MOBILE FLOOR CLEANING DEVICE AND METHOD FOR OPERATING SUCH A DEVICE

Description

[0001] The invention relates firstly to an automatically movable floor cleaning device, designed with a device for creating photographic images of a surface to be traveled over, wherein the floor cleaning device has an evaluation unit which is designed to evaluate an image created by the device with regard to the type of floor covering, and to use a floor covering identified by the evaluation of the evaluation unit with regard to a travel strategy and / or with regard to a setting of cleaning parameters, such as floor clearance and / or size of secondary air openings and / or brush speed.

[0002] Furthermore, the invention relates to a method for operating an automatically movable floor cleaning device which has a device for creating photographic images of a surface to be traveled over, wherein images taken by the device are evaluated with regard to the type of floor covering and, depending on the type of floor covering identified by the evaluation, a travel strategy and / or an adjustment of cleaning parameters, such as distance from the floor and / or size of secondary air openings and / or brush speed, is carried out.

[0003] Floor cleaning devices of the type in question are known, particularly in the form of attachments for handheld household vacuum cleaners or self-propelled robots for cleaning and / or maintaining floors or floor coverings. For example, DE 10357637 A1 discloses a self-propelled floor cleaning device, which is preferably further configured for wet floor cleaning, in particular hard floors, such as tiled floors. A vacuum / sweeping robot is known, for example, from DE 10242257 A1. However, the device can also be, for example, a robotic lawnmower.

[0004] Also known are floor cleaning devices in the form of self-propelled floor cleaning devices which are provided with a camera, in particular a digital camera, in order to create photographic images of the area to be traveled over, in particular the floor area, and furthermore in particular of the floor area in the usual direction of travel of the device in front of the device, in order to detect abysses, such as stairs, in advance.

[0005] Furthermore, US 2005 / 0000543 A1, for example, discloses a cleaning robot that can be operated in different modes, such as a hard floor mode and a carpet cleaning mode. A floor sensor can be used to select one of the modes.

[0006] The document US 2007 / 0267570 A1 discloses a method for detecting an object using structured light, wherein the robot calculates a height difference between a position onto which the structured light is projected and a reference position.

[0007] The document JP 2003-346150 A further discloses a walking robot which has a sensor for determining a ground type in order to be able to balance itself on different surfaces.

[0008] In addition, cleaning devices with ultrasonic sensors for determining soil type are also known in the prior art, for example from EP 1 360 922 A2.

[0009] Against the background of the aforementioned prior art, the invention is concerned with the task of specifying a movable floor cleaning, processing or floor care device, as well as a method for operating such a device, which enables a favorable adjustment of the device with regard to the surfaces to be treated.

[0010] A possible solution to the problem is provided according to a first inventive idea in a floor cleaning device in which the evaluation unit is designed to recognize and classify surfaces recorded with the photographic images (FA) with regard to differences in their condition through the targeted use of digital image processing on the basis of predetermined features, wherein the floor cleaning device is set up by means of these differences to interrupt a cleaning process altogether and / or to leave or avoid a recognized floor area depending on the type of floor covering detected.

[0011] With regard to the method, it is proposed that, through the targeted use of digital image processing based on predetermined characteristics, areas captured with the photographic images are recognized and classified with regard to differences in their condition, whereby these differences, depending on the type of floor covering recognized, result in a cleaning process being interrupted altogether and / or a recognized floor area being left or avoided.

[0012] Based on the evaluation of the surface being driven on, it is possible to distinguish between ceramic, stone, tile or glass surfaces, natural wood or laminate surfaces, plastic surfaces or, for example, textile surfaces, short or long pile carpet surfaces.

[0013] With mobile floor cleaning devices, such as autonomous floor cleaning devices, the problem often arises of limiting or concentrating the cleaning activity on one or more specific areas. For example, with robots for wet floor cleaning, so-called mopping robots, it must be ensured that they do not apply water or moisture to floor coverings that are intolerant of water or moisture, such as carpets.

[0014] According to the proposed solution, through the targeted use of digital image processing based on predefined characteristics, surfaces captured in the images can be identified and classified with regard to differences in their condition. These differences make it possible to take appropriate measures regarding the floor cleaning device. One such measure is to interrupt the cleaning process altogether or, particularly in the case of autonomously operating floor cleaning devices, to leave or avoid the detected floor area. An additional measure can be, for example, adjusting the brush speed if a rotating brush is present, adjusting the suction power if a suction fan is present, or adjusting the moisture output if a water or moisture application device is present.The measures described above can also be combined, for example adjusting the brush speed and adjusting the suction power accordingly.

[0015] The input information is preferably a digital photograph of the area to be traveled by the floor cleaning device, or, if applicable, of the floor area directly traveled over. The camera on or in the device can be positioned so that the area in front of the device in the usual direction of travel is photographed. However, it can also be positioned or positioned underneath the device and thus primarily capture the area directly traveled over.

[0016] Further preferably, photographic images are taken regularly, more preferably at fixed time intervals of up to 2 seconds, more preferably up to 1 second. In this context, the time interval between two photographic images is more preferably dependent on the travel speed of the device. Thus, at higher travel speeds, more images are provided per unit of time than at lower travel speeds.

[0017] The preferably digital photo is evaluated by software, which more preferably runs on a microprocessor / microcontroller, particularly with regard to the degree of reflection and / or a pattern detected in the photo. In a preferred embodiment, comparison values ​​or comparison patterns are stored in a database for these determined values ​​or patterns, with cleaning parameters assigned to these comparison values. This makes it possible, for example, to distinguish between a carpet and a hard floor.

[0018] Furthermore, the proposed solution allows for the limitation of the working area of ​​self-propelled floor cleaning devices without having to prepare the surface, i.e., the floor to be traveled over and the surrounding area, with markings, and preferably without the use of environmental maps. For example, the device can be instructed not to leave a carpeted or non-slip floor. Furthermore, the use of digital image processing allows the device to independently select suitable cleaning modes, thus achieving improved cleaning results while preferably maintaining the device's energy efficiency.

[0019] Supplementary or alternative features related to the proposed method can also be found in the description of device-related features. This also applies vice versa.

[0020] Thus, in a preferred embodiment, the evaluation can be carried out in the floor cleaning device itself. Preferably, the software required to evaluate the photographic recordings, in particular a digital image recording, is contained in an electronic, more preferably non-volatile, memory of the movable device, for example the floor cleaning device. It can also preferably be evaluated using a microcontroller provided on the device. Preferably, the commands for setting cleaning or processing parameters or the travel strategy are also generated in the device itself. They can be immediately converted into control signals relating to, for example, the brush or the like. However, they can also be displayed to a user for modification and / or approval, for example.

[0021] In an alternative embodiment, the evaluation is performed externally with respect to the floor cleaning device, for example, through preferred wireless transmission of the photographic image taken by the floor cleaning device to a preferably stationary or mobile device. A stationary device can, for example, be a charging station of a self-propelled floor cleaning device, at which station the device's batteries can preferably be charged. Alternatively, the evaluating device can be a computer, such as a PC, notebook, tablet PC, or even a smartphone, with a wireless connection, for example, Wi-Fi or Bluetooth, also preferably provided.In this case, bidirectional communication is preferably provided between the floor cleaning device and the device evaluating the recording, so that commands for setting the travel strategy and / or cleaning parameters can be sent to the floor cleaning device via the same wireless connection paths.

[0022] Interaction can also be implemented in a "smart home" manner, for example, using a television to visualize the results. In a further development, the floor cleaning device itself can have a display and controls.

[0023] In a further preferred embodiment, the area to be photographed can be illuminated by a lighting element provided in the floor cleaning device. In one embodiment, this is a conventional light, for example, a halogen light or a preferably white LED. In this context, it is further preferred that the lighting element emits a defined white light.

[0024] The lighting can be supplemented or alternatively provided by infrared light.

[0025] Alternatively or in combination, the camera operates in the invisible spectral range. This reduces the influence of ambient light and can also contribute to the analysis. The spectral values ​​of the light emitted by the light element are then adapted to the camera's operating range.

[0026] It is also preferred that the image is a color image and can be evaluated with regard to color.

[0027] Alternatively, or in combination, the evaluation of overexposed areas (especially white areas) in an image can be performed. For example, surfaces that are sealed, i.e., largely waterproof and smooth, exhibit high gloss and thus a high degree of reflection.

[0028] Gloss and color measurements for surface characterization are well known. Gloss can also be recognized in a digital image as an indicator of a closed surface. Carpets, i.e., floors with a pile containing pile threads, do not exhibit a glossy core in the corresponding digital image because the carpet fibers strongly scatter light, resulting in diffuse reflection.

[0029] For example, wooden floors or other flooring with characteristic patterns or grains can also be identified through analysis, with corresponding areas in the color space being determined using statistical methods. For example, an analysis of the color distribution within a photograph is preferably carried out by counting how many pixels of the image lie within a given color range representing wood.

[0030] Two approaches in particular are provided for the analysis of a joint structure, for example in a tiled floor.

[0031] In this way, depth information can be calculated from two images (taken either simultaneously with a stereo camera system or sequentially with a monocular camera system; the camera systems are also important independently). The algorithms required for this are well-known textbook knowledge. In the depth image, for example, tiles can be recognized as smooth surfaces in a single plane. Joints are visible as interruptions and slight depressions. If necessary, predefined data on typical joint widths, for example, can also be incorporated into the analysis.

[0032] Alternatively, or in combination with this depth information analysis, an analysis of edge and angle information can be performed. The joints are visible as relatively long edges in the image, which usually run in two orthogonal preferred directions. The detection of the edges and the estimation of the edge orientation can be performed using known edge filters (e.g., Sobel filters or Canny filters). Approaches for suppressing short edges are also known. This approach is particularly suitable for monocular camera systems. If a downward-facing camera system is used, the right angles between the joints are also displayed as such. In addition, the width of the joints can also be evaluated.

[0033] Based on depth information, the floor level on which the floor cleaning device moves can be adjusted. On very smooth or flat surfaces, there are hardly any deviations from this level; on soft, long-pile floors, for example, there are usually significant deviations between the actual height in the depth image and the adjusted level.

[0034] In addition, well-known methods for texture analysis are also provided. Filter banks, such as Gabor filters, are often used for this purpose. With appropriate filter parameterization, longer edge segments can be detected. If these are distributed across the entire surface and largely oriented in a preferred direction, there is a high probability that this is wood grain.

[0035] In addition, the characteristic fiber structure of a carpet can also be recognized, especially due to identification via a known pattern.

[0036] By using multiple features, i.e. feature combinations, the risk of misclassification can be significantly reduced.

[0037] As an alternative to, or in combination with, the independent evaluation of the photographic image by the floor cleaning device or an external device communicating with the floor cleaning device, a further development of the invention provides that a photograph of an area can be defined by the user as being to be driven over or not to be driven over. Thus, in one embodiment, before starting up the floor cleaning device, the user shows it the areas that should either be avoided (creation of a negative list) and / or the areas on which the device may / should be located (creation of a positive list). In an exemplary living space environment, the user preferably creates lists on site by placing the device on corresponding surfaces, having the device characterize the surface, and adding this surface to the positive or negative list.

[0038] In a preferred embodiment, positive and / or negative lists are stored in a non-volatile memory of the floor cleaning device and / or the associated base station and / or a computer or the like that communicates with the device directly or indirectly, for example via the base station.

[0039] The floor cleaning device is preferably equipped with a monocular camera system directed forward in the direction of travel or, more preferably, directly downward. Thus, the arrangement of such a camera in the floor area of ​​the floor cleaning device is further preferred. Monocular camera systems consist of a single camera. With only one camera image, such a system can only provide intensity information. To calculate depth information, two consecutively recorded images are preferably combined.

[0040] Alternatively, a light pattern can be actively projected, for example through laser lines or infrared spots, whose position in the image allows conclusions to be drawn about the depth information and thus about the 3D structure of the area currently being driven on.

[0041] In a further preferred embodiment, the device comprises multiple devices for creating photographic images, i.e., multiple, in particular, digital cameras, which preferably image the same surface area at different angles and / or from different directions. This preferably results in a stereo camera system. The two or more cameras provided are arranged offset from one another with a partially overlapping field of view. Such a stereo camera system provides intensity information and allows the calculation of depth information in the overlapping area of ​​the fields of view without relying on chronologically successive camera images.

[0042] A camera system directed forward in the usual direction of travel can also be used for other purposes, such as for the navigation of a self-propelled floor cleaning device or for control by gestures of the user.

[0043] A downward-facing camera system is preferably used solely for surface classification. The preferred location near the bottom of the device allows for favorable shadowing of the camera system against environmental influences, such as changes in lighting or reflections on the surface. Constant lighting conditions can be achieved through active illumination, for example, with LEDs.

[0044] The one or more cameras are preferably configurable in their position, height orientation, and more preferably also in terms of viewing angle and direction.

[0045] With regard to the method, before implementing a travel strategy or the like, depending on an evaluation of the images, a travel over an area to be cleaned is preferably carried out while taking the images, during which no immediate reaction is made to different surface formations detected. In this learning or exploration phase, particularly when the floor cleaning device is designed as a self-propelled floor cleaning device, the entire floor of a living room or apartment is preferably traveled over without simultaneously cleaning the floor. During this learning or exploration phase, data on preferably all surface formations (soil types) that occur are recorded via the one or more cameras and the provided evaluation unit.If the floor cleaning device is designed as a self-propelled floor cleaning device, the movement to record the data is carried out automatically or under the guidance of the user, for example via a remote control.

[0046] An autonomous exploration is preferably based on a travel strategy for cleaning floor surfaces, in particular according to DE 102010000174 A1.

[0047] During such exploration, the entire living space is preferably covered, ensuring that all existing floors are included for data collection. The user does not need to supervise the exploration.

[0048] During user-controlled exploration (teach-in), the self-propelled floor cleaning device is preferably controlled remotely by the user, for example, using a remote control or a mobile computer. Such a user-controlled exploration is known from DE 102009024990 A1 or DE 102009052629 A1.

[0049] During user-controlled exploration, the user must ensure that all floors present in the living space are adequately covered. Accordingly, continuous monitoring of the self-propelled floor cleaning device is necessary during such a sweep.

[0050] The recorded images are preferably analyzed using evaluation software, possibly after the actual flooring has been completed. This analysis preferably takes place during an evaluation phase, in which the collected data is categorized by the evaluation software into different floor types, such as tiles, laminate, parquet, or soft flooring, each of which requires different cleaning techniques (e.g., heavily damp mopping, lightly damp mopping, or vacuuming only). This assigns a unique floor type to each floor or floor section. The categorization of the collected sensor data into different floor types is preferably carried out according to factory-specified characteristics.Features that can be used include, for example, intensity information (e.g., color, gray values, texture, pattern, gloss, or contrast), features derived from depth information (e.g., smoothness of the floor, surface structure of the floor, or presence of joints), or combinations thereof.

[0051] The evaluation phase takes place either after the learning or exploration phase (offline method) or in parallel (online method). With the online method, soil classification is preferably performed only after all data from the entire floor area in the living space has been collected. In contrast, with the offline method, soil classification is performed immediately during the learning or exploration phase.

[0052] In a further preferred embodiment, the evaluation result is visualized on a display or communicated to the user via a voice output. The user can then confirm the correctness of the classification or make corrections if necessary.

[0053] Visualization of the results and interaction with the user can be done via a mobile device, a conventional computer or via a display and control elements that are provided directly on the floor cleaning device.

[0054] For each detected floor, the user is shown the floor, preferably in the form of a photograph (image) or as a marked area on a map, the floor type suggested by the system, and the cleaning parameters assigned by the system. Based on this display, the user can confirm the classification or make changes. This detects and corrects any misclassifications, thus preventing the risk of incorrect cleaning.

[0055] Thus, tracking a travel strategy based on an evaluation is only carried out after the user has confirmed the evaluation results transmitted to the user. The configuration generated in this way is preferably permanently stored in the floor cleaning device and can be used during the cleaning phase. The previously described steps—learning or exploration phase, evaluation phase, and user confirmation—therefore do not need to be performed before each cleaning run.

[0056] However, if the self-propelled floor cleaning device is transported to a different environment, it may be necessary or advisable to expand the existing configuration or to create a new configuration.

[0057] During the cleaning phase, the floor cleaning device systematically covers the entire floor area of ​​one or more rooms, cleaning them according to the cleaning parameters configured in the previous step. The cleaning phase can be performed immediately after the previous steps. However, it can also be performed independently, as the configuration created in the previously described steps is permanently stored in the floor cleaning device. The preferred cleaning strategy is the one described in DE 102010000174 A1.

[0058] Different approaches are possible to adjust the cleaning parameters according to the configuration saved in the previously performed steps.

[0059] For this reason, it is preferable to initially adjust the cleaning parameters based on the current position of the floor cleaning device in the room. To achieve this, the self-propelled floor cleaning device continuously locates itself using a map of its surroundings known to it, allowing its position to be determined. In this case, the cleaning parameters are adjusted independently of the intended and described sensor technology for detecting and analyzing surfaces.

[0060] Alternatively, the cleaning parameters can be adjusted by continuously scanning the area currently being traveled with the sensors and continuously classifying the current sensor data. The cleaning parameters are then immediately adjusted based on the currently detected floor type. Such a system can operate without locating the floor cleaning device in the environment and independently of an environmental map.

[0061] A combination of the approaches described above is preferred. For example, when performing a cleaning run on an area that has already been traversed and analyzed, additional images are taken and evaluated. The floor cleaning device locates itself on the surrounding map, adopts the configuration stored for the position, and verifies it with the currently determined sensor data from the photographic images. If the stored configurations and the currently detected configurations do not match, a reaction is triggered, for example, in the form of a message to the user. Alternatively, or in combination, universally applicable cleaning parameters for this area can be used in such a case ("careful cleaning"). This area can also be left uncleaned.

[0062] The described system, in particular consisting of a camera system and a miniature computer for evaluating the sensor data, is designed as a sensor module, which can be used in addition to a self-propelled floor cleaning device, in attachments for household vacuum cleaners or attachments for handheld vacuum cleaners or above-floor cleaning devices.

[0063] Such a system can also be used to prevent user errors caused by improper cleaning. In the event of incorrect cleaning, such a system can, for example, provide the user with visual feedback (e.g., a red light), optical feedback (e.g., shaking the handle), or alternatively, turn off the floor cleaning device.

[0064] Furthermore, such a sensor module can be used to automatically adjust the suction power of a conventional vacuum cleaner to the surface to be cleaned (e.g. curtains, upholstery, work surfaces, etc.).

[0065] The features described above can also be applied to so-called robotic lawnmowers, in particular to ensure that the robot moves alone on a lawn. By evaluating the image of the surface being driven over, the device prevents it from moving into areas adjacent to the lawn. Preferably, the evaluation of the photographic image is carried out based on identification in the color space and / or via the structure. Identifying lawns via the color space proves advantageous in this case, as a particularly characteristic area is defined by the pigment (leaf green). The lawn structure also reveals a clear pattern, which is preferably evaluated as additional information to the color space information.

[0066] With regard to the creation of a positive or negative list to be carried out by the user in one embodiment, it is further advantageous if, in the case of a robotic lawnmower, the inclusion of the floor area "lawn" is already entered into a positive list at the factory.

[0067] The invention is explained below with reference to the accompanying drawing, which, however, represents only one exemplary embodiment. The drawing shows: Fig. 1: a perspective view of a self-propelled floor cleaning device, with a schematic representation of different floor surfaces to be traveled over; Fig. 2: the bottom view of the floor cleaning device; Fig. 3: a schematic representation of the creation of a photographic image of the floor surface to be traveled over and the evaluation of the image; Fig. 4: one of the Figure 3 appropriate presentation, recording and evaluation of a data set compared to the Figure 3changed floor area; Fig. 5 shows a schematic vertical sectional view of the recording area of ​​the floor cleaning device; Fig. 6 shows an exemplary photographic image of a floor area with a reflection core area within a reflection area identified during the evaluation; Fig. 7 shows one of the Figure 6 corresponding illustration, but relating to a recording of a floor area without a reflection core area; Fig. 8 a flow chart relating to the preferably software-based decision-making process for setting a travel strategy and / or cleaning parameters; Fig. 9 the schematic floor plan of a multi-room apartment during a learning phase of the floor cleaning device for analyzing the floor areas; Fig. 10 one of the Figure 9 corresponding representation, but concerning the cleaning phase.

[0068] It is presented and described first with reference to Figure 1a floor cleaning device 1 in the form of a self-propelled floor cleaning device with a chassis which, on its underside facing the floor 2 to be cleaned, carries electric motor-driven moving wheels 3 and a brush 4 protruding beyond the lower edge of the chassis floor, also driven by an electric motor. The chassis is covered by a device hood 5, whereby the floor cleaning device 1 of the illustrated embodiment has a preferred circular layout.

[0069] Furthermore, the floor cleaning device 1 preferably has a suction mouth opening 6 in addition to or as an alternative to the brush 4. In this case, an electric motor-driven suction fan motor is also arranged in the floor cleaning device 1.

[0070] The electrical supply of the individual electrical components of the floor cleaning device 1, such as in particular the electric motor of the travel wheels 3, the electric drive of the brush 4, more preferably the suction fan and, in addition, further electronics provided in the device for controlling the same, is achieved via a rechargeable battery (not shown).

[0071] The floor cleaning device 1 preferably has a travel strategy that can be adapted, for example, to the detected environment and / or to a degree of soiling of the floor 2.

[0072] There is a need to adapt this travel strategy, preferably in combination with it or alternatively, cleaning parameters such as the speed of the brush 4 and / or the suction power to the type of floor covering.

[0073] For this purpose, a device 7 in the form of an electronic camera 8 is preferably provided on the bottom of the device's chassis for taking photographs of the floor area traveled by the floor cleaning device 1. In addition to the electronic camera 8, the device 7 also includes a lighting element 9.

[0074] The lighting element 9 preferably emits a defined white light and is more preferably a white LED.

[0075] As can be seen particularly from the schematic representation in Figure 5 As can be seen, the camera 8 and the light element 9 are preferably aligned at the same or comparable angle to a vertical directed on the floor 2, with a preferred alignment angle alpha or beta to the vertical of 30 to 60 degrees.

[0076] The photographic image of the floor surface created is preferably evaluated for specific features within the floor cleaning device, more preferably within an integrated evaluation unit 10. In particular, digital image processing methods and algorithms are available for feature extraction from the individual images. In particular, the gloss, color, and structure of the photographic image are considered. Analyzed and evaluated images are preferably compared with features of a more preferably stored database (see schematic representations in the Figures 3 and 4 Depending on the evaluated and detected floor pattern, a specific movement strategy (e.g., leaving the current floor covering) is initiated and / or predefined cleaning parameters are adapted to the detected floor covering.

[0077] The identification of sealed surfaces (e.g., tiled floors) is preferably carried out by examining their gloss. Floor surfaces that are sealed, i.e., largely waterproof and smooth, exhibit a high gloss, i.e., a high degree of reflection. Gloss is also recognizable in the preferably digital FA image as an indicator of a closed surface. Figure 6 A photographic image FA of a tiled or wooden floor area is shown. Within the illuminated area R, shown in dashed lines, a bright (white) core area K, shown in the figure as a solid line, is detected. This core area K is preferably detected by pattern recognition algorithms.

[0078] Figure 7shows a photograph FA of a carpet. This also initially shows an illuminated area R. However, since the carpet fibers strongly scatter the light, resulting in diffuse light, no core area K can be determined here.

[0079] Wooden floors or wood-coated floors are preferentially identified in color space. For example, known gesture recognition algorithms identify a person's face and hands by combining pixels that occupy a color space typical for human skin into a single area. In a similar way, wood, for example, can be uniquely identified, provided that corresponding color space areas have been determined using statistical methods.

[0080] Carpets are preferably identified by their fiber structure. A characteristic feature of a carpet is its fiber structure. While this structure is difficult to detect in photographs of velour, a clear pattern can be seen in carpets with loose-ended pile threads.

[0081] The identification methods described above are preferably combined with each other in order to clearly identify the floor covering.

[0082] Preferably, the floor cleaning device 1 determines the type of floor covering during automatic travel and cleaning of the floor 2 and reacts independently based on predefined parameters. This is particularly useful for floor cleaning devices that are capable of wet cleaning the floor 2 if necessary. If, for example, carpeting is detected, this area is preferentially avoided by a so-called mopping robot or at least the application of moisture is interrupted.

[0083] Figure 8shows a decision flow diagram for an automatically deciding floor cleaning device 1. Starting from point A, it is first determined whether a white core area K can be seen in the illuminated area R, corresponding to a reflection of the light element 9. If this is the case, it is concluded that the surface is hard and at point B it is determined from the color within the photographic image whether this is a wood color. If a wood color was determined, this is in particular a sealed wooden floor or, for example, laminate (point C). In a preferred embodiment, with regard to the cleaning parameters to be set, this results in a low brush power, i.e. low brush speed, being set and only a small amount of moisture being applied, provided a liquid application device is provided.

[0084] If no wood color can be determined, the floor is linoleum or PVC, for example, or alternatively, tile, granite, or sealed stoneware (point D). Here, too, a relatively low brush speed is automatically selected. The amount of moisture applied is normal in this case.

[0085] If no core area K was identified under point A (point E), this initially leads to the fluid application being deactivated, provided a corresponding device is available. This also initially leads to an analysis of whether a wood color tone can be detected in the image. If such a wood color tone is present, it is preferably an open-pored and thus diffusely reflective wooden floor (point F), whereupon the brush power is preferably automatically set to a low level.

[0086] If no wood color tone was detected, the photograph is checked at the next point G to determine whether a fiber structure is recognizable. If fibers are detected in the image (point H), the carpet is preferably a long-fiber carpet, with the result that brush 4 is preferably deactivated. Such a carpet is preferably vacuumed alone. If no fibers are detectable in the image (point J), the carpet is preferably a velour carpet, which should be cleaned with a higher brush power.

[0087] Figure 9shows a schematic example of a floor plan of a multi-room apartment. The floor areas of the individual rooms are designed differently, in particular by the formation of a parquet floor (Pa), tiled floors (Fl), laminate floor (La), and a soft floor (We). Furthermore, in a room with a laminate floor (La), a carpet runner is laid, correspondingly representing a soft floor (We).

[0088] During a learning or exploration run of the floor cleaning device 1 (compare curved line L in Figure 9 ) Data is collected via the camera recordings and the subsequent analysis, which are then evaluated either simultaneously or subsequently. To collect the data, all existing soil surfaces and soil types are traveled over. The trajectory of a user-controlled exploration is shown, in which the floor cleaning device 1 is controlled by the user, for example, using a remote control.

[0089] The determined floor data are stored in a preferably non-volatile memory of the floor cleaning device 1, further preferably in combination with the environmental data of a further stored environmental map of the premises.

[0090] One like in Figure 10 The cleaning run of the floor cleaning device 1 shown is preferably carried out automatically using an environmental map, with a predetermined travel strategy being used more preferably.

[0091] In Figure 10 An example cleaning run is shown. During the cleaning run, the various cleaning parameters are adjusted to the determined and stored soil data.

[0092] Thus, with the appropriate design of the floor cleaning device 1, tiled floors Fl are subjected to very wet mopping (represented by the dashed line), while laminate floors La are subjected to light damp mopping with a lower proportion of liquid per m² than tiled floors (represented by the dotted line). Parquet floors Pa and soft floors We are preferably only vacuumed, not mopped (represented by the solid line).

[0093] Furthermore, when carrying out a cleaning run on an area that has already been examined and analyzed, additional FA images are created and evaluated. Figures 9 and 10A laminate floor La is partially covered with a soft floor We in the form of a freely laid carpet runner. In this partial floor area, preferential cleaning occurs, here – as shown by the solid line in the area of ​​the soft floor We – as a result of vacuuming alone.

[0094] A shift of such an exemplary carpet bridge to a different position relative to the laminate surface is detected in the event of a continuous analysis of the floor or surface during the cleaning run, which results in a reaction from the floor cleaning device 1. If a change compared to the stored values ​​is detected, the floor cleaning device 1 reacts, for example, with gentle cleaning, for example with the brush and wet cleaning deactivated, and furthermore, for example, with low suction power. Alternatively, this area can also be passed over without being cleaned, and a corresponding message is sent to the user. In this regard, it is also possible to compare the currently detected configuration with a stored configuration in the immediate vicinity of the current position of the floor cleaning device 1.If these configurations match, the relevant cleaning parameters are applied to the currently scanned area. List of reference symbols:

[0095] 1 Floor cleaning device FA Recording 2 Floor F1 tiled floor 3 Bicycle La Laminate flooring 4 brush Pa Parquet flooring 5 Device cover We soft floor 6 Suction mouth opening 7 Furnishings 8 camera 9 Lighting element 10 Evaluation unit α angle β angle A Point B Point C Point D Point E Point F Point G Point H Point I Point J Point K Core area L line R Area

Claims

1. An automatically traveling floor cleaning appliance (1), designed with a device (7) for creating photographic pictures (FA) of a surface to be travelled over, the floor cleaning appliance (1) having an evaluation unit (10) which is set up to evaluate a picture (FA) with regard to the type of floor covering produced by the device (7), and to use a floor covering recognised by the evaluation of the evaluation unit (10) with regard to a traversing strategy and / or with regard to an adjustment of cleaning parameters, such as floor spacing and / or size of secondary air openings and / or brush speed, characterised in that the evaluation unit (10) is designed to detect and classify surfaces detected with the photographic pictures (FA) by the targeted use of digital image processing differences with regard to differences in their properties means of on the basis of predetermined features, whereby, based on these differences, the cleaning appliance (1) is enabled to interrupt a cleaning process as a whole or / and to leave or avoid a detected floor surface depending on the type of floor covering detected.

2. The floor cleaning appliance (1) according to claim 1, characterised in that the evaluation can be carried out in the cleaning appliance (1) itself.

3. The floor cleaning appliance (1) according to one of the preceding claims, characterised in that the photographically captured surface (R) can be illuminated by a lighting element (9) that is provided in the floor cleaning appliance (1).

4. The floor cleaning appliance (1) according to claim 3, characterised in that the lighting element (9) emits a defined white light.

5. The floor cleaning appliance (1) according to one of the preceding claims, characterised in that the picture (FA) is a colour picture and can be evaluated with regard to the colour.

6. The floor cleaning appliance (1) according to one of the preceding claims, characterised in that the evaluation can be performed with regard to overexposed areas, namely white areas, in an image.

7. The floor cleaning appliance (1) according to one of the preceding claims, characterised in that a picture (FA) of a floor covering can be defined by the user as whether to be travelled on or not to be travelled on.

8. A method for operating an automatically movable floor cleaning appliance (1), which has a device (7) for creating photographic pictures (FA) of a surface to be travelled over, wherein pictures (FA) that have been taken by the appliance (1) are evaluated with regard to the type of floor covering and, depending on the type of floor recognised by the device (7) by the evaluation, a traveling strategy and / or a setting of cleaning parameters, such as floor clearance and / or size of secondary air openings and / or brush speed is carried out, characterised in that by the targeted use of digital pictures processing on the basis of predetermined features, surfaces detected with the photographic pictures (FA) are recognised and classified with regard to differences in their properties, whereby in view of these differences, depending on the type of floor covering detected a cleaning process as a whole is interrupted or / and a detected floor surface is left out or avoided.

9. The method according to claim 8, characterised in that prior to carrying out a traveling strategy or the like and in dependence on an evaluation of the pictures (FA), traveling over a surface to be cleaned is carried out while creating the pictures (FA), with no immediate reaction taking place to surface formations detected as being different.

10. The method according to one of the claims 8 or 9, characterised in that said traveling is carried out automatically or under the guidance of the user.

11. The method according to one of the claims 8 to 10, characterised in that the created pictures (FA) are analysed by means of an evaluation software, optionally at a time following traveling over the surface.

12. The method according to one of the claims 8 to 11, characterised in that a result of the evaluation is visualised to the user on a display or is communicated using a voice output.

13. The method according to one of the claims 8 to 12, characterised in that the pursuing a traveling strategy dependence on a performed evaluation is undertaken only after the user confirms the result of the evaluation communicated to the user.

14. The method according to one of the claims 8 to 13, characterised in that, when carrying out a cleaning run on a surface which has already been travelled over for analysing and which has been analysed, additional pictures (FA) are created and evaluated.