Method and system for learning a route of an autonomously driving agricultural driving robot

Abstract

The invention relates to a method for learning a route of an autonomously driving agricultural driving robot (1). The method comprises the following steps: - providing an environmental map; - driving along a route in the region of the environmental map from a starting point to an end point by manually controlling the driving robot (1); - locating the driving robot (1) while driving along the route with the aid of at least one sensor arranged on the driving robot (1); - recording coordinates of waypoints (31) along the route on the basis of the locating process that has been carried out; - validating the route by autonomously driving along the route taking the stored coordinates of the waypoints (31) into account, wherein the route is manually confirmed; - marking the confirmed route as being able to be driven along autonomously. The invention also relates to a system comprising an agricultural driving robot (1) with a control device for autonomous driving, and a remote control that is coupled to the control device and is intended to manually control the driving robot (1). The system is distinguished by the fact that the control device is configured to carry out such a method.

Description

[0001] Method and system for teaching a route of an autonomously driving agricultural robot

[0002] The invention relates to a method for teaching a route for an autonomously driving agricultural robot, particularly in a stable or farmyard area. The invention further relates to a system with an autonomously driving agricultural robot with which a route can be taught.

[0003] Many tasks in a farm's stable or yard involve the transport of materials. For example, animal feeding systems are often used. In a central area, the so-called "kitchen," feed rations are mixed from various stored basic ingredients as needed and promptly, and then distributed along so-called "feed lanes" for feeding the animals. Another example concerns the removal of animal excrement. Due to the size of the areas, cleaning farmyards or stables is also usually carried out using vehicles.

[0004] In order to be able to carry out these tasks as autonomously as possible and with minimal personnel input, automated systems and devices have become established for these different applications in the agricultural sector.

[0005] For example, WO 2008 / 097080 A1 describes an autonomously operating feeding system for animals, such as cows. A central component of this system is an autonomously driving vehicle with a feed container that can be automatically filled in a central "kitchen area." The feed can be mixed while driving from the feed containers to the feed unloading point. At the unloading point, feed is automatically dispensed by tipping the container. Various options for the vehicle's movement are described, for example, a predetermined route via pre-laid rails. Another alternative is autonomous navigation using sensors or path markers. Navigation based on a radio positioning system, such as GPS (Global Positioning System), is also described.Especially if a route is not specified by laid rails or other routes, a learning process is necessary to enable navigation in the stable or yard area at all.

[0006] Such a training method for an autonomously driving agricultural robot is known from WO 2018 / 074 917 A2. In a first step, the autonomously driving robot is manually controlled along a desired route using an external mobile device, such as a tablet computer. During manual control, sensor data collected by the driving robot is sent to the mobile device and displayed on a map along with the position of the driving robot. Transmitted sensor data includes, for example, measured current distances to objects. The route traveled is saved and can then be automatically followed by the driving robot in an autonomous navigation process.

[0007] Displaying additional information on the mobile device used as a remote control may be helpful, but it also distracts from careful observation of the driving robot during the training run.

[0008] It is an object of the present invention to provide a training method for a driving robot used in agriculture, in which a route can be trained intuitively and with full concentration on the driving robot and the route can nevertheless be checked and corrected, for example, with regard to maintained safety distances or the like.

[0009] This problem is solved by a method or system having the features of the respective independent claim. Advantageous embodiments and further developments are the subject of the dependent claims.

[0010] A method according to the invention of the type mentioned at the outset comprises the following steps: A map of the environment is provided, and a route located within the area of ​​the map of the environment is manually driven by the driving robot from a starting point to an end point under manual control by a user. While driving the route, the driving robot is localized using at least one sensor arranged on the driving robot, and coordinates of waypoints along the route are recorded based on the localization. In a next step, the driving robot autonomously drives the route, taking into account the stored coordinates of the waypoints, with the route being manually confirmed by the user. After this step of validating the route, the route is marked as confirmed and is thus available for subsequent autonomous driving.

[0011] With this method, the route is taught through controlled manual driving, independently of any validation of the route that occurs during subsequent driving. During manually controlled driving of the desired route, the user can thus fully concentrate on the route itself and the driving robot during the driving. If the route is then automatically driven for the first time for validation purposes, the user can concentrate on the driving robot's surroundings directly or on displayed measured values, checking required distances, etc. Only once the route has been confirmed by the user is it marked as autonomously drivable and is then available for independent, autonomous navigation.By dividing the learning process into a single step of manually specifying the route and subsequent validation, the user's attention is not overwhelmed and the learning process itself can be carried out without the risk of a collision or other error.

[0012] In an advantageous embodiment of the method, the driving robot is manually guided back to the starting point before the validation step, and during the validation step the route is autonomously traveled in the same direction in which it was specified in the manual travel step. Routes are generally taught in the direction in which they will later be traveled. Because the route is validated in the same direction as it was taught during manual travel, validation is also carried out in the direction of later use and is therefore practical. Alternatively, it is also conceivable to perform the validation of the route in a direction of travel that is opposite to the direction in which the route was specified in the manual travel step. If necessary, it can also be provided to validate both directions of travel, particularly if a route is also used in both directions during autonomous productive operation.In a further advantageous embodiment of the method, the route is confirmed section by section during validation. This can require active or passive confirmation.

[0013] With active confirmation, the route or a section of the route is considered confirmed if the user actively performs an action during the validation step. An active action can, for example, be the actuation of a control element on the driving robot itself or on a remote control, which was preferably previously used to control the driving robot to manually drive the route. Such a remote control is, for example, wirelessly linked to a control device of the driving robot.

[0014] In the case of passive confirmation, the route or a section of the route is considered confirmed if no user intervention occurs during the validation step to correct or stop the movement of the driving robot.

[0015] In a further advantageous embodiment of the method, a distance to a surrounding object is measured via at least one distance sensor during the manually controlled and / or automatic travel along the route. The distance sensor can be the sensor used for navigation or localization and / or an independent sensor. Preferably, the measured distance to the surrounding object is compared with a predetermined or predeterminable safety distance, with an acoustic and / or visual warning being issued if the safety distance is not reached. Further preferably, during the route learning process, i.e. while the route is being specified by manually traveling along it, no additional information, such as measured distances to obstacles, is displayed on the user's remote control, which could distract the user.Nevertheless, both during training and validation, provision can be made to signal any breach of safety distances to detected obstacles by means of an acoustic and / or visual warning signal, preferably on the driving robot itself. This prevents potential collisions without distracting the user's attention from the position and movement of the driving robot. Furthermore, it can be provided that validation of a path element is prevented if breach of the safety distance is detected. In a further advantageous embodiment of the method, at least one designated point can be defined by the user while the driving robot is stationary while manually driving the route. One or more functions to be executed can be assigned to such a designated point, which the driving robot then performs at or from the designated point during subsequent operation.A distinguished point can also represent the start or end point of at least one route.

[0016] In a further advantageous embodiment of the method, it is possible for the route to be modified manually by the user or automatically before the validation step, either in whole or in section, by changing the coordinates of the waypoints. This manual intervention allows the route to be subsequently influenced without having to repeat the entire journey manually. It can be provided that marked points cannot be moved. However, it is also conceivable that the option of manual correction is also available for marked points, possibly with increased security conditions, e.g., only after (additional) user authorization and / or authentication, or by allowing only minor changes to the recorded coordinates.

[0017] The primary goal of automated modification is to smooth the route or a section of it. Recorded route elements that include multiple adjacent waypoints are transformed into smoothed route elements using mathematical functions.

[0018] For this purpose, filter algorithms, in particular low-pass filters, can be used, for example, to shift the recorded waypoints in order to reduce "swerving" during travel. Curve smoothing can also be achieved by completely recalculating the position of waypoints and route elements using suitable parametrically modeled curves. In this case, the recorded position of a waypoint is not taken into account in its recalculation, and the route element is determined only by the position of its endpoints (usually marked points). Mixed forms are also conceivable in which the recorded position of a waypoint is taken into account in its recalculation with an adjustable weighting. For both manual and automatic correction of the position of waypoints, restrictive boundary conditions for changing the coordinates of the waypoints can be specified.Such boundary conditions can refer to maximum displacements, to transitions to the next path elements, e.g. to prevent a bend in the path to the following or preceding path elements, and / or to minimum distances to be maintained from objects.

[0019] A system according to the invention comprising an autonomously driving agricultural robot and a remote control for manually controlling the robot is configured to carry out the aforementioned method. The advantages mentioned in connection with the method are achieved.

[0020] The invention is explained in more detail below using an exemplary embodiment and the accompanying figures. The figures show:

[0021] Fig. 1 a, b each show an oblique view of an example of a driving robot from different viewing directions;

[0022] Fig. 2 a plan of a stable and yard area with a driving robot;

[0023] Fig. 3a-f a section of the plan of Fig. 2 in different phases of a learning method according to the invention;

[0024] Fig. 4 shows the plan of Fig. 2 with a plurality of manually specified routes; and

[0025] Fig. 5 the plan according to Fig. 4 with smoothed sections.

[0026] Figures 1a and 1b show an example of a driving robot 1 for agricultural tasks in an overall view from different perspectives. The route learning procedure described below can be carried out, for example, with this driving robot 1.

[0027] In this example, the driving robot 1 is a so-called "feeding robot," which is designed to pick up feed from a dispensing point, mix it automatically, and unload it at one or more feeding stations. The driving robot 1 is therefore also referred to below as the "feeding robot" or simply "robot."

[0028] The same reference symbols identify the same or similarly functioning elements in all figures. For reasons of clarity, not every element in every figure is provided with a reference symbol. In the description, the terms "right" and "left" refer to the respective representation of the figure. The terms "top" and "bottom", on the other hand, refer to the natural orientation of the driving robot. The terms "front" and "back" refer to a forward direction of travel 10 of the driving robot 1. The forward direction of travel 10, which is indicated by a directional arrow in Fig. 1 a, b, represents the main direction of travel of the driving robot 1.

[0029] The driving robot 1 has two main components, a chassis 100 and a body 110.

[0030] The chassis 100 is preferably universally applicable and can, if necessary, be used together with various functional units. Only cladding and / or protective elements of the chassis 100 are visible in Figs. 1a, b, specifically a circumferential apron 103 and two bumpers 104, as well as two of a total of four wheels, specifically one of two drive wheels 101 (in Figs. 1a, b) and one of two swivel wheels 102 (in Fig. 1b). Another of the swivel wheels is located at the front in the forward direction of travel 10 and is hidden under the apron 103 in Figs. 1a, b. In the present embodiment of a feeding robot, the apron 103 also functions as a feed pusher, with which already unloaded feed can be pushed together.

[0031] The structure 110 essentially determines the functionality of the driving robot and thus its intended use within the stable or farm area.

[0032] In the mobile robot 1, which in this case is equipped as a feeding robot, the structure 110 has a feed container 111 as a key component. The feed to be distributed is received in the feed container 111 and can be mixed, if necessary, during filling, in a loading station 28 (see Fig. 2) and / or during travel using a mixing device 112 (not visible in Figs. 1a and 1b). A feed conveyor 113, which is moved using a conveyor belt, is provided for distributing the feed. Depending on the direction of travel of the conveyor belt, feed can be distributed to either side of the feeding robot. The arrangement of the feed container 111 and the feed conveyor 113 represents the functional unit of the mobile robot 1, as they provide its specific functionality and thus define it as a feeding robot.The structure 110 further comprises a casing made up of a plurality of casing elements, typically casing panels 114. The casing panels 114 can preferably be removed separately to gain access to underlying components and their maintenance or replacement. Externally accessible elements are integrated into the casing, for example, charging contacts 115 (see Fig. 1 a) and operating and / or display elements 116 (see Fig. 1 b). The driving robot 1 is configured to automatically drive into the charging station 28 (see Fig. 2), where the charging contacts 115 are contacted to recharge batteries or other power storage devices of the driving robot 1.

[0033] Driving robot 1 is also equipped with a navigation system that enables navigation in the stable or yard area without permanently installed infrastructure elements such as rails or guide cables. For this purpose, the driving robot is equipped with a number of sensors that are either integrated into the cladding or protrude from the cladding.

[0034] 1 a and 1 b show two LIDAR (light detection and ranging) sensors 117, which are used for object detection to assist navigation. The two LIDAR sensors 117 are arranged at the front and rear of the driving robot, respectively. Alternatively or additionally, provision can be made to arrange an optical camera at the front and, if necessary, also at the rear in the direction of travel. The cameras then serve for object or step detection or as additional navigation support. The cameras can be tilted downwards in order to be able to record and thus monitor the ground area directly in front of the driving robot 1 in both directions of travel (i.e. when driving forwards and backwards). Furthermore, ultrasonic sensors 118 are distributed around the circumference of the driving robot 1 in the lower region of the cladding as distance sensors to nearby obstacles.

[0035] Further sensors, not visible here, are mechanical sensors that detect a force acting on one or both bumpers 104. For this purpose, the respective bumper 104 can, for example, be movably mounted so that one of possibly several sensors is actuated when moved against a spring force. In an alternative embodiment, the bumper 104 can be formed in an outer region from an elastically deformable material, in particular a foam, into which a sensor is incorporated, which detects a deformation preferably along the entire edge of the bumper 104. A collision with an obstacle is advantageously dampened and detected at the same time. In one embodiment, for example, two spaced-apart electrodes can be embedded in the elastic material along the edge of the bumper 104, between which a capacitance is detected. A change in the capacitance indicates a deformation of the material.In a further embodiment, a pull chain can be incorporated into the elastic material, which is coupled to a switch or sensor. Deformation of the elastic material leads to a pull on the pull chain, which is detected by the switch or sensor.

[0036] The driving robot 1 has at least one control device that controls the driving robot's actuators, including drive motors, and reads and evaluates signals from the sensors. The control device also performs navigation tasks and maintains a stored map of the environment, which is used, among other things, to localize the driving robot in its surroundings. The map is preferably created by the driving robot 1 itself by evaluating the sensor data recorded during various journeys using a so-called SLAM (Simultaneous Localization and Mapping) process. In addition, the control device is equipped with or coupled to communication interfaces, in particular for wireless communication. The communication interfaces serve, for example, to connect to a higher-level operations management system that coordinates the use of the driving robot 1.

[0037] Furthermore, the communication interfaces can be used to control the driving robot 1 using a remote control.

[0038] Fig. 2 shows a two-dimensional plan of a farm 2 on which a mobile robot 1 is to be used as a feeding robot. In principle, several mobile robots can be used on a farm to deliver feed from the feed bunkers to the animals. The mobile robot 1 can be designed for this purpose, for example, as shown in Figs. 1a and 1b.

[0039] In the example shown here, farm 2 comprises two stables, specifically a first stable 20 and a second stable 21, for animal husbandry, for example for keeping cows, and a surrounding farmyard area. By way of example, the two stables 20, 21 differ in their size, with the second stable 21 being an annex to the larger first stable 20, which is to be regarded as the main stable. The number of two stables 20, 21 of farm 2, as well as their size and layout, are purely exemplary. Both stables 20, 21 have walls 22 on the outside and a plurality of supporting pillars 23 inside. This is also purely exemplary. The stables 20, 21 could equally well be provided with walls on the inside. Furthermore, a gate 24 is provided for each of the stables 20, 21.

[0040] Within each of the two stables 20, 21, several animal areas 25 are provided, i.e., areas in which animals, e.g., the aforementioned cows, are kept. Each animal area 25 is assigned a so-called "feeding grid" 26, in front of which feed is distributed, which the animals can consume from the animal area 25.

[0041] Outside the stables 20, 21, i.e., in the yard area of ​​farm 2, several feed bunkers 27 are located, in which different types of feed are stored for the animals. Three larger feed bunkers 27 are shown as examples, which serve, for example, to hold silage feed. A smaller feed bunker 27, shown as round in the schematic diagram of Fig. 1, serves to hold concentrate feed. The feed bunkers 27 each have feed conveyors with which the collected feed or concentrate feed can be dispensed.

[0042] The plan of the farm 2 shown in Fig. 2 is a schematic drawing, but essentially reflects what the driving robot 1 detects of its surroundings using the lidar sensors 117. The two lidar sensors 117 arranged on the driving robot 1 scan the surroundings in a two-dimensional plane that is aligned parallel to the chassis 100 and thus essentially parallel to the ground on which the driving robot 1 is moving. Accordingly, the driving robot 1 uses the lidar sensors 117 to only detect features that are located in the scan plane of the lidar sensors 117. In the example shown, this plane is located at a height of approximately 1.5 - 2.5 m (meters) above the ground. For this reason, the feed bunkers 27 only detect contours that are at this height, which is why, for example, hoses or conveyor belts that serve to dispense the collected feed or concentrate are not visible in this plan.Also not visible are grids or other barriers formed around the animal areas 25. The invisible elements also include the feeding grids 26, which for this reason are shown in dashed lines in Fig. 2. In order to be able to use the driving robot 1 on the farm 2, according to the invention, routes, i.e., possible paths of the driving robot 1, are defined in a training process. During operation of the driving robot 1, a navigation process can then compile a route from the trained routes, depending on the task to be completed, along which the driving robot moves in order to be able to fulfill its tasks.

[0043] A method according to the invention is explained below using the example of the operation 2 shown in Fig. 2. The driving robot 1 is trained, for example, to be able to use routes that run between a charging station 28, the feed bunkers 27, and the various feed deposition points in front of the feeding grids 26.

[0044] In the illustrated operation 2, the charging station 28 is mounted adjacent to the feed bunkers 27. The charging station 28 is controlled by the driving robot 1 to charge batteries for its power supply. The charging station 28 comprises contacts that, when the driving robot 1 is correctly positioned, make contact with the charging contacts 115 (see Fig. 1 a) and via which charging current is provided to charge the batteries of the driving robot 1. As previously explained, the driving robot 1 uses the signals from the lidar sensors 117 to determine its position during navigation. Signals from other sensors, e.g., from wheel rotation sensors for odometry, can be used to increase positioning accuracy.

[0045] In the vicinity of the charging station 28, an alternative positioning method can be used in which markings on the charging station are optically detected. Such markings include, for example, the reflectors 29 shown in Fig. 2, which are detected by the driving robot 1 using the lidar sensors 117 or other optically operating sensors. In the vicinity of the charging station 28, this allows a higher positioning accuracy required for contacting to be achieved than with contour-based navigation.

[0046] In Fig. 3a, a section of the farm 2 in the area of ​​the feed bunkers 27 and the loading station 28 is shown enlarged.

[0047] In a first step, which is carried out to prepare for the training process according to the application, the driving robot 1 is manually guided into the vicinity of the charging station 28 so that it is at a distance of approximately 1-2 m from the charging station 28 and is positioned in its direction of travel 10 aligned with the charging station 28. The distance and position are selected such that navigation using the reflectors 29 into the charging station 28 is possible.

[0048] To control the driving robot 1, a remote control (not shown here) is used, which preferably, but not necessarily, communicates wirelessly with the driving robot 1. An optical or radio-based communication link can be used for transmission. In particular, a user's universally usable mobile device, e.g., a tablet computer, can be used as the remote control. The communication connection can be established directly between the tablet computer and a receiving device of the driving robot 1 or via a shared communication network, e.g., a WLAN (Wireless Local Area Network) available at the facility 2.

[0049] After the driving robot 1 has been brought into the position shown in Fig. 3a, the user triggers a command that navigates the driving robot 1 into the charging station 28 using the reflectors 29 until the positioning of the driving robot 1 in the charging station 28 shown in Fig. 3b is reached. Upon a further command, the driving robot 1 moves backward out of the charging station - possibly after a completed charging process - while maintaining its orientation by a defined distance, the length of which can be selected if necessary, which is shown in dashed lines in the figures, and thus assumes the position shown in Fig. 3c. To navigate to this position starting from the charging station 28, for example, odometry, orientation using a gyroscope and / or localization using the reflectors 29 can be used.

[0050] In the exemplary embodiment shown, the position thus assumed represents a first designated point, the coordinates of which are stored in the control device of the driving robot 1. The coordinates refer to the environmental map created by the driving robot 1. This first designated point can be regarded as a type of fixed point for a path network 3 to be constructed, since it is structurally fixed by the positioning of the charging station 28. It is therefore also referred to below as "anchor point 30". In an alternative embodiment of the method, it can also be provided that the first designated point, the anchor point 30, is set to the position of the driving robot 1 in the charging station 28. At the same time, the anchor point 30 represents a starting point for the first section of the path network 3 to be learned.To define the route, the user subsequently manually approaches further designated points on the farm 2 using the remote control, e.g., a designated point 30a located in front of a first of the feed bunkers 27 and representing the position in which the driving robot 1 can pick up feed from this first feed bunker 27.

[0051] The position and orientation of the driving robot 1 in front of the first feed bunker 27 is noted on the map guided by the driving robot 1 as a designated point 30a. This designated point is assigned information about its function—here, the intake of feed from the first feed bunker 27. Designated points, which also include the anchor point 30, are hereinafter abbreviated as POI (Point of Interest).

[0052] In the same way, the other feed bunkers 27 are then approached, and corresponding POIs 30b-d are defined in the map of the driving robot 1, and associated functions that the driving robot 1 can perform at these points are stored. A possible sequence of functions is, for example:

[0053] - Stop

[0054] - Start mixing device

[0055] - Request filling of x kg feed from the feed bunker with the number y

[0056] - Wait until the requested amount of feed has been filled

[0057] - Keep going.

[0058] Here, x and y represent the parameter values ​​to be used, which are usually specified by the higher-level farm management system. In addition to a list of functions to be performed, more complex sequences can also be defined, e.g., to specify how to handle errors, such as when no or too little feed is released from the feed bunker.

[0059] This then results in the situation shown in Fig. 3d, in which the driving robot 1 is positioned in front of the left feed bunker 27.

[0060] While traveling the route from the starting point, the anchor point 30, to an end point, the POI 30d, not only the POIs 30a-d are recorded on the map, but also a plurality of waypoints 31, of which only two, between the POIs 30a and 30b, are marked in Fig. 3d as an example. The waypoints 31 are position points along the traveled route, which are recorded and saved at intervals of a few centimeters (cm) up to approximately 20 cm. The waypoints 31 define the course of a path element 32, which represents the course of the path between two neighboring POIs. The recording of the waypoints 31 enables precise travel of the path elements 32 and thus the learned route.In this sense, the POIs also represent waypoints 31, which, however, are marked in that the previously mentioned functions to be carried out can be linked to them and that correction options (see below) can be more limited than for the other, non-marked waypoints 31.

[0061] The remote control is used only to control the driving robot 1 during manual driving and recording of the route. Then, when the driving robot 1 is stationary, it is used to mark a designated point and, if necessary, define actions. This can also be done subsequently. The user's concentration can thus be fully focused on the driving robot 1 and the distances to obstacles, etc.

[0062] Before a recorded route specified by waypoints 31 or POIs 30, 30a-d can be driven autonomously, the invention provides for validation of this route by an autonomously controlled drive, which, however, takes place under the supervision of the user. Only after this validation step has been performed for a specific route or route section is this route or route section marked as autonomously driveable and can be used as part of an automatic navigation process.

[0063] The previously specified route between anchor point 30 and POI 30d is intended for travel in both directions. For such a route, it may be stipulated that validation must also be performed in both directions. Alternatively, it may also be stipulated that validation in one direction is sufficient, although this direction does not necessarily have to be the direction in which this route was specified.

[0064] It may also be necessary to pre-revise the route or sections of the route to be validated. Manual control during route training may result in unplanned swerves or similar deviations, leading to a suboptimal route. A suboptimal route results in a longer distance traveled and leads to avoidable cornering, acceleration, and braking maneuvers, which consume energy and place unnecessary strain on the driving robot's equipment.

[0065] To correct the route, in an advantageous embodiment of the method, the recorded route of the previously created route network 3 can be displayed on a display, in particular a display of the remote control of the driving robot 1, prior to validation. The user has the option of manually or automatically correcting route elements 32 or waypoints 31 and sections of route elements 32 between POIs 30, 30a-d. Manual correction of coordinates of POIs 30a-d can also be provided.

[0066] Manual correction can, for example, involve moving a waypoint 31 or a POI 30a-d. It can be provided that this movement is only possible by a certain distance in order to prevent the risk of moving waypoints into the area of ​​obstacles. For POIs 30a-d, the maximum possible movement of the coordinates can be more strictly limited in order to avoid the risk of the associated function (e.g., picking up food) being impaired after a movement. Furthermore, it can be provided that distances between waypoints 31 and boundaries are determined on the map, and that certain safety distances between waypoints 31 and boundary objects, e.g., walls 22 or supporting pillars 23, must be maintained during movement.

[0067] In addition to a manual correction of path elements 32 or parts of the path elements 32, automated correction options may also be available.

[0068] One possibility for automated correction is provided by a smoothing function. In this case, path elements 32 are smoothed using filter algorithms, e.g., a low-pass filter, to reduce potential "snaking" or swerving. The path elements 32 are thereby replaced by smoothed curves, taking into account certain boundary conditions, such as a continuous, non-bending connection to following or preceding path elements 32, e.g., by maintaining tangent gradients in the POIs 30, 30a-d. An upper limit for the maximum displacement of waypoints 31 resulting from the automated smoothing can also be specified.

[0069] Another option for automated correction, which also results in curve smoothing, is a complete recalculation of the position of the waypoints 31 of path elements 32 using suitable parametrically modeled curves. For example, path elements 32 or parts of path elements 32 can be replaced by so-called "splines" or "Bezier curves." Splines and Bezier curves are mathematically determined curve elements composed of polynomial functions. As with smoothing using a filter function, certain boundary conditions can also be specified for smoothing using fully calculated waypoints, such as a continuous, non-kinking connection to following or preceding path elements 32. Furthermore, it can also be specified here that minimum distances to limiting objects, e.g., the walls 22 or the supporting pillars 23, are maintained in order to prevent collisions and to correctly avoid the objects.In addition, it can be provided that certain waypoints 31 are considered unchangeable in order to be able to manually influence the route.

[0070] As a result, the path elements 32 are converted into smoothed path elements 33 by manual or one of the described automatic corrections, as shown in Fig. 3e.

[0071] During the actual validation step, the driving robot 1 then travels back along the recorded and, if necessary, smoothed route or again in the recording direction, whereby the user must confirm traveled route elements 32. For this purpose, it can be provided, for example, that the automatic travel continues as long as a button on the remote control is held down. Each section traveled again is marked as "successfully validated". If a problem arises during travel, the user can release the button, whereupon the driving robot 1 immediately stops and switches from the route validation step back to the manual recording step. In this way, a route or at least a section of it can then be specified again with corrections.

[0072] If the route is fully validated by traveling and confirming, it is then available for autonomous travel. Fig. 3f shows the situation after validation of the smoothed path elements 33, after the driving robot 1 has automatically traveled the route from its position at the POI 30d back to the anchor point 30 to validate the path network 3 learned up to that point.

[0073] Fig. 4 shows, again for the entire area of ​​the farm 2, how the route network 3 is expanded with additional routes in subsequent training phases. Routes lead to both the first stable 20 and the second stable 21. Additional designated points 30e to 30q are defined, which are associated with various actions. For example, the POI 30e represents a junction where the route starting from the anchor point 30 branches off into either the first stable 20 or the second stable 21.

[0074] A navigation system can use the path element 32 between the anchor point 30 and the designated point 30e for both navigation tasks, i.e., a route to the first stable 20 or to the second stable 21. A further branching is possible at POI 30f, where a total of four path elements 32 converge.

[0075] A first possibility, starting from the designated point 30f, runs via POI 30g along the upper animal area 25 of the first stable 20 in Fig. 4. Feeding of the animals located in the corresponding animal area 25 takes place on the path element 32 running between the designated point 30g and a designated point 30h. For this purpose, a corresponding command for ejecting feed along the feeding grid 26 is issued at the designated point 30g. In addition, the driving robot 1 may reduce its driving speed in this area. The feed ejection ends at POI 30h.

[0076] From the designated point 30f, an alternatively traversable path element 32 leads to a designated point 30j, at which feeding takes place in the lower animal area 25 in Fig. 4. The path along the corresponding feeding fence 26 ends at the designated point 30k. From the end points of the feed ejection, the POIs 30h and 30k, path elements 32 lead to a designated point 30i, where the return paths converge after feeding. From the designated point 30i, another path element 32 leads back to the designated point 30f, with the intermediate path element 32 being characterized, for example, by a higher driving speed, which is possible when the feed container 111 of the driving robot 1 is empty.

[0077] In a similar manner to the first stable 20, path elements 32 lead from the designated point 30e into the second stable 21. Here, a branching point is defined as designated point 30I, from which path elements 32 lead via designated points 30m and 30n along the upper feeding fence 26 in Fig. 4 and via designated points 30p and 30q along the lower feeding fence 26 in Fig. 4. Again, a designated point 30o defines a merging point for return journeys, from which a faster return journey with the feed container 111 of the driving robot 1 emptied is possible.

[0078] The recorded path elements 32 shown in Fig. 4 can in turn be smoothed manually and / or automatically in a subsequent step (using filter functions or replaced by mathematically described path elements), which leads, for example, to the path network 3 shown in Fig. 5, in which all or some of the path elements 32 are replaced by smoothed path elements 33.

[0079] Subsequently, a validation run is performed for the recorded path elements 32 and, if applicable, the smoothed path elements 33. During this run, the driving robot 1 travels along the path elements 32, 33 and then marks them as autonomously usable if they are confirmed by the user. Straightening the path elements 32 is particularly useful along the feeding fences 26 to ensure that feed is ejected or pushed at a defined and consistent distance from the feeding fence 26.

[0080] As previously described, during the route learning process, i.e., during the specification of the path elements 32 by manually driving along them, no additional information, such as measured distances to obstacles, is displayed on the user's remote control, which would distract the user. However, both during the learning process and during the validation process, it can be provided to signal a breach of the safety distances to detected obstacles by means of an acoustic and / or visual warning signal on the driving robot 1 itself. This prevents possible collisions without distracting the user's attention from the position and movement of the driving robot 1.

[0081] It can be provided that confirmation of a path element 32 or a section of the path comprising multiple path elements 32, or of the entire path, occurs actively through a user action. Alternatively, it can be provided that confirmation is accepted if the user does not intervene during the validation run to modify the route traveled or stop the driving robot 1.

[0082] List of reference symbols

[0083] 1 driving robot

[0084] 10 Direction of travel

[0085] 100 chassis

[0086] 101 Drive wheel

[0087] 102 swivel wheel

[0088] 103 Apron (feed pusher)

[0089] 104 bumper

[0090] 110 Structure

[0091] 111 feed containers

[0092] 112 Mixing device

[0093] 113 feed conveyors

[0094] 114 cladding panels

[0095] 115 charging contacts

[0096] 116 operating and / or display elements

[0097] 117 Lidar sensor

[0098] 118 Ultrasonic sensor

[0099] 2 Operation

[0100] 20 first stable

[0101] 21 second stable

[0102] 22 Wall

[0103] 23 supporting pillars

[0104] 24 goals

[0105] 25 Animal area

[0106] 26 feeding grids

[0107] 27 feed bunkers

[0108] 28 charging stations

[0109] 29 reflectors

[0110] 3 trail network

[0111] 30, 30a-q excellent point (POI)

[0112] 31 Waypoint

[0113] 32 path element

[0114] 33 smoothed path element

Claims

Claims 1. Method for teaching a route to an autonomously driving agricultural robot (1) comprising the following steps: - Providing a map of the surroundings; - Driving a route located within the area of ​​the environment map from a starting point to an endpoint by manually controlling the driving robot (1 ); - Locating the driving robot (1 ) while traversing the route using at least one sensor arranged on the driving robot (1 ); - Recording coordinates of waypoints (31) along the route based on the localization achieved; - Validating the route by autonomously driving the route taking into account the stored coordinates of the waypoints (31), whereby the route is manually confirmed; - Marking the confirmed route as suitable for autonomous travel.

2. Method according to claim 1, wherein before the validation step the driving robot (1) is manually steered back to the starting point and in the validation step the route is driven autonomously in the same direction in which it was specified in the manual driving step.

3. Method according to claim 1 or 2, wherein in the validation step the route is driven autonomously in a direction opposite to the direction in which the route was specified in the manual driving step.

4. Method according to one of claims 1 to 3, wherein the route is confirmed section by section.

5. Method according to one of claims 1 to 4, wherein the driving robot (1 ) is manually controlled via a remote control when driving along the route.

6. Method according to claim 5, wherein the remote control is wirelessly coupled to a control unit of the driving robot (1 ). A method according to any one of claims 1 to 6, wherein the route or a section of the route is considered confirmed if the user actively performs an action during the validation step. A method according to claims 5 and 7, wherein the action is the actuation of a control element on the remote control. A method according to any one of claims 1 to 6, wherein the route or a section of the route is considered confirmed if the user does not intervene during the validation step to correct or stop the movement of the robot (1). A method according to any one of claims 1 to 9, wherein, during manually controlled and / or automatic traversal of the route, a distance to a surrounding object is measured via at least one distance sensor.A method according to claim 10, wherein the distance to the surrounding object is compared with a safety distance and an acoustic and / or visual warning is issued if the safety distance is breached. A method according to any one of claims 1 to 11, wherein at least one designated point (30, 30a-q) can be defined by the user while the robot (1) is stationary and manually traversing the route. A method according to claim 12, wherein one or more functions to be executed can be assigned to the robot (1) at the designated point (30, 30a-q). A method according to claim 12 or 13, wherein a designated point (30, 30a-q) can represent the start or end point of at least one route. A method according to any one of claims 1 to 14, wherein the route is manually defined in whole or in part by the user before the validation step. The user is modified by changing the coordinates of the waypoints (31). A method according to any one of claims 1 to 14, wherein the route is automatically modified completely or section by section before the validation step by changing the coordinates of the waypoints (31) to smooth the route completely or section by section. A method according to claim 16, wherein path elements (32) comprising several adjacent waypoints (31) are transformed into smoothed path elements (33) by mathematical functions. A method according to any one of claims 15 to 17, wherein restrictive boundary conditions are specified for modifying the coordinates of the waypoints (31).System comprising an agricultural driving robot (1) with a control unit for autonomous driving and a remote control coupled to the control unit for manual control of the driving robot (1), characterized in that the control unit is configured to carry out a method according to the preceding claims.

Images