absence of a hidden object based on the excavation reaction force calculated by the excavation reaction force computational part 31 and a previously determined hidden object estimator learning model. This learning model is described in detail later. For example, when it estimates the presence of a hidden object, the hidden object estimator 32 issues a control command to the control valve E1. Upon receiving a control command from the hidden object estimator 32, the control valve E1 forcibly actuates the flow rate control valves relating to the predetermined hydraulic cylinders to forcibly extend or retract the predetermined hydraulic cylinders, regardless of the content of any operation input to the control device 26. For example, even if the boom control lever is not actuated, the control valve E1 forcibly extends boom cylinder 7 by forcibly moving the flow rate control valve relating to boom cylinder 7. As a result, the excavator l 100 can forcibly raise the boom 4 to decrease the digging depth (or change direction).Alternatively, even when the arm control lever is actuated, the control valve E1 can forcibly stop the arm cylinder 8 by forcibly moving the flow rate control valve that relates to the arm cylinder 8. By forcibly stopping the arm 5, the excavator 100 can avoid contact between the bucket 6 and the hidden object. In this way, the control valve E1 can reduce contact between the attachment AT and the hidden object by forcibly extending, contracting, or stopping at least one of the boom cylinder 7, the arm cylinder 8, or the bucket cylinder 9 in accordance with a control command from the part 32 for estimating hidden objects. The hidden object estimator 32 can issue a control command to the display device 40 when it estimates the presence of a hidden object. Upon receiving the control command from the hidden object estimator 32, the display device 40 shows an estimated position of the hidden object. For example, the display device 40 can show a virtual viewpoint image representing the state of the excavator 100 as seen from a virtual viewpoint directly above the excavator 100, and display the object hidden in the ground by overlaying it onto the virtual viewpoint image. The virtual viewpoint image is generated based on images captured by the front camera 70F, the rear camera 70B, the left camera 70L, and the right camera 70R.Alternatively, the display device 40 can display an image representing a cross-section of the ground on which the excavator 100 is located and can indicate the object hidden in the ground by superimposing it onto the image representing the cross-section. Furthermore, the hidden object estimator 32 can issue a control command to the sound output device 45 when the hidden object is estimated to be present. Controller 30 starts the hidden object estimation function mode according to a start command from switch 76 for hidden object estimation mode. When the hidden object estimation function mode is started, part 32 for estimating hidden objects estimates the presence or absence of a hidden object based on an excavation reaction force calculated by the excavation reaction force calculation part 31. Conversely, controller 30 stops the hidden object estimation function mode according to a stop command from switch 76 for hidden object estimation mode.This can prevent the excavator 100 from erroneously estimating the presence of a hidden object in response to a fluctuation in the excavation reaction force and issuing a control command to the control valve E1, the indicator device 40, or the tone output device 45, even though it is clear that no hidden object is present. When the hidden object estimation mode is stopped, the excavation reaction force calculation unit 31 does not need to calculate an excavation reaction force, thus reducing the computational load. The hidden object estimation function can be executed in both manual (M) and semi-automatic (SA) modes of the excavator 100. However, the hidden object estimation function can only be executed when semi-automatic mode is selected. When semi-automatic mode is selected, the operator can improve hidden object detection accuracy by moving the AT probe along a pre-set target path. Next, with reference to Fig. 5, the movement of excavator 100 when excavator 100 estimates a water pipe U1 to be a hidden object is described as a representative case. Fig. 5 is a diagram showing a cross-section of soil in which the water pipe U1 is hidden. In Fig. 5, the soil ES before it is excavated is indicated by a dashed line. First, the operator of excavator 100 activates the operating mode switch 75 to switch the excavator 100's operating mode to SA (semi-automatic) mode. The operator then manually operates the control device 26 to move the claw tip of the bucket 6 to the desired position (first position PS1). After the claw tip of the bucket 6 has been moved to the desired position, the operator activates switch 76 for the hidden object estimation mode to start the hidden object estimation function. In SA (semi-automatic) mode, the controller 30 moves the attachment AT autonomously. Specifically, the controller 30 automatically extends or contracts at least one of the boom cylinder 7, arm cylinder 8, or bucket cylinder 9 in such a way as to move a predetermined portion of the attachment AT along a previously set target path TP. However, even in the hidden object estimation mode, the controller 30 does not necessarily have to move the attachment AT autonomously and can move the attachment AT according to the operator's instructions at the control device 26. The controller 30 automatically moves the attachment piece AT so that the claw tip of the bucket 6 moves along a previously set first target path TP1 (dash-dot line) during excavation. As the claw tip of the bucket 6 moves along the first target path TP1, the excavation reaction force calculation part 31 recalculates the excavation reaction force based on outputs from the position detection devices M1 and outputs from the excavation pressure sensors S1. The hidden object estimation part 32 recalculates the presence or absence of a hidden object based on the excavation reaction force calculated by the excavation reaction force calculation part 31. When the claw tip of the shovel 6 reaches the end of the first target path TP1, the control unit 30 stops the autonomous movement of the attachment AT. This means that the hidden object estimator 32 has not detected a hidden object up to the point where the claw tip of the shovel 6 reaches the end of the first target path TP1. The operator then activates switch 76 for hidden object estimation mode to stop the hidden object estimation function. The control unit 30 can automatically stop the hidden object estimation function if the operator performs a boom lift operation or a boom lift and rotation operation by manually operating control device 26. After performing a soil dump operation and a boom lower and rotation operation, the operator moves the claw tip of the bucket 6 to the next desired position (second position PS2) to perform the next excavation operation. For example, the second position PS2 is located at a depth D1 from the soil ES before the soil ES was excavated and is almost directly below the first position PS1. After the bucket 6's claw tip has been moved to the second position PS2, the operator activates the hidden object estimation function mode. When the hidden object estimation function mode is activated, the controller 30 automatically moves the attachment AT so that the bucket 6's claw tip moves along a previously set second target path TP2 (dash-dot line). As the bucket 6's claw tip moves along the second target path TP2, the excavation reaction force calculation unit 31 recalculates the excavation reaction force based on outputs from the position detection devices M1 and outputs from the excavation pressure sensors S1. The hidden object estimation unit 32 then recalculates the presence or absence of a hidden object based on the excavation reaction force calculated by the excavation reaction force calculation unit 31. When the claw tip of the shovel 6 reaches the end of the second target path TP2, the control unit 30 stops the autonomous movement of the attachment AT. This means that the part 32 for estimating hidden objects has not detected a hidden object up to the last point where the claw tip of the shovel 6 reaches the end of the second target path TP2. The operator then performs an earth-unloading operation and a boom lowering and rotating operation in the same manner as described above, and then moves the claw tip of bucket 6 again to the next desired position (third position PS3) to perform the next excavation operation. The third position PS3 is located at a depth D2 from a first exposed surface and almost directly below the second position PS2. After moving the claw tip of the bucket 6 to the third position PS3, the operator activates the hidden object estimation function. When the hidden object estimation function is activated, the control unit 30 automatically moves the attachment piece AT so that the claw tip of the bucket 6 moves along a previously set third target path TP3 (dash-dot line). When the claw tip of the bucket 6 moves along the third target path TP3, the excavation reaction force calculation part 31 recalculates the excavation reaction force based on outputs from the position detection devices M1 and outputs from the excavation pressure sensors S1. The hidden object estimation part 32 recalculates the presence or absence of a hidden object based on the excavation reaction force calculated by the excavation reaction force calculation part 31. For example, part 32, used for estimating hidden objects, estimates that a hidden object is present when the claw tip of bucket 6 reaches a fourth position PS4. The fourth position PS4 is located at a depth D3 from a second exposed surface and on the third target path TP3. The fourth position PS4 is a position where the distance between the water conduit U1, which is a hidden object present in one direction along the third target path TP3, and the claw tip of bucket 6 becomes a value AD1. With reference to Fig. 6, a process for estimating the presence or absence of a buried object based on the excavation reaction force is described. Fig. 6 is a graph showing the relationship between the excavation reaction force F and the approach distance AD. The vertical axis of Fig. 6 is the excavation reaction force F, calculated by the excavation reaction force calculation part 31, and the horizontal axis of Fig. 6 is the approach distance AD. The approach distance AD ​​is the distance between the current position of the claw tip of the bucket 6 and the buried object (water pipe U1) in the direction along the target path TP. Fig. 6 shows that the approach distance AD ​​decreases along the horizontal axis from left to right until it becomes zero.This means that the claw tip of the shovel 6 is located further away from the water pipe U1 when the approach distance AD ​​is the value AD0 than when the approach distance AD ​​is the value AD1. In particular, the dashed double-dotted line in Fig. 6 illustrates a learning model that serves as a reference for the excavation reaction force. That is, the learning model is represented by a function or the like that describes the relationship between the change in the approach distance AD ​​and the excavation reaction force during excavation work. The function representing the learning model can be a linear function, a polynomial function, a logarithmic function, or the like. This learning model has learned the excavation reaction force that the attachment piece AT experiences from the excavation target (in the ground) when no hidden object is present. In the learning model, the excavation reaction force F increases as the approach distance AD ​​decreases. This is because the excavation reaction force increases as the amount of earth and sand loaded in the bucket 6 increases as the bucket 6 approaches the machine body (upper rotating body 3). Part 32, on estimating hidden objects, compares this learning model with the excavation reaction force repeatedly calculated during excavation work to estimate the presence or absence of a hidden object. For example, the excavation reaction force as the claw tip of bucket 6 moves along the third target path TP3 in Fig. 5 is shown by a solid line in Fig. 6. In this case, the excavation reaction force fluctuates essentially according to the learning model at the approach distance AD ​​range from AD0 to AD1. However, when the approach distance AD ​​exceeds AD1, the excavation reaction force deviates from the learning model and increases sharply. This is because the water pipe U1 is located on the third target path TP3 of the bucket 6, and because, as the claw tip of the bucket 6 approaches the water pipe U1, the soil and sand between the bucket 6 and the water pipe U1 are compressed.Part 32, for estimating hidden objects, calculates, for example, the difference between the training model and the actual excavation reaction force and determines whether this difference is equal to or greater than a threshold, thus estimating the presence or absence of a hidden object (water pipe U1). Figure 6 shows an example where the difference between the training model and the actual excavation reaction force becomes equal to or greater than the threshold at a position where the approach distance AD2 is equal to or greater than the threshold. That is, the position where the approach distance AD2 is equal to the time at which the claw tip of the bucket 6 reaches the fourth position PS4 on the third target path TP3 in Figure 5. This position AD2 is, for example, a position approximately 20 cm from the hidden object.The position where the approach distance AD ​​AD3 is is a position where the hidden object (water pipe U1) is located. The comparison procedure between the training model and the actual excavation reaction force is not limited to the above, and various methods can be applied. For example, Part 32, for estimating hidden objects, can compare an average rate of increase of the actual excavation reaction force with respect to the approach distance AD ​​to an average rate of increase of the training model. If the average rate of increase of the actual excavation reaction force exceeds the average rate of increase of the training model by more than a predetermined value, it can be estimated that a hidden object is present. Next, with reference to Fig. 7, a System 200 for estimating hidden objects is described for generating the hidden object detection learning model described above and for providing the learning model. Fig. 7 is a block diagram showing the System 200 for estimating hidden objects. The system 200 for estimating hidden objects includes an information processing device 210, a communication network 220, a cloud server 230 and a control unit 30 for the excavator 100. The information processing device 210 is an information processing component for generating a learning model used for detecting hidden objects by the excavator 100. The information processing device 210 can be a known computer comprising a processor 211, a memory 212, an input / output interface 213, and a communication interface 214. The information processing device 210 can consist of one or more computers. The information processing device 210 is intended, for example, for use in a company or organization that manufactures, manages, or uses a work machine, such as the excavator 100 and the like. The information processing device 210 manages information about the usage status and conditions of several excavators 100 or other work machines and provides information to the excavator 100 or other work machines. The information processing device 210 is connected to the cloud server 230 via the communication network 220 and provides and collects information via the cloud server 230.Alternatively, the information processing device 210 can be configured to perform information communication via the cloud server 230 (or directly) with a portable terminal device, such as a computer, smartphone, tablet and the like, which is available for use by a worker. The communication network 220 of the system 200 for estimating hidden objects can be the internet or a dedicated line, such as Ethernet. As described above, in the system 200 for estimating hidden objects according to the embodiment, the information processing device 210 and each excavator 100 exchange information by accessing the cloud server 230 via the communication network 220. For example, if the excavator 100 uploads various types of excavator information, including an excavation reaction force when excavation work is carried out at a construction site, to the cloud server 230, the information processing device 210 retrieves this information from the cloud server 230 at a suitable time. Conversely, when the information processing device 210 uploads work information (not shown) to the cloud server 230, the excavator 100 downloads the work information from the cloud server 230 at a suitable time before work begins. The work information includes location information of the construction site, information on a 3D or 2D model of the construction site, information on the work machines used on the construction site, a training model for detecting hidden objects, and the like. In the System 200 for estimating hidden objects, the Information Processing Device 210 and the Excavator 100 can be configured to communicate directly without the Cloud Server 230 (or via another computer). Alternatively, a worker can provide information about the Excavator 100 to the Information Processing Device 210 by storing the information about the Excavator 100 in an external storage device and connecting the external storage device to the Information Processing Device 210. The same applies to the information that the Information Processing Device 210 is to provide to the Excavator 100. Next, with reference to Fig. 8, a learning model for detecting hidden objects, generated by the information processing device 210, is described. Fig. 8 is a block diagram showing functional parts of the information processing device 210 for generating a learning model. The information processing device 210 generates a learning model for detecting hidden objects, for example, by an unsupervised learning procedure. As an example, the information processing device 210 internally creates a sensor data storage part 215, a training data extraction part, a reaction force model generation part 217, an optimization part 218, and an evaluation part 219, among others, by the processor 211 reading and executing a program stored in the memory 212. The sensor data storage unit 215 stores information acquired by various sensors of the excavator 100. Here, the information processing device 210 stores information about excavators 100 of the same type in the sensor data storage unit 215, linked together, as well as linked sensor values ​​and time information. Information stored in the sensor data storage unit 215 includes information relating to the excavation reaction force (sensor data from the excavation pressure sensors S1 and position detection devices M1). Furthermore, the information stored in the sensor data storage unit 215 can include information about an area outside the excavator 100, detected by the object detection device 70 (a camera, LiDAR, etc.) of the excavator 100.In the case of implementing drive support for the excavator 100, the sensor data storage unit 215 can further store simulation data obtained by simulating operation of the excavator 100, including excavation work, by the attachment AT. The simulation data can be data actually used by an excavator 100 during excavation work, or it can be data obtained from a simulation previously performed by the information processing device 210 to be made available to the excavator 100. The training data extraction part 216 is the input layer for machine learning, used to extract data for generating a hidden object detection training model from various types of information stored in the sensor data storage part 215, and to provide the data to the reaction force model generation part 217. The data used to generate the training model includes the information described above regarding the excavation reaction force (the pressures from the excavation pressure sensors S1 and the like), information regarding the position of the excavator 100, and information from the object detection device 70. The reaction force model generation part 217 creates a hidden object detection training model based on data provided by the training data extraction part 216. Here, the excavation target dug by the excavator 100 has different soil properties, such as earth, sand, a mixture of earth and sand, hard bedrock, soft rock, and so on. Since different soil properties produce different excavation reaction forces, the reaction force model generation part 217 creates a training model for each different soil property. Therefore, training models are created, including an earth training model used for earth excavation, a sand training model used for sand excavation, a gravel training model used for gravel excavation, and so on. For example, the reaction force model generator part 217 detects a pattern of sequential excavation reaction force data using a time series analysis neural network and an extracted feature parameter of the excavation reaction force. The reaction force model generator part 217 extracts a feature parameter from topographic data based on information (captured image information) from a camera or LiDAR representing the actual excavation work performed by the excavator 100, or based on construction data. The reaction force model generator part 217 then generates a reaction force prediction model by combining the excavation reaction force pattern and the feature parameter of the topographic data in the fully connected layer.The reaction force prediction model, for example, is a function or tabular information that indicates a change in the excavation reaction force with respect to time or a change in the approach distance AD. The reaction force prediction model is calculated for each soil type (earth, sand, earth and sand, and the like) based on the feature size of the topographic data. Optimization Part 218 optimizes reaction force prediction models and generates training models (an earth-based training model, a sand-based training model, an earth-and-sand training model, and the like) based on one or more reaction force prediction models generated by the reaction force model generation part 217, which are input. For example, Optimization Part 218 performs the Bayesian optimization process based on the reaction force prediction models and the feature set of the topographic data or another excavation work condition. Optimization Part 218 can obtain a highly accurate training model by using multiple reaction force prediction models generated by the reaction force model generation part 217. The optimization process of Optimization Part 218 is not limited to the Bayesian optimization process, and various techniques, such as grid search, can be used. The evaluation part 219 assesses the quality of a generated learning model and outputs the evaluation result to the reaction force model generation part 217. For example, the evaluation part 219 can evaluate the learning model based on information about the actual presence or absence of a hidden object, confirmed by a subsequently described uncovering operation. Alternatively, the evaluation of the learning model can be performed by optimizing a hyperparameter that can be set or adjusted by a user of the information processing device 210. Thus, a learning model can be generated in a manner that fulfills a desired requirement, such as improving the performance of the learning model, easily obtaining a learning model, and the like. When the information processing device 210 generates one or more learning models (an earth learning model, a sand learning model, and an earth and sand learning model) for detecting hidden objects through the process described above, the information processing device 210 uploads the learning models to the cloud server 230. Thus, the excavator 100 at the construction site can download the learning models by accessing the cloud server 230. The excavator 100's control unit 30 estimates the presence or absence of a hidden object during the excavation work described above, using the learning models downloaded during the excavation. Here, the excavator 100 can estimate the presence of a hidden object based on the soil conditions by selecting a learning model from among several types (the soil learning model, the sand learning model, and the soil and earth learning model) that corresponds to the soil conditions of the excavation target. For example, when the operator of the excavator 100 recognizes the soil conditions of the excavation target, the operator uses the input part 42 of the display device 40 and enters the soil conditions of the excavation target. Thus, the control unit 30 can select the learning model to be used from among the several learning model types. Alternatively, the control unit 30 can be configured to automatically select the soil composition of the excavation target based on image information captured by a camera and similar device of the excavator 100's object detection system 70. This allows for an immediate switch to a suitable training model, even if, for example, the soil composition changes midway through excavation and the operator does not notice. However, the hardness and viscosity of the excavated material by the excavator 100 during an actual operating phase change due to the influence of air temperature and humidity, or the moisture content in the soil, which are related to these conditions, and the like. Therefore, the excavator 100 may not be able to detect a hidden object simply using the learning model available to it, which may fail because the excavation reaction force may not correspond to the actual construction site. Therefore, the excavator 100's control unit 30 performs a process to tune (adjust) the learning model to match the actual excavation target by excavating the target at the actual construction site to confirm the condition of the excavation target. Fig. 9 is an explanatory diagram showing the acquisition of a learning model by the excavator 100, the tuning of the learning model, and the use of the learning model during excavation work. As shown in Fig. 9, the excavator 100 performs preliminary excavation work in an area where it has been determined that there are no hidden objects on the construction site. During the preliminary excavation work, the excavator 100 detects an actual excavation reaction force using various sensors (the position detection devices M1 and the excavation pressure sensors S1). Controller 30 can determine whether the learning model needs to be adjusted by comparing the rate of change in the excavation reaction force at the actual construction site with the learning model recorded by Controller 30. If it is determined that the learning model needs adjustment, Controller 30 adjusts the learning model using the actual excavation reaction force at the construction site. During the adjustment of the learning model, the learning model (predicted excavation reaction force), represented by a linear function, a polynomial function, a logarithmic function, or the like, is fitted to recorded data of the actual excavation reaction force at the construction site. As a result, the learning model is appropriately corrected to a model that would indicate a change in the excavation reaction force corresponding to the actual construction site.In the following, the corrected learning model will also be referred to as a corrected learning model. On-site adjustments can be performed for a single, pre-selected learning model (e.g., an earth learning model) or for all of the available learning model types (e.g., an earth learning model, a sand learning model, and a gravel learning model). Soil conditions can change during excavation work on-site. If the controller has calculated 30 corrected learning models for multiple learning model types, model switching can be performed smoothly. After completing the previous excavation work, the controller 30 performs the excavation work shown in Fig. 5 using the corrected learning model. That is, by performing the excavation work in SA (semi-automatic) mode, the excavator 100 moves the attachment AT autonomously, while the excavation pressure sensors S1 and the position sensing devices M1 record the respective detection values. The excavation reaction force calculation unit 31 of the controller 30 calculates the excavation reaction force based on these sensors. Part 32, used for estimating hidden objects in control 30, compares the calculated excavation reaction force with the corrected learning model to estimate the presence or absence of a hidden object in the excavation target during excavation work by the attachment AT. As described above, control 30 can accurately estimate the presence of a hidden object in the excavation target because the corrected learning model has been calibrated to the actual construction site. If it is estimated that a hidden object is present in the excavation target, the excavator 100 can autonomously control the movement of the attachment piece AT so that the attachment piece AT does not come into contact with the hidden object. Specifically, the control unit 30 deactivates the arm closing operation based on the detection of the hidden object to stop the movement of the attachment piece AT. Alternatively, the control unit 30 can change directions so that the claw tip of the bucket 6 does not come into contact with the hidden object by automatically extending or retracting the boom cylinder 7 to raise the boom 4. If it is estimated that a hidden object is present in SA (semi-automatic) mode or M (manual) mode, it is preferred that the controller 30 alerts the operator so that the attachment AT does not approach the hidden object. For example, the controller 30 can inform the operator via the tone output device 45 about the level of the distance between the claw tip of the bucket 6 and the hidden object. In this case, the controller 30 can shorten the interval between the intermittent tones as the distance between the bucket 6 and the hidden object decreases. If the claw tip of the bucket 6 gets very close to the hidden object, the controller 30 can emit a loud alarm to the operator via the tone output device 45. If the excavator 100's control unit 30 estimates the presence of a hidden object, a hidden object exposure operation can be performed on-site to confirm its presence or absence. During this operation, for example, the worker uses a tool to dig at the location where an object is estimated to be hidden, thereby exposing any hidden object from the excavation target. Alternatively, during the exposure operation, the excavator 100 operator can confirm the presence or absence of a hidden object by carefully excavating the area where the object is estimated to be hidden by operating the attachment AT.If the presence or absence of a hidden object is confirmed by the excavation work at the location where a hidden object is estimated to be present, the worker can transmit this information to the excavator 100's control unit 30 and / or the cloud server 230 using an information terminal device available to the worker. As a result, the excavator 100's control unit 30 and / or the cloud server 230 store the data of the excavation reaction force at which an object is estimated to be hidden and the result indicating the presence or absence of a hidden object during the excavation work, linked together. In a case where information is received indicating that a hidden object is indeed present during the excavation work, the controller 30 or the cloud server 230 can recognize that the hidden object detection using the learning model (corrected learning model) functioned correctly. Conversely, if information is received indicating that no hidden object is present during the excavation work, the controller 30 or the cloud server 230 can recognize that the hidden object detection using the learning model (corrected learning model) was faulty. Here, the controller 30 can recalibrate the learning model (or the threshold for determining a hidden object) based on the received information. Thus, the excavator 100 can execute the hidden object estimation function mode using the corrected learning model or the newly corrected threshold. Furthermore, the controller 30 can display information relating to hidden object estimation on the display device 40, which serves as a user interface when the hidden object estimation function mode is executed. Fig. 10 is a diagram illustrating image information 400 that is displayed on the display device 40 when the hidden object estimation function mode is executed. The image information 400 can be displayed automatically, independently of the SA (semi-automatic) mode or the M (manual) mode, along with the execution of the hidden object estimation function mode. The image information 400 includes a mode selection display section 410, a selected model display section 420, an abnormality degree indicator 430, a model parameter 440, a model accuracy display section 450, a reaction force display section 460, an environment display section 470, an external connection setting display section 480, and a detail setting display section 490. The mode selection display section 410 allows the operator to select from several types of learning modes, a tuning mode for adjusting the learning mode, or the operating mode for estimating a hidden object, in which actual excavation work is carried out. For example, the operator activates a switch located on lever 26A or lever 26B of the operating device 26 to turn on the mode selection display section 410. Upon activation, the mode selection display section 410 shows a selection window containing various modes, from which the operator can select the desired mode. The selected model display section 420 shows various learning models selected by the operator (an earth learning model, a sand learning model, a gravel learning model, and the like). After the tuning mode has been executed, the selected model display section 420 can show an indicator that the model has been tuned (a pictogram representing a corrected learning model, and the like). When the hidden object estimation function is executed during actual excavation work, the abnormality level indicator 430 converts the probability of a hidden object being present into an abnormality level and communicates this to the operator. The abnormality level is obtained, for example, as the ratio between the difference in the excavation reaction force from the training model and the threshold value. In the illustrated example, the abnormality level is communicated by a circular pie chart and a numerical value. The color of the circular pie chart can change according to the abnormality level, for example, changing to green when the abnormality level is in the range of 0% to 50%, to yellow when the abnormality level is in the range of 50% to 80%, and to red when the abnormality level is in the range of 80% to 100%. If the degree of abnormality is high (in other words, if the difference from the threshold is large), the controller 30 can inform the operator about the hidden object via the abnormality indicator 430. If the degree of abnormality is high, the controller 30 can automatically stop the movement of the attachment AT or change its direction as described above. Alternatively, the operator can monitor the degree of abnormality and, for example, specify that the movement of the attachment AT should be stopped at an abnormality level of 70%, that the movement of the attachment AT should be stopped at an abnormality level of 90%, or similar. Model parameter 440 displays the hyperparameters in the learning model (corrected learning model) in such a way that the operator can adjust them. These hyperparameters include, for example, the soil conditions at the construction site, the detection accuracy of the hidden object estimation function, and the operating mode of excavator 100. The Model Accuracy Display section 450 shows the result of an evaluation of a currently applied learning model (corrected learning model). For example, when evaluating the learning model, the score is increased in a case where no hidden object is actually present, while the absence of a hidden object is detected, and the score is decreased in a case where a hidden object is present, while the absence of a hidden object is detected. Alternatively, when evaluating the learning model, the score is increased in a case where a hidden object is actually present, when the presence of a hidden object is detected, and the score is decreased in a case where no hidden object is actually present, when the presence of a hidden object is detected. The reaction force display section 460 shows the actual learning model used during excavation (corrected learning model) and the excavation reaction force detected by the sensors (position detection devices M1 and excavation pressure sensors S1) of the excavator 100 in the form of a graph. For example, the graph displays information showing a change in the excavation reaction force in relation to a change in excavation distance, with the excavation distance on the horizontal axis and the excavation reaction force on the vertical axis. Alternatively, the graph displays information showing a change in the excavation reaction force in relation to time, with the excavation time on the horizontal axis and the excavation reaction force on the vertical axis. Thus, the operator can operate the excavator 100 while observing the learning model and changes in the actual excavation reaction force in real time. The environmental display section 470 shows environmental factor variables. These variables include, for example, the date and time the excavator performs excavation work, weather, operator ID, and similar information. Display area 480 for external connection settings shows the communication environment between the controller 30 and the external communication network. For example, when connecting to the cloud server 230, the operator can activate display section 480 for external connection settings to configure the connection status with the cloud server 230. The detailed display section 490 allows the operator to change settings other than those described above in connection with the execution of the hidden object estimation function. When the image information 400 described above is displayed on the display device 40, the operator of the excavator 100 can make settings related to the learning model for the hidden object estimation function and can clearly see the relationship between the learning model and the excavation reaction force during the actual excavation work. In particular, according to the embodiment, the excavator 100 can inform the operator by displaying the learning model that the excavator 100 has determined that a hidden object is present because the excavation reaction force deviates from the learning model during the excavation work. The excavator 100 according to the embodiment is configured essentially as described above, and its operation is described below with reference to the flowcharts in Fig. 11A and Fig. 11B. Fig. 11A is a flowchart showing a process sequence prior to the actual execution of the excavation work. Fig. 11B is a flowchart showing a procedure for estimating hidden objects during the actual execution of the excavation work. In the method for estimating a hidden object, the excavator's control unit 30 performs 100 steps S101 to S109 of Fig. 11A and Fig. 11B. Controller 30 accesses cloud server 230 to acquire various learning models stored on cloud server 230 (step S101). The acquisition pattern for the learning models is not limited to this, and a storage device that stores the learning models can be connected to controller 30, allowing controller 30 to acquire the learning models. Alternatively, controller 30 can store the learning models in advance. In this case, step S101 can be omitted. The controller 30 displays a message prompting the operator to calibrate the learning model to the actual construction site and instructs the operator of excavator 100 to perform a preliminary excavation operation, thereby calibrating the learning model (step S102). By performing this calibration, the controller 30 can acquire a corrected learning model that corresponds to the actual construction site. However, the operator can choose not to perform calibration, in which case the learning model available to the controller 30 is used as is. Furthermore, if, for example, several excavators 100 are operating at the same construction site, calibration can be performed by one excavator 100, and information about the calibration result (corrected learning model) can be stored on the cloud server 230. The other excavators 100 then acquire the calibration result from the single excavator 100 on the cloud server 230.Therefore, voting by multiple excavators 100 can be omitted. Alternatively, voting by multiple excavators 100 can be performed, and the voting results can be stored in the cloud server 230. This allows the learning model to be corrected using the multiple voting results stored in the cloud server 230 and then transferred to each excavator 100. This improves the voting accuracy of each excavator 100's learning model. Then the excavator 100 switches to excavation work at the actual construction site. The control unit 30 determines whether the operator has selected the operating mode, the function mode for estimating hidden objects, and the like (step S103). If the function mode for estimating hidden objects has been selected (step S103: YES), the process continues to step S104. In step S104, the controller 30 reads the stored learning model (corrected learning model). Here, the controller 30 displays the image information 400 described above so that the operator can see the learning model to be used. Then, based on the control from the control 30 or operation by the operator, the excavator 100 moves the attachment piece AT to carry out the excavation work at the excavation target (step S105). During this excavation operation, the controller acquires 30 sensor data points from the position detection devices M1 and the excavation pressure sensors S1, calculates an excavation reaction force from this data, compares the calculated excavation reaction force with the learning model, and determines the presence or absence of a hidden object (step S106). If it is determined that no hidden object is present (step S106: NO), the process proceeds to step S107. At step S107, controller 30 determines whether the excavation work should be terminated. If the excavation work is not terminated (step S107: NO), the process returns to step S105 to continue the excavation work. If, however, the excavation work is terminated (step S107: YES), the current process flow is ended. If step S106 determines that a hidden object is present (step S106: YES), the process proceeds to step S108. In step S108, the control system 30 performs the following actions as described above: stopping the movement of the attachment piece AT, changing direction, and so forth, and displays information indicating the presence of a hidden object on the display device 40. Thus, the operator can see that the excavator 100 has stopped working because a hidden object is present at the excavation target during the excavation process. Here, the operator of excavator 100 can inform nearby workers on the construction site that a hidden object has been detected, prompting the workers to begin excavation work to uncover any such object. The workers then provide feedback to control unit 30 or cloud server 230 regarding whether or not a hidden object is actually present (step S109). As a result, the controller 30 can evaluate the used learning model (corrected learning model) based on information from the workers about the actual presence or absence of a hidden object, and thereby correct the learning model or the threshold to be used in the next excavation operation. For example, if there is no hidden object, the controller 30 can correct the function represented by the learning model (corrected learning model), taking into account the changes in the excavation reaction force that occurred this time. As described above, the excavator 100 and the hidden object estimation method can appropriately detect an object concealed within the excavation target by performing the excavation work using the learning model trained by the information processing device 210. Furthermore, the excavator 100 uses a corrected learning model that corresponds to the actual construction site by aligning the learning model acquired by the cloud server 230 with the actual site conditions. As a result, the excavator 100 can more accurately estimate the presence or absence of a hidden object. Next, with reference to Fig. 12, a configuration example of an operating system SYS according to another embodiment is described. Fig. 12 is a schematic diagram showing the configuration example of the operating system SYS. The operating system SYS comprises an excavator 100, a remote control office RC, and an administration center MC. The excavator 100 shown in Fig. 12 has the same configuration as the excavator 100 shown in Fig. 1. The excavator 100, the remote control office (RC), and the management center (MC) are connected to exchange data via the NW communication network. The excavator 100, the remote control office (RC), and the management center (MC) can also be connected to exchange data directly without using the NW communication network. In the illustrated example, the excavator 100 transmits information about the construction site to the remote control office (RC). This allows the remote operator (RO) at the remote control office (RC) to understand the status of the construction site based on the information from the excavator 100. The Bagger 100 is equipped with sensors capable of three-dimensionally detecting the position and shape of objects on the construction site. For example, the Bagger 100 is equipped with a spatial recognition device. Therefore, the Bagger 100 can transmit the results of the three-dimensional survey of the construction site to the remote control office (RC). The spatial detection device is a device for detecting the space around the excavator 100. In the illustrated example, the spatial detection device is a LiDAR. The LiDAR measures, for example, the distance between each of the one million or more points located in the monitored area and the LiDAR. The spatial detection device can be any device capable of measuring the distance to an object. For example, the spatial detection device could be a stereo camera or a combination of an imaging device and a rangefinder, such as millimeter-wave radar. One or more Excavator 100s can be included in the SYS operating system. If multiple Excavator 100s are included, the remote operator (RO) of a specific Excavator 100 can collect information about the construction site covered by that specific Excavator 100, as well as information about the construction site covered by the other(s) of the other Excavator 100s. The remote control office RC contains a communication device T2, a remote control RCC, an operating device 26E, an operating sensor 43, a display device D1E, an internal audio output device SP2E, and an internal audio recording device M2E. A control seat DS is installed in the remote control office RC, where the remote operator RO sits to remotely control the excavator 100. The communication device T2 is configured to communicate with a communication device attached to the excavator 100. The RCC remote control is a computing device for performing various calculations. In the present embodiment, the RCC remote control is comprised of a microcomputer including a CPU and memory. Various functions of the RCC remote control are implemented by the CPU, which executes a program stored in the memory. The display device D1E is a device capable of displaying various types of information. The display device D1E shows an image based on information transmitted by the excavator 100, allowing the remote operator RO in the remote control office RC to visually assess the excavator 100's surroundings. In the illustrated example, the display device D1E is a liquid crystal display for showing an image captured by the imaging device mounted on the excavator 100. The display device D1E can be a display or projector for achieving stereopsis with the naked eye, and can also be VR glasses or similar devices. The SP2E internal sound output device is a device capable of outputting sound information. Based on information transmitted by the excavator 100, the SP2E internal sound output device emits a sound, allowing the remote operator (RO) in the remote control office (RC) to hear the sound occurring at the construction site. The control device 26E is equipped with the operating sensor 43 for detecting the content of an operation on the control device 26E. The operating sensor 43 is, for example, a tilt sensor for detecting the tilt angle of an operating lever, an angle sensor for detecting the swivel angle of the operating lever about a swivel axis, and the like. The operating sensor 43 can be constructed from any other sensor, such as a pressure sensor, a current sensor, a voltage sensor, a distance sensor, and the like. The operating sensor 43 outputs information about the detected content of an operation on the control device 26E to the remote control RCC. Based on the received information, the remote control RCC generates an operating signal and transmits the generated operating signal to the excavator 100. The operating sensor 43 can be configured to generate the operating signal.In this case, the operating sensor 43 can output the operating signal to the communication device T2, bypassing the remote control RCC. With such a configuration, the remote operator RO can remotely control the excavator 100 from the remote control office RC. The MC control center is a facility equipped with various devices for managing the excavator 100 at the construction site or for remote control of the excavator 100 by the remote operator RO in the remote control office RC, and so on. In the illustrated example, the MC control center is installed in a location separate from both the excavator 100 construction site and the remote control office RC. The MC control center contains a 300 control device, an SP2C internal audio output device, and an M2C internal audio collection device. The Management Device 300 is an example of the control unit and is, for example, a server computer (commonly referred to as a cloud server) or an edge server. The Management Device 300 is typically a stationary terminal device, but can also be a portable terminal device (for example, a laptop computer, a tablet, a smartphone, or the like). Even with the SYS operating system described above, the excavator 100 can perform hidden object estimation using the learning model. When the remote operator RO executes the hidden object estimation function during excavation work at the construction site, the excavator 100's control unit 30 or the remote control RCC reads the learning model described above and compares it to the excavation reaction force. This makes it possible to perform excavation work at the construction site by appropriately determining the presence or absence of a hidden object. In the system 200 for estimating hidden objects according to the embodiment, the information processing part for generating a learning model is provided in the information processing device 210 outside the excavator 100. However, the information processing part can also be provided in the controller 30 of the excavator 100. In other words, the system 200 for estimating hidden objects can be implemented solely through the configuration of the excavator 100. The technical idea and effects of the present disclosure, as described in the above embodiments, are described below. A first aspect of the present disclosure is the system 200 for estimating hidden objects, which estimates the probability that a hidden object is present in an excavation target, for the excavator 100, comprising: the lower travel body 1; the upper rotating body 3, which is rotatably mounted on the lower travel body 1; the attachment AT, which is mounted on the upper rotating body 3 and configured to excavate the excavation target; and the control unit (controller 30), which is configured to control a movement of the attachment AT, wherein the system 200 for estimating hidden objects includes an information processing unit (information processing device 210 or information processing computer) configured to: train a learning model for detecting a hidden object by acquiring information relating to an excavation reaction force during excavation work by the attachment (AT);and transmits the learning model to the control unit, wherein the control unit estimates the presence or absence of a hidden object in the excavation target based on an excavation reaction force recorded during actual excavation work and the learning model.; As described above, the system 200 for estimating hidden objects during excavation work by the excavator 100 can adequately estimate the presence or absence of a hidden object in the excavation target by using the learning model trained on information related to the excavation reaction force. That is, the system 200 for estimating hidden objects can estimate the presence or absence of a hidden object with a higher probability than a configuration that merely monitors the excavation reaction force detected by the sensors of the excavator 100, based on changes in the excavation reaction force relative to the learning model. Furthermore, the system 200 for estimating hidden objects, which accumulates sensor data each time excavation work is performed, can improve the learning model.As a result, a hidden object can be estimated more accurately during excavation work carried out by the excavator 100. Furthermore, the control unit (Control 30) acquires information regarding the condition of the excavation target where the actual excavation work is being carried out and adjusts the learning model available to the control unit (Control 30) based on this information. Thus, according to System 200 for estimating hidden objects, it is possible to adjust the learning model to match the condition of the excavation target (soil composition, moisture content, viscosity, weather, etc.) at the actual construction site. As a result, the presence or absence of a hidden object can be estimated according to the condition of the excavation target, and a hidden object can be estimated more accurately. Furthermore, the information regarding the condition of the excavation target includes information about the excavation reaction force, which is recorded when the excavation target, where the actual excavation work is to be carried out, is excavated beforehand. Thus, the system 200 for estimating hidden objects can adjust the learning model to the actual construction site based on the information about the excavation reaction force recorded when the excavation target is excavated beforehand. The information processing unit (information processing device 210) generates several learning models for different types of soil conditions at the excavation target, and the control unit (control 30) selects the learning model to be used based on the soil conditions at the excavation target where the actual excavation work is being carried out. Thus, the excavator 100 can adequately monitor the excavation response force during the actual excavation work using the learning model that corresponds to the soil conditions at the excavation target. After determining that a hidden object is present during the actual excavation work, the control unit (controller 30) or the information processing unit (information processing device 210) acquires information confirming the presence or absence of a hidden object and evaluates the training model based on this information. Thus, the hidden object estimation system 200 can improve the training model based on the presence or absence of a hidden object confirmed by the excavation work and can further improve the accuracy of the next hidden object estimation. The control unit (control 30) calculates the difference between the excavation reaction force during actual excavation work and the learning model. It determines that a hidden object is not present if the difference is less than a threshold, and that a hidden object is present if the difference is greater than or equal to the threshold. Thus, the hidden object estimation system can easily and accurately estimate the presence or absence of a hidden object. If a hidden object is detected, the control unit (control 30) stops the movement of the attachment piece AT or changes its direction. This allows the excavator 100 to prevent the attachment piece AT from coming into contact with the suspected hidden object. The display device 40 for showing the learning model used during actual excavation work is also provided. This allows the operator of the excavator 100 to easily identify the learning model used when estimating the location of a hidden object. The display device 40 also shows the image information 400, which includes information relating to the result of estimations of the presence or absence of a hidden object. Thus, the operator of the excavator 100 can immediately recognize the result of estimations of the presence or absence of a hidden object and take the necessary action. A second aspect of the present disclosure is the excavator 100, comprising the lower drive body 1; the upper rotating body 3, which is rotatably mounted on the lower drive body 1; the attachment AT, which is mounted on the upper rotating body 3 and configured to excavate a target; and a control unit (control 30) configured to control a movement of the attachment AT, wherein the control unit is configured to: acquire a learning model for detecting a hidden object by acquiring information relating to an excavation reaction force during excavation work by the attachment AT; and estimate the presence or absence of the hidden object in the target based on an excavation reaction force acquired during actual excavation work and the learning model. Again, a hidden object in the target can be suitably estimated during excavation work by the excavator 100. The System 200 for estimating hidden objects and the excavator 100 according to the embodiments disclosed herein are exemplary and in no way limiting. The embodiments can be modified and improved in various ways without departing from the scope and spirit of the appended claims. The details described in the aforementioned several embodiments can be configured in other ways, provided no contradiction arises, and can be combined to the extent that no contradiction arises. DESCRIPTION OF REFERENCE MARK 1 lower drive body 3 upper rotating body 30 control unit 40 display device 100 excavator 200 system for estimating hidden objects 210 information processing device 400 image information AT attachment

Claims

System for estimating hidden objects for an excavator, configured to estimate the probability of the presence of a hidden object in an excavation target, wherein the excavator comprises: a lower travel body; an upper rotating body rotatably mounted on the lower travel body; an attachment AT mounted on the upper rotating body and configured for excavating the excavation target;and a controller comprising a processor and memory configured to control movement of the attachment, and wherein the system for estimating hidden object comprises: an information processing computer comprising a processor and memory configured to train a learning model for detecting the hidden object by acquiring information relating to an excavation reaction force during excavation work performed by the attachment and transmitting the learning model to the controller, wherein the controller estimates the presence or absence of the hidden object in the excavation target based on an excavation reaction force acquired during actual excavation work and the learning model. System for estimating a hidden object according to claim 1, wherein the controller acquires information relating to a state of the excavation target at which the actual excavation work is carried out and adjusts the learning model acquired by the controller on the basis of the information relating to the state of the excavation target. System for estimating a hidden object according to claim 2, wherein the information relating to the condition of the excavation target is information about an excavation reaction force that is captured when the excavation target at which the actual excavation work is to be carried out is previously excavated. System for estimating hidden object according to one of claims 1 to 3, wherein the information processing computer generates multiple learning models for respective types of soil conditions of the excavation target, each of the multiple learning models being the learning model, and the controller selects the learning model to be used based on a soil condition of the excavation target at which the actual excavation work is carried out. System for estimating a hidden object according to any one of claims 1 to 3, wherein, after the controller estimates that the hidden object is present during the actual excavation work, the controller or information processing computer acquires information confirming the presence or absence of the hidden object, and the learning model evaluates on the basis of the information confirming the presence or absence of the hidden object. System for estimating hidden object according to any one of claims 1 to 3, wherein the control calculates a difference between the excavation reaction force during the actual excavation work and the learning model, determines that the hidden object is not present if the difference is less than a threshold, and determines that the hidden object is present if the difference is greater than or equal to the threshold. System for estimating hidden objects according to claim 6, wherein the control, when it is determined that the hidden object is present, stops the movement of the attachment or changes directions. System for estimating a hidden object according to any one of claims 1 to 3, further comprising: a display device configured to display the learning model used in the actual excavation work. System for estimating a hidden object according to claim 8, wherein the display device displays image information that includes information relating to a result of estimating the presence or absence of the hidden object. Excavator comprising: a lower drive body; an upper rotating body provided to rotate on the lower drive body; an attachment provided on the upper rotating body and configured for excavating a target; and a controller comprising a processor and memory and configured to control movement of the attachment, the controller being configured to: acquire a learning model for detecting a hidden object by acquiring information about an excavation reaction force during excavation work performed by the attachment; and estimate the presence or absence of the hidden object in the target based on an excavation reaction force acquired during actual excavation work and the learning model.

Images