Trainable timer for garden equipment with user rating

Description

[0001] The invention relates to the control of mobile gardening equipment for lawn care. Such gardening equipment can be robotic lawnmowers, garden tractors, lawn mowers, or other mobile or stationary tools used in gardening, green space maintenance, and landscaping.

[0002] In practice, lawns are often mowed manually by a user or at predetermined regular intervals, such as daily or weekly. It is still common practice, especially with robotic lawnmowers, for lawns to be mowed daily.

[0003] US 2017 / 0020064 A1 discloses a method for autonomously mowing a lawn. The robotic lawnmower includes a sensor for measuring grass height. The mowing schedule can be adjusted based on fluctuating weather information.

[0004] The invention concerns the intelligent deployment planning and control of gardening equipment. There is a need for the deployment of gardening equipment, e.g., a robotic lawnmower, to be automatically adapted to changing environmental conditions and user preferences. For example, the lawn should be mowed automatically whenever a certain grass height is reached. Furthermore, the lawn should not be mowed during rainy or other restricted periods. The deployment planning and configuration can relate to either a fully automated gardening equipment (e.g., a robotic lawnmower) or a semi-automated gardening equipment (e.g., a garden tractor with a mowing deck) that is operated with technical assistance by a human.

[0005] In the case of robotic lawnmowers, in addition to the start time, the mowing duration is also particularly important. Chaotically navigating robotic lawnmowers that mow the lawn in more or less random paths require sufficient time to mow the lawn completely and evenly, whereby the speed can also depend on the grass height. Therefore, to achieve optimal mowing duration, an optimization between efficiency and the quality of the mowing result is desirable.

[0006] It is known from the state of the art to plan the working hours of garden equipment by processing rain sensor data or weather data.

[0007] It is also known to estimate grass growth using a grass growth simulation, i.e., a model-based calculation, to predict the next mowing session. The grass growth simulation can be fed with weather data and data from the lawn area under consideration. Static or dynamic exclusion periods (e.g., Sundays and public holidays, rainy seasons, nighttime periods) can also be taken into account. Such methods for determining a suitable working time window based on a grass growth simulation are referred to here as "simulation methods."

[0008] The disadvantage of simulation methods is that they do not yet offer sufficient accuracy, consideration of local peculiarities or adaptability to user needs.

[0009] The object of the invention is to provide an improved technology for planning and controlling gardening equipment. The invention solves this problem with the features of the independent claims.

[0010] The invention is based on the fundamental idea that operational time planning can be improved by training an AI system based on working time windows generated by simulation and evaluated by users. Alternatively or in addition to working time windows generated by simulation, other initial work planning methods can also be used. For training, working time windows for the next use of the gardening tool are preferably first generated using a simulative method, i.e., based on a grass growth simulation. These working time windows are then provided to the user for evaluation before and / or after use of the gardening tool. The user is prompted to evaluate the working time window and, if necessary, other parameters in order to generate training data sets for training an AI system.Based on the training data sets, the AI ​​system, preferably an artificial neural network, in particular a time series model, can be trained to determine the optimal working time window for future uses of the gardening tool instead of the simulative method with grass growth simulation (or another initial work planning method).

[0011] The invention can be applied with various initial work scheduling methods. The initial work scheduling method refers to a method for determining working time windows, which is used at least temporarily to generate training data, and then to train the AI ​​system with the training data. After sufficient training, the AI ​​system can replace and / or supplement the initial work scheduling method.

[0012] The AI ​​system can also be referred to as an AI model, mowing scheduling model, or intelligent mowing scheduling (smart mowing). Several known trainable models (e.g., artificial neural networks) are available to those skilled in the art. The AI ​​system is preferably designed as a time series model. Simply put, the AI ​​system provides the desired output data (start time, working time, etc.) for appropriately structured input data (weather data, gardening equipment data, lawn property data, etc.). The present disclosure encompasses both the initial method for the preparatory collection of training data (e.g., for the initial training of the AI ​​system prior to its use) and the operating method of the trained AI system. Preferably, even during use of the AI ​​system, further training data is continuously collected to improve the AI ​​system.The disclosure also includes embodiments as a mere method of operating the AI ​​system without further user evaluation and / or generation of training data.

[0013] In the following disclosure of the invention and the various embodiments, the simulative method using a grass growth simulation as the initial work planning method is described by way of example. In all described or claimed embodiments, another initial work planning method or only the AI ​​system can be used as an alternative or in addition to the simulative method.

[0014] Another possible initial work planning procedure involves manually specifying working time windows. These manually specified working time windows can be used, if necessary with the addition of additional usage data (e.g., sensor data from the gardening tool), to generate evaluation queries. When the AI ​​model training function is activated, training data is generated based on the evaluated working time windows.

[0015] Alternatively or additionally, a user-configurable rule can be used as the initial work scheduling procedure, e.g., by specifying a time interval and duration. These work time windows generated by simple rules can be improved through evaluation queries and / or additional usage data (e.g., sensor data from the gardening equipment).

[0016] Another possible initial work scheduling method uses randomly generated work time windows. In particular, work time windows created using a heuristic and / or user-configurable rule can be varied with randomly generated values. These randomly generated work time windows (possibly with deployment data) can be used for evaluation queries and the generation of training data.

[0017] The trained AI system preferably uses similar or the same input data as the simulation process. The working time window is preferably determined based on input data that includes weather data (e.g., measurement data and / or forecasts from an online weather service for the location of the lawn to be maintained), gardening equipment data (e.g., type of gardening equipment, area performance, etc.), user profile data (e.g., user settings, personal cut-off times, grass height thresholds, etc.), lawn property data (e.g., location, area, and shape of the lawn, grass type, irrigation, soil condition, etc.), historical work data (e.g., time of the last lawn care), and / or calendar data (e.g., Sundays and public holidays).

[0018] Preferably, both the simulative method with the grass growth simulation and the AI ​​system determine a working time window with a start time and / or a working duration for a future use of the gardening tool. In the simulative method, the grass growth since the last lawn care is estimated using a grass growth simulation based on the weather and lawn data. As soon as the simulation predicts a grass height, i.e., cumulative grass growth, above a configurable threshold, the next use of the gardening tool can be planned. The exact start time and duration of use can be adapted to other conditions (e.g., wetness during and after rain, exclusion of rest periods). Various possible embodiments of a simulative method with grass growth simulation within the meaning of this disclosure can be found in European patent application EP23172108.5.

[0019] The term "working time window" is used in this disclosure as a generic term for a proposed working time window and, if applicable, corrected working time windows. Depending on the embodiment, the working time window can consist, in particular, of only a start time (e.g., for hand-held lawn mowers) or a start time and a working time (e.g., for automated robotic lawn mowers). The working time can be determined dynamically or be fixed.

[0020] The trained AI system has the advantage that, by training with data from evaluated, past deployments, diverse influences (e.g., local characteristics of the lawn, user preferences, model errors, configuration errors) can be better taken into account, even without explicit configuration by the user. Over time, the user automatically receives improved time suggestions for maintaining their garden, which better meet their personal preferences and actual conditions.

[0021] The rating query can be carried out in several ways depending on the embodiment and the time of the query, i.e. before or after use. The rating query is provided on a user's device, e.g. via an app on a smartphone or via a web app in a web browser. The rating query is structured such that the user is prompted to enter the predetermined user rating data. Preferably, the data to be rated (e.g. the suggested working time window) is displayed to the user. The rating query can comprise several query steps that can build on one another and include both qualitative (e.g. good, bad, longer, shorter, earlier, later) and quantitative assessments (e.g. desired start time, desired working duration, x hours / days earlier / later, x hours shorter / longer).

[0022] In an advantageous embodiment, the user rating data includes at least a qualitative rating (e.g., like / dislike) of the start time and / or the working duration of the working time window. The working time window can be in the past, present, or future at the time of the rating. If the user rates an aspect of the working time window negatively, further user rating data on this aspect is preferably requested (e.g., the desired working time window, a desired start time, a desired working duration, earlier or later, more or less grass growth until mowing).

[0023] By evaluating the machine-suggested working time window, a training dataset with additional information on the suitability of the suggested working time window can be generated, which can then be used to improve the system using machine learning methods. Qualitative assessments can be used, for example, for reinforcement learning methods. Alternatively or additionally, quantitative assessments, especially with explicit desired values ​​for the working time window as the "true" output data, can be combined with the corresponding input data to generate training datasets for supervised learning.

[0024] Thanks to the method according to the invention, the user can train the system to generate suitable working time windows for their gardening equipment without the manufacturer having to manually generate training data. Through their evaluations, which relate to their personal preferences, their individual garden, and their individual gardening equipment, the user can achieve user-specific improvements to the system during operation. Training is carried out intuitively for the user by answering the evaluation questions.

[0025] Alternatively or additionally, the user can also use their ratings to improve a system deployed across multiple users and devices. The ratings submitted by a large number of users allow large amounts of training data to be collected, which can be used in whole or in part to train a shared system.

[0026] If the user evaluates the working time window in advance, the user evaluation data can be used to correct the working time window for controlling the gardening tool. Preferably, a corrected working time window is generated based on the user evaluation data, which is then used to control the gardening tool instead of the originally suggested working time window. To generate training data, corrected working time windows can also be generated based on subsequent user evaluations, even if the tool has already been used.

[0027] If the user subsequently evaluates the working time window, the actual working time window and / or the work result (e.g., evenness of the mowed area) can be evaluated. Furthermore, operational data (e.g., sensor data from the gardening device or actually recorded weather data) can be taken into account.

[0028] In particular, multiple evaluation queries can be performed for the same use of the gardening tool. For example, one evaluation query can be performed in advance at the predicted working time window, another evaluation query shortly before on the same day of the working time window, and another evaluation query after the end of the use. Depending on the time of the user query, the user and the system have different information at their disposal. For example, the user can briefly estimate on the day of the planned working time window whether the current grass height already corresponds to the desired grass growth threshold at which mowing should begin. After the end of the use, the user can assess the duration of the mowing operation based on the condition of the mowed lawn.

[0029] The rating query can, in particular, include a push notification at one or more predetermined times. The rating query can be updated if, for example, the working time window has been recalculated due to a changed weather forecast or if the assignment has since occurred. Alternatively or additionally, the user can also enter user rating data on their own initiative, for example, via a deployment history.

[0030] The working time windows generated either by simulation or by the AI ​​system will be used to control the gardening equipment. In the case of an automated gardening equipment (e.g., a robotic lawnmower), the working time window can be transferred to the gardening equipment's control system via a device interface. For semi-automated gardening equipment (e.g., lawnmowers or manually operated garden tractors), the working time window can be transferred via the device interface to the gardening equipment itself or a user device (e.g., a smartphone) as a user aid.

[0031] Based on the predicted working time window and user rating data, training data sets can be generated and stored in a training database. This training data can be used to train an AI system that can generate improved working time suggestions based on new input data. Training can be performed either initially, when the AI ​​system is first put into operation, or for incremental improvement of the AI ​​system. The system is preferably trained in regular update cycles.

[0032] Preferably, a shared AI system is used for multiple users and / or devices. The AI ​​system can be trained with training data from a large number of users or devices. The AI ​​system is advantageously configured to process user-specific input data. In particular, the combination of a cross-user AI system and / or cross-user training and user-specific input data enables effective system improvement through large amounts of training data and user-specific work schedules.

[0033] Alternatively or additionally, a customized AI system can be deployed for the user or a user scenario (a combination of user, gardening tool, and / or lawn area), which can be specifically trained for the user or user scenario. Generic pre-trained AI systems are more advantageous, and can then be retrained with training data specific to the user or user scenario.

[0034] The AI ​​system can be used instead of or in addition to the simulation method with grass growth simulation to determine suitable working time windows. In particular, it is possible to compare the results of the simulation method and the AI ​​system and / or have them evaluated by the user.

[0035] The invention enables a time-based control of gardening tools that is more efficient in terms of resource utilization and is better adapted to the individual circumstances of the gardening tool's usage scenario.

[0036] Further embodiments and advantageous features of the invention are described below with reference to the drawings.

[0037] The invention is illustrated schematically and by way of example in the drawings. These show: Figure 1: a flowchart of a possible embodiment of the method with user evaluation in advance and correction of the working time window before use of the gardening tool; Figure 2: a flowchart of another embodiment of the method with user evaluation after use of the gardening tool; Figure 3: functionality of a simulative method with grass growth simulation; Figure 4: representation of a planned working time window for a robotic lawnmower in an app; Figure 5: representation of a schedule for rainy seasons, mowing seasons, and exclusion periods in an app; Figure 6: representation of a rating query in an app; Figure 7: process of a structured rating query with input of user rating data in an app; Figure 8: process of a structured rating query with input of user rating data in an app.

[0038] Two possible embodiments of the computer-implemented method are described in the Figures 1 and 2The computer-implemented process can be described as a series of steps, which can be performed in a different order than the one shown and described here.

[0039] At the beginning of the illustrated process, work planning (200) takes place. In this process, a working time window (TWO) for a future use of the gardening device (100) is determined in a simulative process with a grass growth simulation (201) and / or with a trained AI system (202). To determine the working time window (TWO), several input data (ID) are processed. Preferably, weather data (WD), lawn property data (LCD), gardening device data (GDD), historical work data (HOD) and / or calendar data (CD) are processed as input data (ID). The disclosure explicitly includes any possible combination of the input data (ID) disclosed here. The functioning of the simulative process is described further below with reference to Figure 3 explained in more detail.

[0040] The next step is Figure 1 an evaluation query (210). An evaluation query (EQ) of the working time window (TWO) is generated and provided on a terminal (110). The user can select the suggested working time window (TWO), for example, as shown in the Figures 4, 5 and 6displayed in an app. The user can start the rating query, in particular by pressing a corresponding button. The user enters their user rating data (UED). The user rating data (UED) can include various qualitative and / or quantitative ratings. Preferably, a qualitative rating (QE) of the working time window (TWO) or the start time (ST) and / or the working duration (DO) is entered. The qualitative rating can, for example, include a positive ("like", thumbs up, etc.) or negative ("dislike", thumbs down, etc.) rating. Alternatively or additionally, the user rating data can include quantitative ratings, for example a working time window (DTWO), a desired start time (DST), or a desired working duration (DDO) desired by the user.Instead of explicit times or durations, the user can also enter other forms of evaluation to adjust the working time window.

[0041] The user evaluation data (UED) is stored, preferably in conjunction with the associated evaluated working time window (TWO) and / or the associated input data (ID) used to determine the working time window.

[0042] In a further step (220), a training data set (TD) is generated. The training data set (TD) preferably comprises the evaluated working time window (TWO), the associated user evaluation data (UED), the associated input data (ID), and, if applicable, a corrected working time window (RTWO). The training data set (TD) can also comprise only a portion of this data and / or additional data. In particular, training data can be derived from the user evaluation data (UED) in an intermediate step. For example, a target value can be calculated from a qualitative evaluation (e.g., "later start time") and / or a quantitative evaluation (e.g., "mowing 1 hour longer") in an intermediate calculation step.

[0043] The training data set (TD) is preferably stored in a training database (250). The training database (250) can be specifically designed for the user or the respective user scenario.

[0044] In a particularly advantageous embodiment, the training data sets (TD) can be weighted depending on the user rating data (UED) and / or other weighting factors (e.g., user trustworthiness, plausibility of the rating data). The assigned weight can be used, in particular, in the training process of the AI ​​system (202) to give greater consideration to certain training data. In this way, ratings from certain users, in particular, can be incorporated more heavily into the system's training than others.

[0045] In a further optional step (230), a corrected working time window (RTWO) is calculated. Preferably, the corrected working time window (RTWO), the corrected start time (RST), and / or the corrected working duration (RDO) are determined on the basis of the user evaluation data (UED), if necessary with intermediate calculation steps. In the simplest case, the working time window, the start time, and / or the working duration are replaced with a working time window (DTWO), a desired start time (DST), and / or a desired working duration (DDO) explicitly requested by the user. Alternatively or additionally, the values ​​of the corrected working time window can be determined from intermediate calculations, in particular based on qualitative evaluations with a direction indication (e.g., later, earlier, longer, shorter) using predefined calculation methods or heuristics.

[0046] Preferably, the input data (e.g. exclusion times from the user profile or rainy periods) are also taken into account when calculating the corrected working time windows (RTWO).

[0047] The step of calculating the corrected working time window (230) preferably takes place before the training data set (TD) is saved so that the corrected working time window (RTWO) can be included in the training data set (TD).

[0048] In a further step (300), the gardening device (100) is controlled. The gardening device (100), preferably a robotic lawnmower (101), a garden tractor (102), or a hand-pushed lawnmower (103), is controlled via a control interface (401, 402). In the case of an automated robotic lawnmower (101), the working time window (TWO) or a corrected working time window (RTWO) can be transmitted directly via a control interface (401) to a controller of the gardening device so that the gardening device is activated at the start time. In the case of a semi-automated gardening device (e.g., a garden tractor), the working time window can be provided via a message interface (402) on a user's terminal device. The user can be given recommendations for operating the gardening device, which the user can then implement.

[0049] In an advantageous embodiment, fleets of multiple gardening tools of the same or different types can also be controlled. The working time windows can be created for a single tool or for multiple tools. Alternatively or additionally, the working time windows can be coordinated between the gardening tools and / or lawns.

[0050] After a sufficient number of training data sets (TD) have been collected, the AI ​​system (202) can be trained for the first time or again. In a step (500), the AI ​​system (202) is trained with training data (TD). Training preferably takes place asynchronously with the operation of the method, for example, during maintenance intervals or after a sufficient number of training data sets has been achieved.

[0051] Once the AI ​​system (202) has been trained at least initially, the AI ​​system (202) can be used in addition to or instead of the simulative method with grass growth simulation (201) to determine suitable working time windows.

[0052] Figure 2 shows an embodiment of the method in which an evaluation query (210) is made after the device control (300), ie after the use of the gardening device (100) in a previously determined working time window (TWO) or corrected working time window (RTWO).

[0053] If the evaluated use of the gardening tool has already taken place or is still in progress, the user evaluation (210) can take into account not only the working time window (TWO) but also usage data (OD). The usage data (OD) can be obtained from a gardening tool (100), for example, from a control unit or a sensor system of the robotic lawnmower, and / or entered via a terminal device (110). The usage data (OD) can include operating data of the gardening tool (100) and / or data on the course or result of the use of the gardening tool (100).

[0054] In possible embodiments, operational data (OD) can be recorded, for example, as the mowing resistance of a robotic lawnmower during operation or the actual duration of a garden tractor's mowing operation. Alternatively or additionally, the user can enter the result of the operation, e.g., the quality of the cut pattern or the grass height, as operational data.

[0055] The operational data (OD) can be the subject of the evaluation query (EQ) either alternatively or in addition to the working time window (TWO). For example, the user can be asked whether they are satisfied with the mowing result or whether the duration of the mowing operation was appropriate, since the mowing resistance already indicates early on that the lawn has been completely mowed.

[0056] The operational data (OD) can be incorporated into the training dataset (TD) (with or without user evaluation). By incorporating operational data (OD), particularly operating data from the gardening tool and / or user input data regarding the operational outcome, the AI ​​system can perform even more accurate operational planning through training than would be possible with the simulation method. This allows more input variables and boundary conditions, as well as influencing variables that cannot be explicitly considered in the simulation method, to be incorporated into the operational planning.

[0057] Figure 3shows a schematic of the function of a simulative method with grass growth simulation. Based on the input data (ID), in particular the weather data (WD), the lawn property data (LCD), and the historical work data (HOD), the grass growth simulation (201) is first used to estimate the grass growth (gg) since the last mowing operation. The grass growth (gg) is estimated and accumulated for several time periods. The point in time at which the accumulated grass growth exceeds a grass growth threshold (tgg) can be regarded as the earliest mowing time (EMT). The start time (ST) is planned according to this earliest mowing time (EMT). When planning the working time window (TWO), the working duration (DO) and various exclusion times (ET), e.g. wet periods during and after rain, rest periods, or closed periods due to the operation of other garden equipment, can be taken into account.The working time window can be planned in one or more parts, whereby a minimum duration can also be taken into account for each part.

[0058] In the Figure 3 In the simulation shown, grass growth (gg) is simulated at multiple time intervals (e.g., hourly or daily) based on the input data (weather data, lawn property data, etc.) and accumulated over time. On the day designated as d1, the user-configured grass growth threshold (tgg) is expected to be exceeded. The system plans a working duration (DO) of 5 hours. The system shown here plans single-part working time windows. The grass growth threshold (tgg) can preferably be set by the user as the maximum additional grass height until the next mowing operation.

[0059] The system considers nighttimes before and after sunset as exclusion times (ET), as well as a predicted rainy period on day 2 (d2) between 10 a.m. and 11 a.m., after which a buffer time for drying is taken into account. In this way, the simulation process uses the grass growth simulation (201) to plan the next working time window for the robotic lawnmower on day 2 (d2) between 12 p.m. and 5 p.m.

[0060] The working time window (TWO) proposed by the simulation process can now be presented to the user for evaluation before, during, or after the deployment in an evaluation query (EQ) in order to generate suitable training data (TD) for improving deployment planning based on the user evaluation data. After sufficient training of the AI ​​system (202), instead of the Figure 3 illustrated simulation procedure, a suggestion for the next working time window can be made by the AI ​​system.

[0061] The Figures 4, 5 and 6show possible representations of the working time windows (TWO) suggested by the system in an app on a user's device. Preferably, the user has already configured his system at the time shown, i.e., he has set his gardening equipment, the lawn area to be worked on, and his user preferences. The system suggests Figure 4 The next working time window for the gardening device ("smart mowing time") is Friday, October 2nd, between 3 pm and 9 pm (3 pm - 9 pm). The user can obtain more information about the creation of this schedule by clicking on the buttons ("tell me more", "how do you know"). The grass growth threshold (tgg) is set by the user in the case of Figure 4The mowing height is set to 3 cm. This means that the lawn (regardless of the cutting height set on the device) should grow by 3 cm since the last mowing session before it is mowed again. This can significantly increase the efficiency of the robotic lawnmower in terms of energy consumption and wear and tear compared to daily mowing. The user also has the option of selecting past mowing sessions ("mowing track") using a corresponding button.

[0062] Figure 5 shows a possible representation of a multi-part working time window on November 2, 2023. Rain is forecast between 8:00 a.m. and 10:00 a.m. The system schedules the first part of the mowing operation between 11:30 a.m. and 12:30 p.m. An exclusion time between 1:00 p.m. and 3:00 p.m., obtained via an IFTTT interface, is considered an interruption to the mowing operation. The second part of the working time window occurs between 3:00 p.m. and 8:00 p.m.

[0063] The rating query can be triggered in several ways by the system and / or the user. For example, the user can initiate a rating query (EQ) by clicking a rating button (e.g., the "thumbs up" symbol). Alternatively or additionally, the system can prompt the user (e.g., via a push notification) to enter user rating data (UED).

[0064] Figure 6 shows an example of an evaluation query (EQ) in an app. The evaluation query includes a suggested work window (TWO) with a start time (ST) on November 2nd at 3 pm and a work duration (DO) of 4 hours. The user can qualitatively and / or quantitatively rate the start time and / or work duration using appropriate buttons. Figure 6 the user can rate the start time and / or the working time as positive ("thumbs up") or negative ("thumbs down").

[0065] In an advantageous embodiment, the evaluation query is structured in several steps. The steps preferably build on each other depending on the evaluation result. For example, a qualitative evaluation can be requested first, and only in the case of a negative evaluation can a further query be requested, e.g., a qualitative evaluation with direction or a quantitative evaluation with desired explicit values.

[0066] The Figures 7 and 8 show an example of a sequence of representations in an app for multi-level evaluations of a working time window.

[0067] In Figure 7 The user negatively evaluates the suggested start time (ST) of 3 p.m. on November 2nd and positively evaluates the work duration (DO) of 4 hours. The user is then prompted to enter the desired start time (DST) by entering the desired date and time.

[0068] In Figure 8 On the other hand, the user rates the start time on November 2nd at 3 p.m. as positive, but the suggested working time (DO) of 4 hours as negative. Accordingly, the user is asked for their desired working time (DDO), which they can enter on a subsequent screen.

[0069] In addition to the embodiments shown in the drawings, the method can be implemented in a variety of other embodiments. The disclosure also encompasses all combinable embodiments resulting from combining the individual features disclosed herein. In particular, the sequence of the steps disclosed herein can be varied. The claimed method can also be defined using the steps disclosed in the description, whereby the subject matter of the claim is not limited to the sequence of these steps. Reference symbol 100 Garden tool garden device 101 Robotic lawnmower mowing robot 102 garden tractor garden tractor 103 mowing machine mower 110 End device terminal device 200 Work planning operational planning 201 Grass growth simulation grass growth simulation 202 AI system AI system 210 Rating query evaluation processing 220 Training data generation training data generation 250 Training database training database 230 Time window correction time window rectification 300 Device control device control 401 Control interface control interface 402 Message interface message interface 500 training training 600 Training data filtering TWO Working time window time window for operation ST Start time start time DO Working hours duration of operation RTWO corrected working time window rectified time window for operation DTWO desired working time window desired time window for operation DST desired start time desired start time DDO desired working time desired duration of operation QE qualitative assessment qualitative evaluation d1, d2 Day 1, Day 2, ... day 1, day 2, ... 99 Grass growth, per period grass growth, per period tgg Grass growth threshold threshold for grass growth ET Exclusion periods exclusion times EMT earliest mowing time earliest mowing time ID Input data input data GDD Garden equipment data garden device data wn Weather data weather data UPD User profile data user profile data LCD Lawn property data lawn characteristics data HOD historical work data historic operating data OD Operational data operational data TD Training dataset training data set UED User rating data user evaluation data

Claims

1. Computer-implemented method for determining a work time window (TWO) for a gardening device (100) for maintaining a lawn area, preferably for a robotic lawnmower (101), a garden tractor (102) or a lawnmower (103), wherein the work time window (TWO) is determined on the basis of input data (ID) and by means of a grass growth simulation (201) and the work time window (TWO) comprises at least a start time (ST) and / or a work duration (DO) for the use of the gardening device, wherein an evaluation query (EQ) is generated for evaluating the work time window (TWO) and the evaluation query (EQ) is provided on a terminal device (100) of a user (110) and user evaluation data (UED) of the work time window (TWO) are queried to generate a training dataset (TD) for an AI system (202).

2. Computer-implemented method according to claim 1, wherein the work time window (TWO) is further determined by means of the AI system (202).

3. Computer-implemented method for determining a work time window (TWO) for a gardening device (100) for maintaining a lawn area, preferably for a robotic lawnmower (101), a garden tractor (102) or a lawnmower (103), wherein the work time window (TWO) is determined on the basis of input data (ID) and by means of a trained AI system (202) and the work time window (TWO) comprises at least a start time (ST) and / or a work duration (DO) for the use of the gardening device, wherein an evaluation query (EQ) is generated for evaluating the work time window (TWO) and the evaluation query (EQ) is provided on a terminal device (110) of a user and user evaluation data (UED) of the work time window (TWO) are queried to generate a training dataset (TD), wherein the training dataset (TD) is for the AI system (202) or for an individual AI system that is specifically trained for the user or the user scenario and generates a rectified work time window that can then be used instead of the originally proposed work time window to control the gardening device.

4. Computer-implemented method according to claim 3, wherein the work time window (TWO) is further determined by means of a grass growth simulation (201).

5. A computer-implemented method according to one of the preceding claims, wherein the user evaluation data (UED) comprises at least one of the following elements: a desired work time window (DTWO), a desired start time (DST), a desired work duration (DDO) and / or a qualitative evaluation (QE) of the work time window.

6. Computer-implemented method according to one of the preceding claims, wherein the work time window (TWO) is rectified to a rectified work time window (RTWO) on the basis of the user evaluation data (UED), in particular if the user evaluation takes place before the work time window (TWO).

7. Computer-implemented method according to one of the preceding claims, wherein the work time window (TWO) or a rectified work time window (RTWO) for controlling the gardening device (100) is provided at a control interface.

8. Computer-implemented method according to one of the preceding claims, wherein the training dataset (TD') comprises at least one of the following elements: the input data (ID), the work time window (TWO), the rectified work time window (RTWO), the user evaluation data (UED) and / or operational data (OD).

9. Computer-implemented method according to one of the preceding claims, wherein the input data (ID) preferably comprise at least one of the following elements: weather data (WD), gardening device data (GDD), user profile data (UPD), lawn characteristic data (LCD), historical work data (HOD) and / or calendar data (CD).

10. Computer-implemented method according to one of the preceding claims, wherein one or more evaluation queries (EQ) are generated before and / or after the use of the gardening device (100).

11. Computer-implemented method according to one of the preceding claims, wherein - operational data (OD) from the gardening device (100) used and / or the user's terminal device (110) is obtained when the gardening device has already been used; AND / OR - wherein the operational data (OD) is processed to generate the evaluation query (EQ) and / or to generate the training dataset (TD).

12. Computer-implemented method according to one of the preceding claims, wherein a lack of user evaluation is interpreted as an implicitly positive evaluation of the work time window.

13. Computer-implemented method according to one of the preceding claims, wherein at least one training dataset (TD') for training the AI system (202) is stored in a training database (250).

14. Computer-implemented method according to one of the preceding claims, wherein the AI system (202) is trained with a plurality of training data (TD) from the training database (250).

15. Computer-implemented method according to one of the preceding claims, wherein the training datasets (TD) are filtered with a plausibility filter.