Learning method and system for a driving route for autonomously driving an agricultural driving robot

Abstract

The present invention relates to a method for learning a driving route for autonomously driving an agricultural driving robot (1). The method includes the following steps. A step of providing an environmental map, a step of driving along a route in an area from a starting point to an ending point in the environmental map by manually controlling the driving robot (1), a step of identifying the position of the driving robot (1) while driving along the route with the help of at least one sensor provided on the driving robot (1), a step of recording the coordinates of waypoints (31) along the route based on the executed position identification process, a step of authenticating the route by autonomously driving along the route considering the saved coordinates of the waypoints (31), the route being manually confirmed, and a step of marking the confirmed route as autonomously drivable. The present invention further relates to a system including an agricultural driving robot (1), the agricultural driving robot (1) having a control device for autonomous driving and a remote control connected to the control device and intended to manually control the driving robot (1). The system is characterized in that the control device is configured to execute the above method.

Description

Technical Field

[0001] The present invention relates to a method for learning a driving route for autonomously driving an agricultural driving robot, particularly in a livestock shed and on a farm premises, and further to a system for autonomously driving an agricultural driving robot that enables learning of a driving route.

Background Art

[0002] Many operations in livestock sheds and on farm premises are related to the transportation of materials. For example, a feeding system is often used to feed livestock. In a feeding system, the amount of feed to be given to livestock is made by mixing various basic components in a central area, i.e., a so-called "kitchen," as needed and at an appropriate time, and is given to livestock by being distributed through a so-called "feed array." Another example relates to the removal of livestock excrement. Cleaning of the farm premises or livestock shed area is usually carried out using a vehicle according to the size of the area.

[0003] In the agricultural field, in order to perform these operations as autonomously as possible and with minimal use of human labor, automated systems and devices for these various applications have been established.

[0004] For example, an autonomous feeding system for livestock such as cows is known from International Publication WO2008 / 097080A1. The central component of this system is an autonomously driving vehicle, which has a feed container that can be automatically filled in a central so-called "kitchen area." Feed can be mixed during the movement from the feed container to the unloading point of the feed. At the unloading point, feed is automatically supplied by tilting the container. Various options are described for the movement of the vehicle. For example, the route is predetermined by pre-laid rails. Another option described is autonomous navigation using sensors or route marking. Navigation based on a wireless positioning system such as GPS (Global Positioning System: Global Positioning System) is also described.

[0005] In particular, when the route is not defined by a laid track or other routing, a learning procedure is required to enable navigation in a livestock shed or on a farm premises.

[0006] International Publication WO2018 / 074917A2 discloses such a learning method for autonomously operating an agricultural driving robot. In this method, in the first step, an external mobile device, for example, a tablet computer, is used to manually control an autonomous driving robot along a desired route. During manual control, sensor data recorded by the driving robot is transmitted to the mobile device and displayed on a map together with the position of the driving robot on the mobile device. The transmitted sensor data includes, for example, the measured current distance to an object. The traveled route is saved and then can be automatically traveled by the driving robot in an autonomous navigation process.

[0007] Displaying additional information on a mobile device used as a remote control can be helpful, but it may also distract the user from carefully observing the driving robot during the learning drive.

Prior Art Documents

Patent Documents

[0008]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0009] The present invention is a learning method for an agricultural driving robot, which can learn a route intuitively and with complete concentration on the driving robot, can further check the route, and can correct the route as needed, for example, regarding a safety distance. The purpose is to create a learning method.

Means for Solving the Problems

[0010] The above object is solved by a method or a system having the features described in the respective independent claims. Advantageous designs and further developments are the subject matter of the dependent claims.

[0011] The method according to the present invention of the type described above has the following steps. An environmental map is provided, and a driving route extending in the area within the environmental map is manually driven by the driving robot from the starting point to the end point under the manual control of the user. During the driving of the driving route, the driving robot is located using at least one sensor provided on the driving robot, and the coordinates of the waypoints along the driving route are recorded based on the location determined above. In the next step, the driving route is autonomously driven by the driving robot in consideration of the saved coordinates of the waypoints. In this case, the driving route is manually confirmed by the user. After this step of authenticating the driving route, the driving route is marked as confirmed and can thus be used for autonomous driving thereafter.

[0012] In this method, the learning of the driving route, hereinafter abbreviated as the route, is performed by a controlled manual operation independent of the authentication of the route. The authentication is performed during the new operation. When the desired route is driven in a manually controlled manner, it is possible to fully concentrate on the route itself and the driving robot during driving. When the route is automatically driven for the first time for authentication thereafter, the user can concentrate directly around the driving robot or on the displayed measurement values, and can check, for example, the distance to be maintained. Only when the route is confirmed by the user, the route is marked as autonomously drivable and can then be used for independent autonomous navigation. By dividing the learning process into a step of manually specifying the route and then a step of authenticating it, the user is not overburdened with attention, and the learning process itself can be executed without the risk of collisions or other errors.

[0013] In a beneficial design of this method, the driving robot is manually advanced to the starting point before the authentication step, and in the authentication step, the route is autonomously driven in the same direction as the direction specified in the manually driven step. The route is usually learned in the direction in which it will be driven later. Since the route is authenticated in the same direction as it was learned in the manually driven step, the authentication is also performed in the direction of the next use, which is therefore practical. Alternatively, it is also conceivable to authenticate the route in the direction opposite to the direction specified in the manually driven step. If necessary, especially when the route is used in both directions during autonomous and productive operation, both directions of travel can be authenticated.

[0014] In a further beneficial design of this method, the route is confirmed part by part during authentication. Active or passive confirmation may be required.

[0015] In active confirmation, it is considered that the route or a part of the route is confirmed when the user acts actively during the authentication step. Active action can, for example, be to operate the driving robot itself or a control element on the remote control. Remote controls have also been preferably used before to control a driving robot to manually drive a route. Such a remote control is, for example, wirelessly connected to the control device of the driving robot.

[0016] In passive verification, it is considered that a route or a part of the route has been verified when there is no user intervention to correct or stop the movement of the driving robot during the authentication process.

[0017] In a further advantageous design of the method, during manual and / or automatic driving of the route, the distance to surrounding objects is measured via at least one distance sensor. The distance sensor can be a sensor used for navigation or positioning and / or a sensor independent thereof. Preferably, the measured distance to surrounding objects is compared with a predetermined or pre-determinable safety distance. In this case, when the distance is shorter than the safety distance, an acoustic and / or visual warning is issued. While the route is being learned, i.e., while the route is pre-defined by manual driving, it is preferable that additional information such as the measured distance to an obstacle is not displayed on the user's remote control to distract the user. However, in both learning and authentication, it is possible to indicate that the safety distance to a detected obstacle is preferably not reached by an acoustic and / or visual warning signal provided on the driving robot itself. This prevents the possibility of a collision without distracting the user's attention from the position and movement of the driving robot. Furthermore, if an infringement of the safety distance is detected, the authentication of the route element can be blocked.

[0018] In a further advantageous design of the method, at least one marking point can be defined by the user during the manual driving of the route while the driving robot is stationary. One or more functions to be executed can be assigned to such marking points. The driving robot then executes the function at or from the marking point during the next movement. The marking point can also represent the starting point or the end point of at least one route.

[0019] In a further advantageous design of the method, it is possible to change the route in whole or in part manually or automatically by the user by changing the coordinates of the waypoints before the step of authentication. Subsequently, manual intervention can be used to affect the routing without the need to newly drive it manually in full. In this case, the marking point may be immovable. However, the option of making manual corrections to the marking point can also be considered, for example only after (additional) authorization and / or approval by the user, or by allowing only minor changes to the recorded coordinates, in which case the security status is probably also high.

[0020] The main purpose of the automatic correction is to smooth the route or a part of the route. Recorded route elements including several adjacent waypoints are converted into smoothed route elements using a mathematical function.

[0021] For example, in order to reduce the "deviation" during movement, the recorded waypoints can be shifted using a filter algorithm, in particular a low-pass filter. Curve smoothing can also be achieved by completely recalculating the positions of the waypoints of the route element using an appropriate curve modeled with parameters. In this case, the recorded positions of the waypoints are not considered in the recalculation, and the root element is determined only by the position of the end point (usually the mark point). A hybrid form can also be considered in which the recorded positions of the waypoints are considered in the recalculation with adjustable weighting. In order to change the coordinates of the waypoints in both manual and automatic correction of the waypoint positions, the respective boundary conditions can be specified. Such boundary conditions are related to the longest movement, the transition to the next waypoint, for example, preventing a bent connection to a subsequent or previous waypoint, and / or the shortest distance to the object to be maintained.

[0022] The system according to the present invention, which includes autonomously driving an agricultural driving robot and a remote control for manually controlling the driving robot, is set to execute the above method. As a result, the advantages described above in relation to the above method are obtained.

[0023] The present invention will be described in more detail below with reference to the drawings according to exemplary embodiments. The drawings are as follows.

Brief Description of the Drawings

[0024]

Figure 1a

Figure 1b

Figure 2

Figure 3a

Figure 3b

Figure 3c

Figure 3d

Figure 3e

Figure 3f

Figure 4

Figure 5

DETAILED DESCRIPTION OF THE INVENTION

[0025] FIG. 1a and FIG. 1b are diagrams showing a schematic view of an example of the agricultural work driving robot 1 from different angles. The route learning method described below can be executed by, for example, the driving robot 1.

[0026] In this embodiment, the driving robot 1 is a so-called "feeding robot", which is set to pick up food from a supply point, automatically mix it, and unload it at one or more feeding points. Therefore, hereinafter, the driving robot 1 will also be referred to as the "feeding robot" or simply the "robot".

[0027] In all the drawings, the same reference numerals indicate components having the same or similar effects. For the sake of simplicity, reference numerals are not assigned to all components in all the drawings. In the following description, the terms "right" and "left" indicate the right and left in the drawings, respectively. On the other hand, the terms "up" and "down" indicate the up and down in the natural orientation of the driving robot. The terms "front" and "rear" mean the forward direction in the driving direction 10 of the driving robot 1. The forward direction in the driving direction 10 is indicated by an arrow in FIGS. 1a and 1b and represents the main driving direction of the driving robot 1.

[0028] The driving robot 1 has two main components, namely a chassis 100 and a body 110.

[0029] The chassis 100 is preferably applicable to anything and can be used with various functional units if necessary. In FIGS. 1a and 1b, only clad elements and / or protective elements can be seen on the chassis 100, specifically the enclosure apron 103 and two bumpers 104. Further, one of the four wheels, specifically one of the two drive wheels 101 (FIGS. 1a and 1b) and one of the two swivel wheels 102 (FIG. 1b) can be seen. The other swivel wheel is located at the front part in the forward direction of the driving direction 10 and is hidden under the apron 103 in FIGS. 1a and 1b. In this exemplary embodiment of the feeding robot, the apron 103 also functions as a feed pusher, by which the already unloaded feed can be pushed out together.

[0030] The body 110 essentially determines the function of the driving robot and thus determines the intended usage of the driving robot in the livestock shed and on the farm premises.

[0031] In the case of the driving robot 1 provided as a feeding robot as in this case, the body 110 has a feed container 111 as the main component. The feed to be distributed is placed in the feed container 111 and can be mixed during charging at the charging station 28 (see FIG. 2) and / or during movement using the mixing device 112. The mixing device 112 is not visible in FIGS. 1a and 1b. A feed conveyor 113 is provided for supplying feed. The feed conveyor 113 operates with the aid of a conveyor belt. Depending on the running direction of the conveyor belt, the feed can be supplied to either side of the feeding robot. The structures of the feed container 111 and the feed conveyor 113 represent functional units of the driving robot 1. This is because they provide the specific functions of the driving robot 1 and thus define the driving robot 1 as a feeding robot.

[0032] The body 110 further includes a clad part formed using a plurality of clad elements, typically a clad panel 114. The clad panel 114 is preferably removable as a separate part to access the components below and for maintenance and replacement of those components. Elements accessible from the outside are integrated into the clad part, for example, the charging contact 115 (see Fig. 1a) and the operation and / or display element 116 (see Fig. 1b). The driving robot 1 is set to automatically enter the charging station 28 (see Fig. 2), and within the charging station 28, the charging contact 115 comes into contact to recharge the battery or other power storage device of the driving robot 1.

[0033] The driving robot 1 further includes a navigation system that enables navigation in a livestock shed or on a farm premises even without fixed infrastructure elements such as rails or guide cables. For this purpose, the driving robot is equipped with a plurality of sensors integrated into or protruding from the clad part.

[0034] Figs. 1a and 1b show two LiDAR (Light Detection and Ranging) sensors 117. These are used to detect objects and support navigation. Two LiDAR sensors 117 are respectively provided at the front and rear of the driving robot. Alternatively or in addition to this, an optical camera can also be provided at the front in the traveling direction and, if necessary, at the rear. In that case, the camera is used for object or step detection, or to provide additional support for navigation. In order to enable recording and monitoring of the ground immediately in front of the driving robot 1 in both directions of the traveling direction (i.e., when driving forward and backward), it is also possible to tilt the camera downward. Furthermore, ultrasonic sensors 118 are distributed in the lower region of the clad portion around the driving robot 1 to serve as distance sensors for nearby obstacles.

[0035] Other sensors not visible here are mechanical sensors that detect the application of force to one or both of the bumpers 104. For this purpose, the bumpers 104 can each be removably attachable. For example, when moving against the spring force, one of several possible sensors is activated. In another design, the bumper 104 can be formed in the outer region with an elastically deformable material, particularly a foamed material, in which a sensor is incorporated. This sensor preferably detects deformation along the entire edge of the bumper 104. In this way, the collision with an obstacle is successfully weakened and detected simultaneously. In one design, for example, two electrodes provided at intervals can be embedded in an elastic material along the edge of the bumper 104, and the capacitance between the electrodes is detected. A change in capacitance indicates deformation of the material. In yet another design, a tension chain can be incorporated into an elastic material and connected to a switch or sensor. The tension chain is tensioned by the deformation of the elastic material and this is detected by a switch or a sensor.

[0036] The driving robot 1 has at least one control device which controls the actuators of the driving robot including the drive motor and reads and evaluates signals from sensors. The control device further performs navigation tasks and maintains a stored map of the environment. The map is used, for example, to identify the position of the driving robot in the environment. The map is preferably created by the driving robot 1 itself by evaluating sensor data recorded during various movements in a so-called SLAM (Simultaneous Localization and Mapping) process. Furthermore, the control device comprises or is connected to a communication interface for communication, in particular for wireless communication. The communication interface is used, for example, to connect to a high-level operation management system which coordinates the use of the driving robot 1. The communication interface can also be used to control the driving robot 1 using a remote control.

[0037] Figure 2 is a two-dimensional map of the farm 2 in which the driving robot 1 is used as a feeding robot. In principle, several robots can also be used on the farm to deliver feed from the feed bunkers to the livestock. The driving robot 1 can be designed, for example, as shown in Figures 1a and 1b.

[0038] In the embodiment shown here, the farm 2 has, for keeping livestock, for example, two livestock houses for keeping cows, specifically a first livestock house 20 and a second livestock house 21, and a farm compound surrounding them. In this embodiment, the two livestock houses 20 and 21 are of different sizes, and the second livestock house 21 is farther away from the larger first livestock house 20. In this sense, the first livestock house 20 can be considered the main livestock house. Although there are two livestock houses 20 and 21 on the farm 2, this number is only an example, and their sizes and configurations are also only examples.

[0039] Both the livestock houses 20 and 21 have a wall 22 on the outside and a plurality of columns 23 on the inside. This is only an example. The livestock houses 20 and 21 may have internal walls. Furthermore, each of the livestock houses 20 and 21 is provided with a door 24.

[0040] In each of the two livestock houses 20 and 21, there are provided several livestock areas 25, that is, areas where livestock, such as the cows described above, are raised. In each livestock area 25, a so-called "feed fence" 26 is assigned, and feed is placed in front of it. Livestock can pick this up from the livestock area 25.

[0041] Outside the livestock houses 20 and 21, that is, within the farm area of the farm 2, several feed bunkers 27 are set up, where different types of feed are stored for the livestock. For example, three relatively large feed bunkers 27 are shown. These are used, for example, to store silage of feed. The relatively small feed bunker 27 indicated by a circle in FIG. 1, which is a schematic diagram, is used to store concentrated feed. Each feed bunker 27 has a feeding conveyor, by which the received feed or concentrated feed can be supplied.

[0042] The map of the farm 2 shown in FIG. 2 is schematic, but essentially reflects what the driving robot 1 detects about the environment with the help of the LiDAR sensor 117. The two LiDAR sensors 117 provided on the driving robot 1 scan the environment in a two-dimensional plane aligned parallel to the chassis 100, and thus essentially parallel to the ground on which the driving robot 1 moves. Therefore, the driving robot 1 detects only the features located within the scanning plane of the LiDAR sensor 117 via the LiDAR sensor 117. In the illustrated embodiment, this is at a height of approximately 1.5 to 2.5 m (meters) above the ground. For this reason, only the contour of the feed bunker 27 at this height is detected. For example, a hose or conveyor used to supply picked-up feed or concentrated feed is not seen in this map for this reason. Furthermore, not seen are the grids or other barriers formed around the livestock area 25. The invisible element includes the feed fence 26, and therefore the feed fence 26 is shown by a dashed line in FIG. 2.

[0043] To enable the use of the driving robot 1 on the farm 2, a route, i.e., a path along which the driving robot 1 can be used, is defined by the learning method according to the present invention. The navigation method compiles a route according to the task to be completed from the acquired route during the operation of the driving robot 1. The driving robot 1 moves along this route to enable the performance of its task.

[0044] Hereinafter, the method according to the present invention will be described using the farm 2 shown in FIG. 2 as an example. The driving robot 1 is trained, for example, to be able to use a route extending between the charging station 28, the feed bunker 27, and various feed deposit points in front of the feed fence 26.

[0045] In the illustrated farm 2, the charging station 28 is provided adjacent to the feed bunker 27. The charging station 28 is controlled by the driving robot 1 to charge the battery for energy supply. The charging station 28 includes contacts that come into contact with the charging contact 115 (see FIG. 1a) when the driving robot 1 is correctly positioned. To further charge the battery of the driving robot 1, a charging current is supplied through these contacts. As described above, the driving robot 1 uses the signal from the LiDAR sensor 117 to determine its position as part of navigation. The accuracy of the position can also be improved by using signals from other sensors, such as wheel rotation sensors for odometry.

[0046] Another positioning method can also be used near the charging station 28. In this method, the markings on the charging station are optically detected. Such markings are, for example, the reflectors 29 shown in FIG. 2, and the reflectors 29 are detected by the driving robot 1 using the LiDAR sensor 117 or other optically operating sensors. Near the charging station 28, a higher positioning accuracy can be achieved compared to assisting contour-based navigation, which is the positioning accuracy required for contact.

[0047] First, FIG. 3a shows an enlarged view of the farm 2 in the area of the feed bunker 27 and the charging station 28.

[0048] The first step is performed to prepare the learning method according to this application. In the first step, the driving robot 1 is manually advanced near the charging station 28 and positioned at a distance of about 1 to 2 m from the charging station 28 and in the direction along the charging station 28 in the traveling direction 10. The distance and position are selected so that the charging station 28 can be navigated using the reflector 29.

[0049] To control the driving robot 1, a remote control (not shown here) is used. The remote control preferably communicates wirelessly with the driving robot 1, but this is not necessarily required. For transmission, an optical or wireless-based communication link can be used. In particular, a user's mobile end device that can be used universally, such as a tablet computer, can be used as the remote control. The communication link can be established directly between the tablet computer and the receiving device of the driving robot 1, or via a shared communication network, such as a WLAN (Wireless Local Area Network) available on the farm 2. The WLAN is available on the farm 2.

[0050] After the driving robot 1 reaches the position shown in Fig. 3a, a command is issued by the user to navigate the driving robot 1 to the charging station 28 with the help of the reflector 29 and reach the position shown in Fig. 3b within the charging station 28. In response to a further command, the driving robot 1 backs out of the charging station after the charging process is completed if necessary. During this time, the driving robot 1 maintains its orientation at a specified distance. The length of the specified distance can be selected as needed and is shown by a dashed line in the drawing. As a result of the above backward movement, the driving robot 1 reaches the position shown in Fig. 3c. For example, to control the above position from the charging station 28, odometry, orientation with the help of a gyroscope, and / or position identification with the help of the reflector 29 can be used.

[0051] In the illustrated exemplary embodiment, the position taken in this way represents the first mark point. The coordinates of the first marking point are stored in the control unit of the driving robot 1. The coordinates are those on the environmental map created by the driving robot 1. This first marking point can be regarded as a kind of fixed point for the route network 3 to be set. This is because it is fixed by the design according to the positioning of the charging station 28. Therefore, the first marking point is also referred to as the "anchor point 30" hereinafter. In another design of this method, it is also possible to set the first marking point, that is, the anchor point 30, at the position of the driving robot 1 within the charging station 28.

[0052] The anchor point 30 also represents the starting point of the first route of the route network 3 to be learned. In order to define the route, a further marking point on the farm 2, for example, the marking point 30a, is manually approached by the user using remote control. The marking point 30a is located in front of the first feed bunker among the feed bunkers 27 and represents a position where the driving robot 1 can pick up feed from this first feed bunker 27.

[0053] The position and orientation of the driving robot 1 in front of the first feed bunker 27 are marked as the marking point 30a on the map guided by the driving robot 1. Information regarding a function, in this case, the function of picking up feed from the first feed bunker 27, is assigned to this marking point. The marking points, including the anchor point 30, are hereinafter also abbreviated as POI (Point Of Interest).

[0054] Thereafter, the other feed bunkers 27 are similarly approached, and the corresponding POIs 30b to 30d are defined within the map of the driving robot 1. The driving robot 1 can perform functions in these aspects, and the functions assigned to these aspects are saved. One possible sequence of functions is as follows, for example. - Stop - Start the mixing device - Request to fill x kilograms of feed from the feed bunker y times - Wait until the required amount of feed is filled - Continue driving

[0055] Here, x and y are parameter values to be used, and are usually specified by an advanced farm management system. In addition to the list of functions to be executed, more complex sequences, for example, sequences for determining how to handle errors such as when little or no feed is supplied from the feed bunker, can be defined.

[0056] As a result of the above, the situation shown in Figure 3d occurs. In this situation, the driving robot 1 is located in front of the left feed bunker 27.

[0057] While driving along the route from the anchor point 30, which is the starting point, to the POI 30d, which is the end point, not only the POIs 30a to 30d but also a plurality of waypoints 31 are recorded on the map. In Figure 3d, only two of the plurality of waypoints 31 are marked as an example between POIs 30a and 30b. The waypoint 31 is a position point along the route to be traveled, and is recorded and saved at intervals between several centimeters (cm) and about 20 cm. The waypoint 31 defines the course of the route element 32, and the route element 32 represents the course of the route between two adjacent POIs. By recording the waypoint 31, it becomes possible to precisely follow the route element 32, and thus it becomes possible to precisely follow the trained route. In this sense, the POI also represents the waypoint 31, but these are labeled in that the functions executed as described above can be linked to them, and furthermore, the correction option (see below) is labeled in that it can be more severely restricted than in the case of other unlabeled waypoints 31.

[0058] During manual driving and route recording, the remote control is used only to control the driving robot 1, and to mark mark points when the driving robot 1 is stationary, and furthermore to define actions as necessary. The latter can also be done retrospectively. Therefore, the user can focus only on the distance to the driving robot 1 and obstacles, etc.

[0059] According to the present invention, the recorded route specified by the waypoint 31 or the POIs 30 and 30a to 30d is authenticated by an already autonomously controlled route run before it can be run autonomously. However, the above autonomous control is performed under the supervision of the user. Only when this authentication process is performed for a specific route or a part of a route, this route or part of the route is marked as autonomously drivable and can be used as part of the automatic navigation system process.

[0060] The route previously specified between the anchor point 30 and the POI 30d is intended to be traveled in both directions. In the case of such a route, it may be necessary for authentication to be performed in both directions of travel. Alternatively, authentication in one direction may be sufficient, in which case the above direction does not necessarily have to be the direction in which this route was specified.

[0061] The route or part of the route to be authenticated can be reviewed in advance. Manual control when learning a route can result in unintended deviations, etc., which lead to sub-optimal routing. Sub-optimal routes result in longer distances being covered, which leads to avoidable cornering, acceleration, and braking maneuvers in practice. These consume energy and distort the materials of the driving robot 1 unnecessarily.

[0062] In an advantageous design of the method, the recorded routes of the previously created route network 3 can be displayed, in particular on the display of the remote control of the driving robot 1, before authentication in order to correct the route. In this case, the user has the option of manually or automatically correcting the route elements 32 or waypoints 31 between the POIs 30 and 30a to 30d and a part of the route element 32. It is also possible to manually correct the coordinates of the POIs 30a to 30d.

[0063] Manual correction can include, for example, moving the waypoint 31 or the POIs 30a to 30d. This shift may only be possible for a certain distance in order to prevent the risk of moving the waypoint into an obstacle area. In the case of the POIs 30a to 30d, the maximum distance by which the coordinates can be moved can be more strictly limited so as to prevent the risk that the assigned function (e.g., picking up food) is impaired after the movement. Furthermore, the distance between the waypoint 31 on the map and the boundary line can be determined, and a certain safety distance between the waypoint 31 and boundary objects, such as the wall 22 or the support 23, may have to be maintained during movement.

[0064] In addition to the manual correction of the route element 32 or a part of the route element 32, the option of automatic correction may also be available.

[0065] One possibility of automatic correction is provided by a smoothing function. Here, the root element 32 is smoothed by a filter algorithm, for example a low-pass filter, in order to reduce the possibility of "snaking" or deviation.

[0066] Accordingly, the root element 32 is replaced by a smoothed curve. In this case, certain boundary conditions, for example maintaining a tangent gradient from the POIs 30 and 30a to 30d, are taken into account to continuously and non-tortuously connect to subsequent or previous root elements 32. It is also possible to define an upper limit for the maximum movement of the waypoints 31 resulting from automatic smoothing.

[0067] Another option for automatic correction also results in the smoothing of the curve, which is to completely recalculate the positions of the waypoints 31 of the root element 32 using an appropriate curve modeled with parameters. For example, the root element 32 or a part of the root element 32 can be replaced by a so-called "spline curve" or "Bezier curve". Spline curves and Bezier curves are mathematically determined curve elements composed of polynomial functions. Similar to the case of smoothing using a filter function, certain boundary conditions can also be provided for smoothing using fully calculated waypoints. Certain boundary conditions are, for example, to continuously and non-tortuously connect to subsequent or previous root elements 32. Furthermore, in order to prevent collisions and correctly avoid objects, the shortest distance to boundary objects, such as the wall 22 or the support 23, can be maintained. Furthermore, in order to enable manual influence on route guidance, certain waypoints 31 can be considered immutable.

[0068] As a result, the root element 32 is converted into a smoothed root element 32 by the above manual correction or by one of the above automatic corrections, as shown in FIG. 3e.

[0069] In the actual authentication process, the driving robot 1 then reverses along the recorded and potentially smoothed route or reverses again in the recording direction. In this case, the user must confirm that the root element 32 has been traveled. For this purpose, for example, autonomous driving can be performed as long as a button on the remote control is pressed. Each part that has been traveled again is marked as "authentication successful". If there is a problem while driving, the user can release the button. In that case, the driving robot 1 immediately stops and switches from the process of authenticating the route to the recording process by manual control. In this way, the route or at least a part of it can be identified again in the corrected form.

[0070] If the route is fully authenticated by driving on and checking the route, the route becomes available for autonomous driving. FIG. 3f shows the situation after the authentication of the smoothed root element 33 after the driving robot 1 starting from the position of the POI 30d automatically returns to the anchor point 30 along the route and authenticates the route network 3 taught up to that point.

[0071] FIG. 4 shows how the route network 3 is extended by further routes in a further learning process following the above, throughout the farm 2 again. The route connects to both the first livestock shed 20 and the second livestock shed 21 and defines further points of interest 30e to 30q associated with various actions. For example, POI 30e represents a branch point where the route starting from anchor point 30 branches off to the first livestock shed 20 or the second livestock shed 21.

[0072] The navigation system can use the route element 32 between the anchor point 30 and the mark point 30e for both navigation tasks, i.e., the movement to the first livestock shed 20 or the movement to the second livestock shed 21. Further branching is possible at POI 30f. A total of four route elements 32 converge at POI 30f.

[0073] The first possibility starting from mark point 30f extends along the upper livestock area 25 of the first livestock shed 20 in FIG. 4 via POI 30g. Livestock located in the corresponding livestock area 25 are fed on the route element 32 extending between mark point 30g and mark point 30h. For this purpose, a corresponding command to release feed along the feed fence 26 is issued at mark point 30g. Furthermore, the driving robot 1 reduces the driving speed as required in this area. The feed release ends at POI 30h.

[0074] From mark point 30f, another movable route element 32 extends to mark point 30j. At mark point 30j, feeding is performed in the lower livestock area 25 of FIG. 4. The route along the corresponding feed fence 26 ends at mark point 30k.

[0075] The route element 32 extends from POI 30h and 30k, which are the end points of the feed release, respectively, to mark point 30i. The return routes after feeding converge at mark point 30i. In the reverse direction from mark point 30i, a further route element 32 extends to mark point 32f. In this case, the intermediate route element 32 has a feature such as a higher speed. Such a high speed becomes possible due to the feed container 111 of the driving robot 1 becoming empty.

[0076] Similar to up to the first livestock shed 20, the route element 32 extends from the mark point 30e to the second livestock shed 21. Here, the branch point is defined as the mark point 30l, and the route element 32 extends from the mark point 30l along the upper feed fence 26 in FIG. 4 via the mark points 32m and 30n, and further extends along the lower feed fence 26 in FIG. 4 via the mark points 30p and 30q. Here too, the convergence point for the return route is defined by the mark point 30o. From the mark point 30o, due to the feed container 111 of the driving robot 1 becoming empty, a faster return route becomes possible.

[0077] The recorded route element 32 shown in FIG. 4 can be smoothed manually and / or automatically in a later process (by using a filter function or being replaced by a mathematically described route element), and as a result, for example, the route network 3 shown in FIG. 5 is obtained. Here, all or part of the route element 32 is replaced by the smoothed route element 33.

[0078] Thereafter, a run for authentication for the recorded route element 32 and, if necessary, the smoothed route element 33 is performed. During that time, the driving robot 1 runs along the route elements 32 and 33, and then, if confirmed by the user, identifies them as being autonomously usable. Straightening the route element 32 along the feed fence 26 is particularly useful for ensuring that food is discharged and extruded at a certain defined distance from the feed fence 26.

[0079] As described above, during route learning, i.e., during the identification of route elements 32 by manual movement, no additional information, such as the measured distance to an obstacle, is displayed on the user's remote control. However, during both learning and authentication, it is possible to indicate by means of acoustic and / or visual warning signals of the driving robot 1 itself that the safety distance to a detected obstacle has not been violated. This prevents the possibility of a collision without diverting the user's attention from the position and movement of the driving robot 1.

[0080] The confirmation of a route element 32 or a part of a route or the entire route including several route elements 32 can be actively carried out by the user's actions. Alternatively, the confirmation can be made if the user does not intervene by changing the route along which the driving robot 1 travels or by stopping the driving robot 1 during authentication.

Explanation of Signs

[0081] 1 Driving robot 10 Travel direction 100 Chassis 101 Driving wheels 102 Steering wheels 103 Apron (feed pusher) 104 Bumper 110 Body 111 Feed container 112 Mixing device 113 Feed conveyor 114 Clad panel 115 Charging contact 116 Operating and / or display element 117 LiDAR sensor 118 Ultrasonic sensor 2 Farm 20 First livestock shed 21 Second livestock shed 22 Wall 23 Support column 24 gates 25 livestock area 26 feed fence 27 feed bunker 28 charging station 29 reflector 3 route network 30, 30a~30q mark point (POI) 31 waypoint 32 route element 33 smoothed route element

Claims

1. A method for learning a driving route to autonomously operate an agricultural driving robot (1), The process of providing an environmental map, The process of manually controlling the driving robot (1) to drive along a route extending within the area of ​​the environmental map from the starting point to the ending point, A step of determining the position of the driving robot (1) while it is traveling along the route using at least one sensor provided on the driving robot (1), A step of recording the coordinates of waypoints (31) along the route based on the completed location identification, A step of authenticating the route by autonomously driving the route taking into account the saved coordinates of the waypoint (31), wherein the route is manually verified. The process includes marking the confirmed route as autonomous. method.

2. Prior to the authentication step, the driving robot (1) is manually moved to the starting point, and during the authentication step, the route is autonomously driven in the same direction as the direction identified in the manual driving step. The method according to claim 1.

3. In the authentication process, the route is autonomously driven in the opposite direction to the direction identified in the manual driving process. The method according to claim 1.

4. The aforementioned route is checked in sections. The method according to claim 1.

5. The driving robot (1) is manually controlled via remote control when traveling along the route. The method according to claim 1.

6. The remote control is wirelessly connected to the control device of the driving robot (1). The method according to claim 5.

7. The aforementioned route or a portion of the route is considered to have been verified when an action was actively taken by the user during the authentication process. The method according to claim 1.

8. The aforementioned action is the operation of the control element on the remote control. The method according to claim 7.

9. The aforementioned route or a portion of the route is considered authenticated if there is no user intervention during the authentication process to correct or stop the movement of the driving robot (1). The method according to claim 1.

10. The distance to surrounding objects is measured via at least one distance sensor during the manual control and / or automatic driving of the route. The method according to claim 1.

11. The distance to the surrounding object is compared to the safety distance, and if the distance is shorter than the safety distance, an audible and / or optical warning is issued. The method according to claim 10.

12. While the driving robot (1) is stationary, at least one mark point (30 and 30a to 30q) can be designated during manual driving along the route. The method according to claim 1.

13. One or more functions to be performed can be assigned to the driving robot (1) at the mark points (30 and 30a to 30q). The method according to claim 12.

14. The mark points (30 and 30a to 30q) may represent the starting point or the ending point of at least one route. The method according to claim 12.

15. Prior to the authentication process, the route is modified entirely or partially by the user changing the coordinates of the waypoint (31). The method according to claim 1.

16. In order to smooth the course of the aforementioned route, either entirely or partially, prior to the authentication step, the route is automatically modified entirely or partially by changing the coordinates of the waypoints (31). The method according to claim 1.

17. The root element (32), which includes multiple adjacent waypoints (31), is transformed into a smoothed root element (33) by a mathematical function. The method according to claim 16.

18. The boundary conditions for modifying the coordinates of the waypoint (31) are predetermined. The method according to claim 15.

19. A system comprising an agricultural driving robot (1), the agricultural driving robot (1) having a control device for autonomous driving and a remote control connected to the control device and for manually controlling the driving robot (1), wherein the control device is configured to perform the method according to claim 1. system.

Images