JP2019213557A - Work path generation system - Google Patents

Abstract

To provide a work path generation system to be used in an agricultural working vehicle, capable of more flexibly performing autonomous traveling on the basis of an operator's actual operation.SOLUTION: A work path generation system comprises a teacher data storage unit 82 and a control unit 4. The teacher data storage unit 82 stores: farm field information including outline information on a non-rectangular deformed farm field including a work region in which predetermined work is executed by straight traveling of a tractor 1 and a headland region in which turning is permitted; and first work path information being information on a first work path on which the tractor 1 is made to travel by an operator manually in a partial region of the farm field. On the basis of the farm field information and the first work path information, the control unit 4 generates a second work path on which autonomous travel is possible for the remaining region excluding the partial region. The second work path includes a plurality of straight paths. The control unit adjusts lengths of the straight paths of the second work path so that the headland region's width is constant.SELECTED DRAWING: Figure 3

Description

  The present invention mainly relates to a work route generation system used by an agricultural work vehicle configured to be capable of autonomous traveling.

  2. Description of the Related Art Conventionally, an agricultural work vehicle capable of autonomous traveling using a satellite positioning system has been known. Patent Literature 1 discloses a work vehicle of this type. In the work vehicle disclosed in Patent Document 1, every time the vehicle body reaches the edge of the field by the operation of the operator, the vehicle is recognized as a recognition area point, and a polygon formed by connecting the recognition area points with a straight line is set as a recognition area. After the setting of the area, the cultivation path when the field is run by the operator for the first time is displayed on the monitor. When the operator determines that the field is appropriate, the set field area and the generated path are stored, and When judged to be appropriate, a correction is made, autonomous traveling means for autonomously traveling in the recognized recognition area is provided, and at the time of the second and subsequent tilling work, the autonomous traveling along the generated route is performed. ing. According to Patent Document 1, it is assumed that this configuration allows an artificial operation to be performed in only one process, and thereafter, a favorable operation can be performed while maintaining the same operating conditions.

Japanese Patent No. 4801252

  However, in the configuration of Patent Literature 1, autonomous traveling can be performed for a second time or later in a field that has been manually operated, but autonomous traveling cannot be performed flexibly according to the situation such as work content. .

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a work route generation system used by an agricultural work vehicle that can perform autonomous traveling more flexibly based on an actual operation of an operator. It is in.

Means and effects for solving the problems

  The problems to be solved by the present invention are as described above. Next, means for solving the problems and the effects thereof will be described.

  According to an aspect of the present invention, a work route generation system having the following configuration is provided. That is, the work route generation system includes a storage unit and a control unit. The storage unit is capable of storing field information and first work route information. The field information includes outer shape information of a non-rectangular deformed field including a work area in which a predetermined work is performed by the work vehicle traveling straight and a headland area that allows the work vehicle to turn. The first work route information is information on a first work route in which the work vehicle is manually run by an operator in a part of the field. The control unit is capable of generating a second work route capable of autonomous traveling in a remaining area excluding the partial area in the field based on the field information and the first work path information. The second work path includes a plurality of straight paths. The control unit is capable of adjusting the length of the straight traveling path of the second work path so that the width of the headland region is constant.

  This makes it possible to learn only a part of the field and apply it to the entire field, so that it is possible to reduce the time and man-hour required for performing manned traveling and learning by the tractor, and It is possible to keep the width of the turning area for turning the wheel constant.

  In the work route generation system, the following configuration is preferable. That is, the first work path includes a straight path. The length of the straight path of the first work path and the second work path may be different depending on the shape of the field.

  This makes it possible to create the second work route according to the shape of the field based on the operation actually performed by the operator.

FIG. 1 is a side view showing an overall configuration of a robot tractor according to one embodiment of the present invention. The top view of a robot tractor. FIG. 2 is a block diagram showing a main configuration of a control system of the robot tractor. The figure which shows a mode that the working machine was lowered | hung to the working position. FIG. 4 is a conceptual diagram showing teacher data provided for learning a motion prediction model. The figure which showed the remote control device. The side view which showed the mode that the manned robot tractor accompanied the unmanned robot tractor. FIG. 4 is a plan view showing a state where a robot tractor in a manual operation mode travels on a field by an operation of an operator. FIG. 9 is a plan view showing a state in which the robot tractor in the automatic operation mode performs work while traveling unattended, based on the learning result in the manual operation mode in FIG. 8. FIG. 4 is a plan view showing how a robot tractor in a manual operation mode learns an operation of an operator on a part of a field. FIG. 11 is a plan view showing a state in which the robot tractor in the automatic operation mode performs work while traveling unattended in the remaining area of the field based on the learning result in the manual operation mode in FIG. 10. FIG. 3 is a block diagram showing an example in which data relating to learning is transmitted and received between a plurality of robot tractors.

  Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a side view showing an overall configuration of a robot tractor 1 according to one embodiment of the present invention. FIG. 2 is a plan view of the robot tractor 1. FIG. 3 is a block diagram illustrating a main configuration of a control system of the robot tractor 1.

  First, a description will be given of a robot tractor (hereinafter, may be simply referred to as a “tractor”) 1 which is an embodiment of an agricultural work vehicle according to the present invention. As shown in FIGS. 1 and 2, the tractor 1 includes a traveling body (vehicle body) 2 that travels in a field by itself. A working machine 3 is detachably provided on the traveling machine body 2. The work machine 3 is used for farm work. The working machine 3 includes, for example, various working machines such as a cultivator, a mower, a plow, a fertilizer, a sowing machine, and the like. 2 can be attached. In the present embodiment, a rotary tiller is used as the working machine 3. The traveling machine body 2 is configured to be able to change the height and posture of the mounted work machine 3.

  The configuration of the tractor 1 will be described in detail with reference to FIGS. As shown in FIG. 1, the traveling body 2 of the tractor 1 has a front portion supported by a pair of left and right front wheels 7, 7, and a rear portion supported by a pair of left and right rear wheels 8, 8.

  A hood 9 is arranged at the front of the traveling body 2. An engine 10 which is a drive source of the tractor 1 is accommodated in the hood 9. The engine 10 can be configured by, for example, a diesel engine, but is not limited thereto, and may be configured by, for example, a gasoline engine.

  A cabin 11 for an operator to board is arranged behind the hood 9. Inside the cabin 11, a handle 12 for an operator to perform a steering operation, a seat 13 on which the operator sits, and various operation devices for performing various operations are mainly provided. However, the agricultural work vehicle is not limited to the one with the cabin 11 and may be one without the cabin 11.

  Examples of the operating device include a monitor device 14, a throttle lever 15, a shift lever, a PTO switch 17, a PTO shift lever 18, and a plurality of hydraulic shift levers 16 shown in FIG. These operating devices are arranged near the seat 13 or near the handle 12. The monitor device 14 is configured to be able to display various information of the tractor 1. The throttle lever 15 is for setting the rotation speed of the engine 10. The shift lever is used to change the speed ratio of the transmission 22. The PTO switch 17 is for switching between transmitting and shutting off power to a not-shown PTO shaft (power take-out shaft) projecting from the rear end of the transmission 22. That is, when the PTO switch is ON, power is transmitted to the PTO shaft, the PTO shaft rotates, and the work implement 3 is driven. On the other hand, when the PTO switch is OFF, power to the PTO shaft is cut off. The PTO shaft does not rotate, and the work implement 3 is stopped. The PTO speed change lever 18 changes the power input to the work implement 3 and, more specifically, changes the rotation speed of the PTO shaft. The hydraulic shift lever 16 can switch a hydraulic external take-out valve (not shown).

  At the front of the armrest 19 disposed on the right side of the seat 13, operating devices such as a main shift lever 27 and a work implement elevating switch 28 are provided.

  The main transmission lever 27 is for changing the traveling speed of the tractor 1, and is configured so that the traveling speed increases when the main transmission lever 27 is moved forward, and decreases when the main transmission lever 27 is moved backward. The main transmission lever 27 is configured to be capable of stepless operation, and the traveling speed of the tractor 1 is steplessly shifted according to the operation amount of the lever.

  The work implement lifting switch 28 is configured as an electric switch provided on the main shift lever 27 and capable of operating up and down, and is used when the work implement 3 is moved up and down. Thereby, the work implement 3 can be lowered to start the tilling work by the tilling claw 25, or raised to end the tilling work.

  As shown in FIG. 1, a chassis 20 of the tractor 1 is provided below the traveling body 2. The chassis 20 includes a body frame 21, a transmission 22, a front axle 23, a rear axle 24, and the like.

  The body frame 21 is a support member at the front portion of the tractor 1 and supports the engine 10 directly or via a vibration isolation member. The transmission 22 changes the power from the engine 10 and transmits the power to the front axle 23 and the rear axle 24. The front axle 23 is configured to transmit the power input from the transmission 22 to the front wheels 7. The rear axle 24 is configured to transmit the power input from the transmission 22 to the rear wheels 8.

  As shown in FIG. 3, the tractor 1 includes a control unit 4 for controlling the operation (forward, backward, stop, turn, etc.) of the traveling machine body 2 and the operation (elevation, drive, stop, etc.) of the work implement 3. Prepare. A governor device 41, a transmission 42, a lifting actuator 44, and the like are electrically connected to the control unit 4.

  The governor device 41 adjusts the rotation speed of the engine 10. By controlling the governor device 41 by the control unit 4 and appropriately adjusting the rack position, the rotation speed of the engine 10 can be set to a desired rotation speed.

  The transmission 42 is, for example, a hydraulic swash plate type continuously variable transmission that is provided in the transmission 22. By controlling the transmission 42 by the control unit 4 and appropriately adjusting the angle of the swash plate (not shown), the transmission 22 can have a desired transmission ratio.

  The raising / lowering actuator 44 moves the working machine 3 to a retreat position (a position where agricultural work is not performed, for example, a position in FIG. 1) or a working position by operating a three-point link mechanism connecting the working machine 3 to the traveling machine body 2. (Position for performing agricultural work, for example, the position in FIG. 4). In the present embodiment, since the working machine 3 is configured as a rotary tilling device, the agricultural work by the working machine 3 means the tilling work. By controlling the lifting / lowering actuator 44 by the control unit 4 to appropriately raise and lower the working machine 3, farm work can be performed by the working machine 3 at a desired height.

  The tractor 1 including the control unit 4 as described above controls various parts (the traveling body 2, the working machine 3, etc.) of the tractor 1 by the control unit 4 when the operator gets into the cabin 11 and performs various operations. Then, it is configured to be able to perform agricultural work while traveling in a field. In addition, even if the operator does not board the tractor 1, the tractor 1 of the present embodiment travels by instructing forward movement, backward movement, turning, and the like by the remote control device (remote control device) 46 shown in FIGS. It is also possible to cause the tractor 1 to run autonomously.

  Specifically, as shown in FIG. 3, the tractor 1 includes various components in the control unit 4 for enabling autonomous traveling. Further, the tractor 1 has various components such as a positioning antenna 6 necessary for acquiring its own position information based on the positioning system. With such a configuration, the tractor 1 can acquire its own position information based on the positioning system and can autonomously travel on the field.

  Next, the configuration of the tractor 1 for enabling autonomous traveling will be described in detail. Specifically, as shown in FIGS. 1 and 3, the tractor 1 includes a steering actuator 43, a positioning antenna 6, a wireless communication antenna 48, and the like, in addition to the control unit 4.

  The steering actuator 43 is provided, for example, in the middle of the rotation axis (steering shaft) of the steering wheel 12 and adjusts the rotation angle (steering angle) of the steering wheel 12. By controlling the steering actuator 43 by the control unit 4 and operating it appropriately, the steering angle of the steering wheel 12 is set to a desired value, the front wheel 7 which is a steered wheel is turned, and the tractor 1 is turned at a desired turning radius. Can be operated.

  The positioning antenna 6 receives a signal from a positioning satellite constituting a positioning system such as a satellite positioning system (GNSS). As shown in FIG. 1, the positioning antenna 6 is disposed on the upper surface of the roof 92 of the cabin 11 of the tractor 1. The positioning signal received by the positioning antenna 6 is input to the position information calculating unit 49 shown in FIG. 3, and the position information of the tractor 1 (strictly speaking, the positioning antenna 6) is For example, it is calculated as latitude / longitude information. The position information calculated by the position information calculation unit 49 is input to the control unit 4 and used for autonomous traveling.

  Here, as a specific example of the positioning system, there is a satellite positioning system using GPS technology (GPS satellite). Instead, a system using another satellite such as a quasi-zenith satellite or a Glonass satellite is used. It is also possible. In addition, as a positioning system utilizing GPS technology, single positioning, relative positioning, DGPS positioning, RTK-GPS positioning, or the like can be adopted.

  The wireless communication antenna 48 receives a signal from the remote operation device 46 or transmits a signal to the remote operation device 46. As shown in FIG. 1, the wireless communication antenna 48 is arranged on the upper surface of the roof 92 of the cabin 11 of the tractor 1. The signal from the remote control device 46 received by the wireless communication antenna 48 is subjected to signal processing by the wireless communication unit (communication unit) 40 shown in FIG. A signal transmitted from the control unit 4 to the remote operation device 46 is signal-processed by the wireless communication unit 40, transmitted from the wireless communication antenna 48, and received by the remote operation device 46.

  The remote control device 46 is configured as a tablet-type personal computer having a touch panel 39 as shown in FIG. The operator can confirm by referring to information displayed on the display 37 of the remote control device 46 (for example, information on a field necessary for performing autonomous traveling). In addition, the operator operates the touch panel 39 or the hardware key 38 arranged beside the display 37 to transmit a control signal for controlling the tractor 1 to the control unit 4 of the tractor 1. be able to. Note that the remote control device 46 is not limited to a tablet-type personal computer, but may be constituted by, for example, a notebook-type personal computer. Alternatively, when the manned tractor 1x travels along with the unmanned tractor 1 as shown in FIG. 7, the monitor device 46x mounted on the manned tractor 1x may be a remote control device.

  The tractor 1 configured as described above can perform work by the work implement 3 while autonomously traveling along a work path on a field based on an instruction of an operator using the remote control device 46.

  Next, the lifting and lowering of the working machine will be described with reference to FIGS. FIG. 4 is a side view showing a state where the work machine 3 is lowered to the work position from the state of FIG.

  As shown in FIG. 1, a working machine 3 is mounted on a rear part of a traveling body 2 of a tractor 1. As described above, a part of the driving force of the engine 10 is transmitted to the work machine 3 via the PTO shaft, and the work machine 3 can be driven to perform tilling work. In the lower part of the work machine 3, a tilling claw 25 that is driven to rotate about a shaft 26 that is arranged horizontally is provided. By lowering the work machine 3 to the work position shown in FIG. 4, the rotating tillage claw 25 comes into contact with the soil, and the tillage work in the field can be performed. Also, by raising the work implement 3 to the retracted position shown in FIG. 1, the tilling claw 25 can be detached from the soil, and the tilling work can be stopped. The lifting and lowering of the work machine 3 can be performed by an operator operating the work machine lift switch 28, and the control unit 4 can also automatically control the work machine 3.

  Next, the learning function of the control unit 4 will be described with reference to FIGS. 3, 5, 8, and 9. FIG. 8 is a plan view showing how the robot tractor 1 in the manual operation mode travels on the field 61 by an operation of the operator. FIG. 9 is a plan view showing how the robot tractor 1 in the automatic operation mode performs work while traveling unattended, based on the learning result in the manual operation mode in FIG.

  The control unit 4 of the present embodiment is configured to learn the operation of the tractor 1 performed by the operation of the operator, and to cause the tractor 1 to automatically operate based on the learned operation. The configuration will be described below.

  As illustrated in FIG. 3, the control unit 4 includes an operation mode switching unit 70, a learning unit 80, and an automatic operation control unit 71.

  The operation mode switching unit 70 is for switching between the automatic operation mode and the manual operation mode of the tractor 1. For example, the control unit 4 has a RAM (not shown), and an automatic operation mode and a manual operation mode are switched by switching an operation mode flag stored in the RAM. A specific switching operation of the operation mode can be realized by the operator appropriately operating the remote operation device 46. For example, when the operator touches the mode switching button displayed on the display 37 of the remote operation device 46 with a finger or the like, the remote operation device 46 transmits a mode switching signal, and transmits the mode switching signal via the wireless communication unit 40. The received operation mode switching unit 70 changes the operation mode flag to switch between the automatic operation mode and the manual operation mode.

  The learning unit 80 is configured to be able to learn the operation of the tractor 1 operated by the operator in the manual operation mode. The specific configuration of the learning unit 80 will be described later.

  The automatic operation control unit 71 is configured to allow the tractor 1 to perform automatic operation based on the operation of the tractor 1 learned by the learning unit 80.

  The learning unit 80 includes a motion prediction model 81 and a teacher data storage unit (storage unit) 82. The learning unit 80 operates based on the teacher data (data related to the field 61 and data related to the operation of the tractor 1 when the operator operates the tractor 1 in the manual operation mode in the field 61) stored in the teacher data storage unit 82. The prediction model 81 can be made to learn.

  The teacher data storage unit 82 includes an input case data storage unit 85 and an output case data storage unit 86.

  The input case data storage unit 85 can store input case data input by an operator. The input case data is data on the field 61 and includes information on the shape of the field 61.

  Note that the information on the shape of the field 61 can be expressed as, for example, a set of position information of each vertex of a polygon approximating the shape of the field 61a shown in FIG. The position information indicating the shape of the field 61 can be obtained by, for example, sequentially acquiring the position information by the position information calculating unit 49 while running the tractor 1 along the contour of the field 61, It can also be obtained by operating the device 46 and sequentially inputting position information.

  The input case data includes information (work information) on the work content. The work information may include, for example, the type (model) and size of the work machine. The work information can include various types of information for each work machine.

  The output case data storage unit 86 can store output case data which is data relating to the operation of the tractor 1 operated by the operator in the manual operation mode. The output case data includes information on the work route of the tractor 1 (information on the route (travel route 65) in which the tractor 1 traveled in the field 61, and information on the lifting operation of the lifting actuator 44).

  The information on the traveling route 65 can be expressed, for example, as a change in the position of the tractor 1 that has actually traveled in the manual operation mode (a set of position information calculated by the position information calculation unit 49). In addition, after simplifying the traveling route 65 by a straight line and an arc, it can be expressed as position information of the start point and the end point of the straight line, the start point and the end point of the arc, and the position information of the center of the arc.

  In addition, information on the lifting / lowering operation of the lifting / lowering actuator 44 (the lifting / lowering timing of the work implement 3) is obtained by detecting the position of the tractor 1 on the traveling path 65 and the height of the work implement 3 at the position by using an appropriate sensor. And information in which the detected values are associated with each other.

  After the operator inputs the shape and the like of the field 61, when the operator operates in the manual operation mode in the field 61, information on the shape and the like of the field 61 is stored in the input case data storage unit 85, and the tractor 1 operated by the operator is operated. Is detected by the position information calculation unit 49 and various sensors (not shown), and stored in the output case data storage unit 86.

  As shown in FIG. 5, the motion prediction model 81 learns based on teacher data including a set of input case data and output case data stored in the teacher data storage unit 82. More specifically, the above input case data is represented as a multidimensional vector, but if it is used for learning as it is, the amount of information is too large and learning cannot be completed in a realistic time. By performing a process of extracting only essential features (feature extraction process) on the input case data, a vector with a reduced number of dimensions (a multidimensional feature vector) is obtained. By providing a large number of sets of the feature vector and the multidimensional vector representing the output case data to the motion prediction model 81, learning can be performed.

  The learned motion prediction model 81 can output a multidimensional vector corresponding to the input of the feature vector when the feature vector is input. The control unit 4 controls the traveling of the tractor 1 and the elevating operation of the elevating actuator 44 based on the information output from the operation prediction model 81.

  With the above configuration, when the learning unit 80 performs the work again in the field 61 once learned, the tractor 1 can be automatically driven on the same route as the travel route 65 operated by the operator in manned traveling.

  Further, even when an unknown feature vector is given, the learned motion prediction model 81 can estimate and output a multidimensional vector corresponding to the unknown feature vector (generalization ability). Therefore, even when the shape of the unknown field 61b as shown in FIG. 9 is given to the tractor 1 in the automatic operation mode, another field 61 (for example, a shape similar to the above-described field 61b as shown in FIG. 8). In the field 61a), flexible automatic operation of the tractor 1 can be realized based on the result of learning the operation of the operator (the operation of the tractor 1) in the past.

  As described above, the control unit 4 of the present embodiment can learn the operation of the tractor 1 operated by the operator, and can automatically drive the tractor 1 based on the learned operation. This allows the tractor 1 to automatically perform traveling of the tractor 1 and elevation of the elevation actuator 44 according to the traveling route 65 which is closer to when the operator actually drives. As a result, it is possible to realize an automatic operation that precisely responds to the individual requirements of the user that differs for each actual field 61 and each operation.

  The output case data stored in the output case data storage unit 86 includes not only the traveling path 65 of the tractor 1 operated by the operator and the lifting / lowering operation of the lifting / lowering actuator 44, but also the braking operation of the tractor 1 (the left and right wheels). Information on the vehicle speed set on the tractor 1, information on the engine speed set on the tractor 1, and information on the PTO shaft (power of the engine 10 not shown). And information on the driving state of the mechanism (mechanism for transmitting to the machine 3). Note that these pieces of information can be acquired from detection values of various sensors attached to the tractor 1 or setting values input to the control unit 4 and the like. Thereby, the control unit 4 can learn the operation of the operator including the specific operation such as the braking operation, and can make the tractor 1 autonomously travel. These pieces of information can be associated with the position of the tractor 1 on the traveling route 65 in the same manner as the information on the elevating operation of the elevating actuator 44.

  The output case data stored in the output case data storage unit 86 includes, in addition to the above information, information on the vehicle speed of the tractor 1 adjusted according to the inclination of the field 61 and the slip rate of the tractor 1 The information on the vehicle speed of the tractor 1 adjusted as described above, the information on the engine speed adjusted according to the change in the engine load factor, the information on the vehicle speed adjusted according to the change in the engine load factor, and the engine load Information about the elevating operation of the elevating actuator 44 corresponding to the change in the rate is included. Note that these pieces of information can be obtained from detection values of various sensors attached to the tractor 1 and the like. That is, conventionally, the operator has to respond to a specific situation such as approaching the inclined portion of the field 61 during running, slipping the front wheel 7 or the rear wheel 8 or increasing the load factor of the engine 10. Based on the experience, adjustment of the vehicle speed, adjustment of the engine speed, elevating and lowering of the elevating actuator 44, and the like were performed, and the operation required a high degree of skill. In this regard, the control unit 4 of the present embodiment can appropriately control the vehicle speed, the engine speed, and the like during autonomous traveling of the tractor 1 by learning, for example, the operations actually performed by an experienced operator.

  The input case data stored in the input case data storage unit 85 includes not only the shape of the field 61, but also the inclination of the field 61, the shape of the headland 63 in the field 61, trees and swamps in the field 61, and the like. Information on obstacles, the quality of soil in the field 61, and the like are included. Note that these pieces of information can be obtained from, for example, information input by an operator operating the remote control device 46. Thus, the control unit 4 controls the traveling of the tractor 1 and the lifting / lowering operation of the lifting / lowering actuator 44 along the traveling route 65 closer to the case where the operator actually performs the tractor while appropriately responding to the detailed situation of the field 61. 1 can be performed automatically.

  Next, how the control unit 4 learns the operation of the tractor 1 in a part of the field 61 and applies it to the operation of the tractor 1 in the entire field 61 will be described with reference to FIGS. 10 and 11. FIG. 10 is a plan view showing how the robot tractor 1 in the manual operation mode learns the operation of the operator on a part of the field 61. FIG. 11 is a plan view showing a state in which the robot tractor 1 in the automatic operation mode performs work while traveling unattended in the remaining area of the field 61 based on the learning result in the manual operation mode in FIG.

  FIG. 10 shows an example of a field 61c having a slightly complicated shape. In the field 61c, the shapes of the upper right portion where manned traveling has been performed and the lower left portion where no work has been performed are relatively similar. An intermediate portion connecting the upper right and lower left portions of the field 61c is a simple rectangle.

  In such a field 61c, the operator sets the tractor 1 in the manual operation mode, inputs the shape of the field 61 only in the upper right portion in FIG. 10, and performs the work by manually moving the tractor 1 only in that portion. After that, the operator switches to the automatic operation mode, inputs information such as the shape of the middle part and the lower left part of the field 61 to the input case data storage unit 85, and sends the information to the tractor 1 according to the output of the motion prediction model 81 corresponding to the input. Work can be performed automatically. As described above, by manipulating a part of the area of the field 61c and learning the motion prediction model 81, the remaining area is very similar to that performed by the operator at the time of learning, as shown in FIG. It is expected that traveling and work will be performed.

  That is, since it is possible to learn only a part of the field 61 and apply it to the entire field 61, it is possible to reduce the time and man-hours required for the tractor 1 to learn by performing manned traveling.

  Next, FIG. 12 illustrates an example in which the control unit 4 included in the tractor 1 acquires data stored in the control unit 4 included in another tractor 1a by wireless communication and uses the acquired data as data for learning the motion prediction model 81. It will be described using FIG. FIG. 12 is a block diagram illustrating a main configuration of a control system of the tractor 1 when the control unit 4 acquires data learned by the control unit 4 included in another tractor 1a by wireless communication.

  In the example shown in FIG. 12, a tractor 1a different from the tractor 1 is provided with a control unit 4a. Data obtained by the operator operating in the manual operation mode in each of the tractor 1 and the tractor 1a is wirelessly transmitted and received between the two tractors 1 and 1a via the wireless communication antenna 48. be able to. Although not shown, the control unit 4a provided in the tractor 1a also has a learning unit 80 (an operation prediction model 81, an input case data storage unit 85, and an output case data storage unit 86).

  The data exchanged between the control units 4 and 4a may be the teacher data stored in the input case data storage unit 85 and the output case data storage unit 86, or may represent the motion prediction model 81 in which learning based on the teacher data is completed. Data may be used. In general, in order to construct an accurate motion prediction model 81 in the learning unit 80 of the tractor 1, it is necessary to provide a large number of teacher data to the motion prediction model 81 for learning. By utilizing the obtained teacher data, the time and man-hours required for data acquisition can be reduced. Further, when the configuration is such that data of the learned motion prediction model 81 is exchanged, an excellent motion prediction model constructed outside can be obtained, and the time and effort required for learning can be reduced. It is preferable because it can be performed.

  When data relating to learning is exchanged between the plurality of tractors 1 and 1a, information for specifying the model of the tractor 1, information for specifying the model of the work machine 3, and attachment of the positioning antenna 6 to the tractor 1 are provided. It is preferable that information on the position and the like be included in the teacher data (input case data) given to the motion prediction model 81. That is, the elevating operation of the work implement 3 in the tractor 1 differs depending on the performance of the elevating actuator 44 provided in the tractor 1, the weight of the work implement 3 and the like, and the axis of the tilling claw 25 of the work implement 3 from the positioning antenna 6. The distance up to 26 (the distance L shown in FIG. 1) differs depending on the model of the work machine 3, the mounting position of the positioning antenna 6, and the like. Therefore, by performing learning including the above information, it is possible to construct a highly general-purpose motion prediction model 81 applicable to various tractors, and to share learning results among a plurality of tractors. Becomes easier.

  As described above, the tractor 1 of the present embodiment includes the traveling machine body 2, the working machine 3, the teacher data storage unit 82, and the control unit 4. The work machine 3 is mounted on the traveling machine body 2. The work machine 3 is used for agricultural work. The teacher data storage unit 82 stores the field information including the information indicating the shape of the field 61, the past traveling information and the past work in at least a part of the field 61 or at least a part of another field 61. Information can be stored. The control unit 4 can generate a work route in the field 61 based on the field information, the past travel information, and the past work information.

  Thus, the operation by the automatic operation can be performed more flexibly based on the operation actually performed by the operator.

  Further, the tractor 1 of the present embodiment includes a wireless communication unit 40 capable of communicating with another tractor 1a. The control unit 4 obtains at least one of the field information, the past travel information, and the past work information from another tractor 1 a via the wireless communication unit 40, and obtains another information via the wireless communication unit 40. It can also be provided to the tractor 1a.

  Thereby, information for realizing good automatic driving can be easily shared with another tractor 1a.

  The preferred embodiment of the present invention has been described above, but the above configuration can be modified as follows, for example.

  The remote control device 46 of the above embodiment is configured to be able to remotely monitor and operate the work by the autonomous traveling of one tractor 1 with respect to one remote control device 46, but is not limited thereto. The tractor 1 may be configured to be monitored and operated remotely.

  In the above embodiment, an example is shown in which data relating to learning is transmitted and received between the two tractors 1 and 1a. However, the number of tractors 1 is not limited to two and may be any number. For example, the control unit 4 may be connected to a WAN such as the Internet via wireless communication, and may exchange learning-related data with a learning unit 80 provided in the control unit 4 of a distant tractor. In addition, a database that can store learning-related data may be constructed in a server connected to the Internet, and data may be acquired from the database. As described above, as the number of tractors sharing data increases, it becomes easier and easier to realize good automatic driving of the tractor 1.

  The work route may be created without using the machine learning by the learning unit 80. For example, when the shape of the field is almost the same in the past and this time, the work route may be created by simply copying all the past travel routes. Alternatively, a work route may be created by repeatedly arranging a route of a predetermined repetition unit (for example, a straight route and a turning route for one round trip) of the past traveling routes. In this case, when a plurality of traveling routes in the past traveling route appear in the repetition unit, an average traveling route obtained by averaging the turning radii and the like is used as a repeating unit when a new work route is created. May be used.

  When a special shape appears in a part of the shape of the field (for example, a circuit) or when an obstacle is arranged in the field, a portion requiring such special operation (for example, the reverse movement of the tractor 1) For example, only for a turn that does not involve the above or a turn that involves the backward movement of the tractor 1, a work route may be created based on past travel information and past work information.

  Instead of using machine learning, past traveling information on other tractors, past work information, and field information at that time may be received, and a work route may be created based on the received information. Specifically, the type and size of the work machine 3 are the same as those in the case where another tractor has worked in the past, and the difference in the shape of the field is also within a preset threshold range (when the shape of the field is When it can be determined that they are substantially the same), it is conceivable to create a work route by, for example, simply copying a travel route when working with another tractor. Further, in order to be able to be used by another tractor, past traveling information on the own machine, past work information, and field information at that time can be transmitted to the other tractors as necessary so that the tractors can be provided to the other tractors. Is preferred.

  When the tractor 1 travels in the field 61a based on the data on learning acquired from the tractor 1a, the tractor 1 travels not only based on the data on learning in the entire field 61a acquired from the tractor 1a, but also travels on the field 61a acquired from the tractor 1a. Based on any of the data relating to learning in a part of the field 61a, the data relating to learning in the entire other field 61b obtained from the tractor 1a, and the data relating to learning in a part of the other field 61b obtained from the tractor 1a. Can also be run. That is, the tractor 1 acquires the input case data and the output case data when the other tractor 1a travels only on a part of the field 61a, and the operation prediction model 81 learns the learning. It can be applied when the tractor 1 runs. Further, for example, the tractor 1 acquires input case data and output case data when the other tractor 1a travels only in a part of the field 61b, and the operation prediction model 81 learns the learning result. The present invention can be applied to a case where the tractor 1 travels over the entire other field 61a including a portion having a similar shape.

  Although the output example data is data relating to the operation of the tractor 1 operated by the operator in the manual operation mode, it is data relating to the operation of the tractor 1 in the automatic operation mode and data relating to the operation of the other tractor 1a in the automatic operation mode. There may be.

  The present invention is not limited to being applied to a tractor as in the above-described embodiment, but can be applied to, for example, a rice transplanter. In this case, the planting unit (planting device) corresponds to a work machine. Further, the present invention can be applied to various other agricultural work vehicles.

1,1a Robot tractor (work vehicle)
2 traveling body (body part)
3 work machine 4, 4a control unit 40 wireless communication unit (communication unit)
61 (61a, 61b, 61c) Field 82 Teacher data storage unit (storage unit)

Claims (2)

Field information including outer shape information of a non-rectangular deformed field including a work area in which a predetermined work is performed by straight traveling of the work vehicle and a headland area that allows the work vehicle to turn, and a partial area of the field A storage unit capable of storing first work route information, which is information on a first work route in which the work vehicle is manually traveled by an operator,
A control unit capable of generating a second work route capable of autonomous traveling in a remaining region excluding the partial region in the field based on the field information and the first work route information;
With
The second work route includes a plurality of straight-ahead routes,
The work route generation system, wherein the control unit is capable of adjusting a length of the straight traveling route of the second work route so that a width of the headland area is constant. The work route generation system according to claim 1,
The first work path includes a straight path,
The length of the straight path of the first work path and the second work path may be different depending on the shape of the field, respectively.

Images