MAPPING A GROUND SURFACES

Description

[0001] The invention relates to mapping a ground surface. In particular, the invention relates to determining the boundaries of a covering lying on the ground surface.

[0002] US 2017 / 273527 A1 (HAN SEONG JOO [KR] ET AL) September 28, 2017 (2017-09-28) reveals a cleaning robot that creates a map of the environment.

[0003] The invention is defined in the attached independent claims. A cleaning machine is configured to traverse and clean a floor surface in a household. The cleaning process includes, for example, dry and / or wet cleaning of the floor covering. The cleaning machine can, for example, comprise a robotic vacuum cleaner, a robotic mop, or a combined device. Typically, the cleaning machine scans its surroundings and creates a map, which it uses to navigate, for example, between furniture and other obstacles. A user can view and edit the map. For example, a zone can be defined that the cleaning machine should not enter or clean, or a zone requiring particularly intensive cleaning, such as a walkway or the area under a dining table, can be specified.

[0004] The cleaning machine can detect different floor coverings and treat them accordingly. For example, it can distinguish between a long-pile carpet and a short-pile carpet. The carpet can be vacuumed and cleaned with a brush roller, while the rug can be vacuumed only to prevent the brush roller from getting caught in the carpet fibers. Hard floors are also distinct from carpets, as they can be wet-cleaned in addition to dry-cleaning.

[0005] Detecting floor coverings like carpets is often inaccurate, which can limit the usefulness of the mapping. A detected carpet often appears jagged to the user on the map, with uneven edges and "holey" areas in the center. To at least create filled areas, a workaround is sometimes to simply draw an extra-wide line around the carpet marker, but this also makes the overall display of details on the map less reliable. The user may then face considerable effort in trying to derive helpful settings for controlling the processing machine from the floor area scan.

[0006] It has been proposed to process floor surface scans using machine learning methods to improve the detection of a carpet lying on the floor. However, training such a device is very complex and requires a large amount of high-quality training data, which must first be collected and prepared.

[0007] One of the problems underlying the present invention is to provide an improved technique for mapping a floor area in a household. The invention solves this problem by means of the subject matter of the independent claims. Dependent claims describe preferred embodiments.

[0008] A method for mapping a floor area in a household comprises steps of scanning the floor area and determining sections of the floor area where a predetermined first condition of the floor area is present; entering the determined sections into an occupancy map of the floor area; applying a binary morphological filter to the occupancy map; and determining first boundaries of a covering lying on the floor area based on the occupancy map.

[0009] The process can be performed locally within the home and requires no external resources. For example, the process can be carried out partially or entirely using a processing machine. By utilizing established image processing techniques, even scan data that initially contains moderate to severe errors can be further processed to achieve a good approximation of the actual boundaries of the flooring. The necessary operations can also be performed in an acceptable timeframe using a less powerful processing device. The process is deterministic and does not require the use of complex technologies such as artificial neural networks.

[0010] The binary morphological filter operates on a matrix whose cells each contain a first or second predetermined value. This matrix is ​​well-suited for graphical representation of the floor area. One of the two possible values ​​for a cell can indicate that the predetermined condition is present, and the other value indicates that the condition is not present. This condition can be associated with the carpet. The carpet can be graphically represented on the matrix and easily manipulated using the binary morphological filter.

[0011] The matrix can form a standalone map or be implemented as a dedicated layer within a larger map, which may also include further information about the household. The process can also be applied multiple times to identify objects with several different characteristics. A predetermined characteristic can be assigned to a map or map layer, and the processing described therein can be executed independently across the different maps or layers.

[0012] Determining the properties of a ground surface can be performed in a combined manner, so that the ground surface only needs to be scanned once to obtain information about several different objects on the ground. For example, a property can be determined based on a value within a predefined range. This range can be divided into several sections, which are assigned to maps or layers. A property of the ground surface determined in one section can be marked on the map or layer corresponding to that section of the range. A binary morphological filter can then be applied to each of these maps or layers.

[0013] The filtering operation can include dilation followed by erosion. Such a combined operation is also called closing. Dilation can involve extending the boundaries of the graphic object using a predefined mask. Conversely, erosion removes a predetermined edge from the object's boundaries using a predefined mask. A mask can encompass a geometric object such as a square or a circle. Masks for the two operations can be identical in size and shape or be different. Multiple closures can also be performed sequentially, using the same or different masks.

[0014] The closing function can smooth the inner corners of a graphic object, bridge small gaps, and, in particular, close internal holes. Closing can improve the correction of errors in flooring determination and provide a closed representation.

[0015] The floor surface can be scanned using a cleaning machine with a driven brush roller. The condition of the floor surface can be determined based on the current draw of the brush roller's drive. This allows the condition to be determined within a specific area covered by the brush roller, enabling faster scanning of the entire floor surface. Alternatively, a cleaning machine equipped with a dedicated sensor for detecting the floor surface condition can be used. Preferably, the sensor does not perform point-by-point measurements but rather measures a predetermined area within the cleaning machine's range.

[0016] The bristle roller has a predetermined width, which is determined transversely to the direction of travel of the processing machine. The width can be defined, in particular, along a rotational axis of the bristle roller. This axis of rotation extends perpendicular to the direction of travel and parallel to the substrate. The size of a filter mask of the binary morphological filter is selected depending on the width of the bristle roller. This allows the filter to reflect scanning conditions, enabling more accurate or realistic results.

[0017] It is also preferred that a circular filter mask be used, the radius of which lies between half and two widths of the bristle roller. Particularly preferably, the radius of the filter mask corresponds to the width of the bristle roller. Tests have shown that such a combination allows for very efficient and accurate determination of the floor covering boundaries.

[0018] In a further embodiment, secondary boundaries of the floor covering can be determined based on a systematic scanning of its outline. These boundaries can be combined to unite the strengths of both methods. For this purpose, sections where the floor surface's properties change can be defined along regularly staggered, antiparallel scans across the floor area. Secondary boundaries of the covering can be determined based on a multitude of such sections. These boundaries can then be defined based on an intersection of objects defined by the first and second boundaries. This intersection can correspond to a representation of the covering, and the boundaries of this intersection can be provided as the boundaries of the covering.

[0019] Particularly when the surface texture is determined by means of a bristle roller that includes a scanning surface extending transversely to the direction of movement of the processing machine, determining the second boundaries can be inaccurate if the scanning surface moves obliquely over an edge of the coating. In one embodiment, a midpoint between a right and a left boundary of the bristle roller can be defined as the section where the surface texture changes.

[0020] Preferably, a midpoint is determined between sections on two antiparallel paths where the texture changes in opposite directions. Based on a multitude of such midpoints, the secondary boundaries of the covering can then be determined. For this purpose, adjacent midpoints can be connected to each other in the manner of a polygon. If a regularly shaped edge of the covering is detected, for example, if the midpoints lie essentially on a straight line or on a convex curve of a predetermined curvature, then sections located at the edges of the curve where the texture changes can also be taken into account. In this way, many carpets with common shapes can be better represented. For example, a rectangular, a square, a round, an elliptical, or an oval carpet can be easily identified.

[0021] It is preferred that the first boundaries be dilated before the intersection between the objects defined by the first and second boundaries is formed. The dilation preferably follows a closure and can be performed using a previously used or a dedicated mask.

[0022] The defined boundaries, or an object enclosed by the boundaries, on the occupancy map can be used to control a cleaning machine for the floor surface. The machine can be programmed to avoid driving over the object. Depending on its specific characteristics, the object can be driven over, but only in a predetermined way, or not at all. For example, a combined floor cleaning machine that can both vacuum and mop a driven-over floor surface can be programmed to vacuum the specific object only, but not to mop it.

[0023] A graphical representation of the floor area is preferably provided, including an indication of specific flooring boundaries. The flooring map can be viewed as a raster graphic and presented to the user. The user can then check whether the flooring information is correct and complete, and correct or supplement it if necessary. Often, no changes to the information will be required, and the user can confirm its accuracy.

[0024] A surface treatment machine comprises a chassis for moving the machine across the surface; a device for determining whether a section of the surface being treated has a predetermined condition; a storage device for a surface coverage map; and a processing unit. The processing unit is configured to enter specific sections exhibiting the predetermined condition into the surface coverage map; to apply a binary morphological filter to the surface coverage map; and to determine the initial boundaries of a surface covering based on the surface coverage map.

[0025] The processing equipment may be configured to execute all or part of a method described herein. For this purpose, the processing equipment may be electronic and, for example, include a programmable microcomputer or microcontroller, and the method may be in the form of a computer program product with program code. The computer program product may also be stored on a computer-readable data carrier. Features or advantages of the method may be transferred to the equipment or vice versa.

[0026] It should be noted that various processing machines can be used for scanning and processing the floor surface. The processing of the scanned information in the manner proposed herein can be carried out by the first machine, the second machine, or partially by both.

[0027] The invention will now be described in more detail with reference to the accompanying figures, in which: Figure 1: A processing machine on a floor surface; Figure 2: A flowchart of a process; Figure 3: Unprocessed and processed floor surface occupancy maps; Figure 4: A schematic representation of a floor surface scan; Figure 5: A specific outline of a floor covering; and Figure 6: A section of specific boundaries of a floor covering represents.

[0028] Figure 1Figure 1 shows a cleaning machine 100 on a floor surface 105 in a household. The cleaning machine 100 is designed to move across the floor surface 105 and clean it, in particular by dry or wet cleaning. A covering 110, comprising a carpet, runner, or similar object that covers part of the floor surface 105, may be present on the floor surface 105. The covering 110 is not merely temporarily placed on the floor surface 105 and may form part of the household furnishings. The properties of the covering 110 differ from those of the floor surface 105.

[0029] The one in a lower section of Figure 1The processing machine 100, shown in more detail, comprises a processing unit 115, a chassis 120 (here exemplified by two wheels with electric motors), a storage device 125, and a processing unit 130. The processing unit 130 includes a suction unit 135 and a suction nozzle 140. The suction unit 135 is configured to generate an airflow through the suction nozzle 140 to extract dirt from the floor surface 105. The suction nozzle has a predetermined width and extends transversely to the usual direction of travel of the processing machine 100. An electrically driven brush roller 145 is provided in the area of ​​the suction nozzle 140 to process the floor surface 105. Preferably, a controllable mopping device 150 is provided, which can use liquid from a liquid tank 155 to clean a floor surface 105 that is being driven over.The wiping device 150 can comprise a cushion or fleece that can preferably be lowered onto or lifted from the floor surface 105. The supply of liquid can be controllable. In a further embodiment, a device for suctioning liquid from the floor surface 105 is provided, which can also be controllable. The widths of the suction nozzle 140, the bristle roller 145, and the wiping device 150 can correspond to each other in pairs. The processing device 115 is configured to control said components of the processing machine 100.

[0030] To optimally process the floor area 105, the processing machine 100 is to be controlled depending on the floor covering 110. For this purpose, it is first necessary to determine where the covering 110 lies on the floor area 105, or where the boundaries between the covering 110 and the floor area 105 lie. Determining these boundaries is also referred to as mapping. The processing machine 100 can be configured to detect objects such as furniture, people, or pets in the household in order to avoid collisions or to control cleaning. For this purpose, one or more dedicated or implicit sensors can be provided. For example, a camera or a LiDAR sensor can be used to scan the surroundings, or electrical currents from the electric motors of the chassis 120 or the brush roller 145 can be evaluated. The mapping described herein can be integrated with or performed separately from this detection.Findings can each be stored in maps or map layers of a common map in the storage device 125.

[0031] Figure 2 Figure 200 shows a flowchart of a process for mapping a soil area 105. In step 205, the soil area can be surveyed using the processing machine 100. This can involve processing the soil area 105 or it can be a reconnaissance run. The survey can be carried out systematically, for example in a meandering pattern, or irregularly, according to another system, or even randomly.

[0032] In step 210, the properties of the floor surface 105 can be determined. For this purpose, a dedicated sensor can be evaluated, which determines the properties, for example, based on the reflectance of infrared light. However, an electric current is also measured, which flows through a drive of the bristle roller 145. The current will vary depending on the length of a pile of the substrate.

[0033] In a first branch of procedure 200, which in Figure 2As shown on the left, for each section of the ground surface 105, it can be determined what condition the section has, or whether it has a predetermined condition or not. This distinction allows the section in question to be assigned either to the ground surface 105 or to a surface covering 110 laid on top of it. This information can be entered into an occupancy map, which comprises a matrix-like grid representing the ground surface 105. Each cell is assigned to a section of the driven-over subsoil. Cells carry binary values, where a first value indicates that the predetermined condition is present and a second value indicates that it is not. Cells of the occupancy map can be initialized with the second value. Optionally, a third value can also be provided, with which the cells are initialized and which indicates that the corresponding section has not yet been driven over.The scanning process can continue until a sufficient number of sections have been traversed and scanned. If the occupancy map data is to be evaluated, fields still containing the third value can be overwritten with the second value.

[0034] In step 220, a binary morphological filter can be applied to the occupancy map. The occupancy map can be understood as a binary graphical representation of the ground surface 105. Before the filter is applied, the occupancy map—assuming the existence of the ground surface 110 and a not excessive misidentification rate—comprises a first area in which fields predominantly bear the first value, and at least a second area in which fields predominantly bear the second value. The first area is an approximate representation of the ground surface 110, and the second area is the surrounding ground surface 105.

[0035] The filter can, in particular, implement a closure where small areas lying between groups of fields containing the first value are assigned the first value without allowing a structure formed by contiguous fields containing the first value to grow in all directions. The filter comprises a two-dimensional mask that can extend over multiple fields, the shape and size of which control properties of the closure.

[0036] The closure process can involve dilation, where the outline of a structure is extended in all directions by a predetermined amount, followed by erosion, the reverse operation where the outline is eroded in all directions by a predetermined amount. Combined, these processes can not only bridge small areas bearing the second value or close any resulting gaps, but also remove isolated fields with the first value located near the edge. This effect can be controlled by selecting the size ratios of the masks used in dilation and erosion. As a result of this processing, the occupancy map now includes an improved representation of the ground surface 105, on which an object representing the covering 110 is formed by fields with the first value.

[0037] In a second branch of procedure 200, which is in Figure 2As shown on the right, a different evaluation of the scans can be performed. This requires that the floor surface 105 has been traversed in a predetermined manner. The processing machine 100 travels on predetermined straight tracks that are parallel to each other and preferably uniformly spaced. More preferably, the processing machine 100 travels on adjacent tracks in opposite directions. If an electric current flowing through the drive motor of the bristle roller 145 changes during the traversal, a change in the condition of the subsoil in the traversed section of the floor surface 105 can be determined. The transition can be stored, for example, by writing a corresponding value to a field assigned to the section in a further occupancy map.

[0038] If sufficient information is available about sections of the floor area 105 where the nature of the subsoil changes, an outline of the covering 110 can be determined – again, assuming the existence of the covering 110 and a reasonable error rate. By connecting sections where the nature of the subsoil changes, an outline of the covering 110 can be determined. Variations of the procedure, in which points are formed based on the sections to further improve the representation of the outline, are described in more detail below.

[0039] In step 235, a determination of either the left or the right branch can be used to specify the boundaries of the covering 110 on the floor surface 105. Preferably, however, both approaches are combined. For this purpose, an intersection of objects can be formed, each representing the covering 110 on the floor surface 105. It is preferred that the object on the occupancy map is prepared after processing the left branch by binary morphological dilation using a predetermined mask before the intersection is formed. Boundaries of the object formed by the intersection can be considered as the boundaries of the covering 110.

[0040] If the specific characteristics of the covering 110 include, for example, a long-pile surface, it can be stipulated that an area of ​​the floor surface 105 covered by the covering 110 may be driven on and vacuumed, but not brushed or wet-cleaned, provided the processing machine 100 supports this. If the characteristics affect an even longer-pile area, driving on it may be permitted, but any kind of processing may be prohibited. If the characteristics affect an extremely long-pile area, driving on it may also be prohibited.

[0041] In step 240, the specific object can be visualized. Specifically, the object can be displayed or highlighted in a graphic representation that can be presented to a user of the processing machine 100 for review. The representation can depict the information the processing machine 100 has about the floor area 105 in the household and can, for example, indicate a hazard, an obstacle, the location of a base station, or a no-entry area. Instructions on how the processing machine 100 should be controlled depending on the area can be provided. The user can confirm or modify the information. In particular, the user can assign instructions for the processing machine 100 to a specific area of ​​the floor area 105, which can specifically control how close the machine 100 gets to the area, whether it crosses it, and if or how it is allowed to process it.

[0042] In step 245, the processing machine 100 can be controlled via the floor surface 105. Depending on available information, in particular regarding the boundaries of the floor covering 110 and associated instructions, processing of the surface being driven over can be controlled.

[0043] Figure 3 Figure 1 shows unprocessed and processed soil cover maps 300 of a soil area 105. From left to right, three columns depict an unprocessed soil cover map 300, a soil cover map 300 after binary morphological dilation, and a soil cover map 300 after subsequent erosion. The grid of the soil cover map 300 is not discernible in the representation and can be assumed to be correspondingly fine.

[0044] The first row assumes a relatively complete but incomplete representation of the coating 110. The gaps may be caused by misidentifications or obstacles, such as a table or chair leg. A predetermined first mask 305, which is round and relatively small, is used for dilation and erosion.

[0045] The second line assumes an incomplete and fragmentary representation of the surface 110. For example, a larger obstacle such as an armchair may be present in the upper area of ​​the object. A second mask 310, also used, is circular and slightly larger than the first mask 305.

[0046] The third line uses the same representation of the covering 110 as the second line, but uses an even larger and also circular third mask 315.

[0047] It can be seen how the depicted closures can each better approximate a form of the surface 110. A larger mask has a potentially stronger power to replicate inaccessible or blocked areas of the surface 110.

[0048] Figure 4 Figure 1 shows a schematic representation of a scanning of a floor surface 105. The processing machine 100 travels on a meandering path 405 on tracks 410, each preferably about as wide as the processing machine 100 or, more preferably, about as wide as the suction nozzle 140, the bristle roller 145, or the wiping device 150. The directions of travel of the processing machine 100 on adjacent tracks 410 are preferably opposite to each other.

[0049] If the processing machine 100 travels over a boundary of the covering 110 on the floor surface 105, this can be detected, for example, because an electric current flowing through the drive motor of the bristle roller 145 changes in a predetermined way. The current can increase when approaching the long-pile covering 110 and decrease when retracting. In this way, sections or points 415 can be determined where a boundary of the covering 110 is suspected. The point 415 is preferably determined in the middle of a traversed path 410 to account for the possibility that the suction nozzle 140 could cross the boundary at an angle. The points 415 can be connected to form an outline 425 of the covering 110.

[0050] Preferably, the points 415 are further processed before the outline 425 is formed by creating center points 420 between adjacent points 415 where the surface changes in opposite directions due to the different directions of travel. This allows for consideration of the fact that the detection of a transition often occurs with a predetermined delay after the actual crossing of the boundary. Endpoints of a sequence of adjacent points 415, where the sequence extends along a predetermined curve, can be considered in addition to the center points 420. A curve can, in particular, comprise a straight line or a curve with a predetermined curvature. In another embodiment, points 415 lying in maximally separated paths 410 can be considered.Furthermore, points that are at their maximum distance from each other along a lane 410 or with respect to all lanes 410 can be taken into account. However, remaining points 415 need not be considered further if the selected points 415, 420 are joined together in the manner of a polygon to form an outline 425 of the surface 110.

[0051] Figure 5 shows a specific outline 425 of a floor covering 110 after Figure 4 based on center points 420 and selected points 415. The outline 425 determined in this way can be used to crop an object that has been created by applying a binary morphological filter to an occupancy map 300 of certain properties of the ground surface 105 as described herein.

[0052] Figure 6This shows such a cut or the formation of an intersection. As an example, it is assumed that an obstacle 605 is located on a section of the surface 110, preventing a scan of its properties. Initial boundaries 610 are determined based on binary morphological filtering. Advantageously, the position of the obstacle 605 is precisely determined; however, the edges of the surface 110 are not very precisely defined.

[0053] Second boundaries 615 correspond to an outline 425 formed on the basis of points 415 or centers 420. The second boundaries generally run more smoothly along the edges of the surface 110. However, the specifications do not indicate where an obstacle 605 lies on the surface 110. Since no transition between the ground surface 105 and the surface could be determined in this area, numerous obstacles 605 of varying sizes and shapes are conceivable that could extend across this area. Accordingly, the extent of a drivable section of the surface 110 cannot be reliably determined based on the second boundaries 615.

[0054] By forming an intersection of a first object, defined by the first boundaries 610, with a second object, defined by the second boundaries 615, the ability of the first approach to close open areas can be combined with the ability of the second approach to accurately define an outline based on transitions.

[0055] It is even more preferred that the first boundaries 610 are dilated before forming the intersection using a predetermined filter, resulting in first boundaries 610'. A formed intersection is in Figure 6 The area is shown hatched. The boundaries of this intersection can be defined as the boundaries of the pavement 110 or as the boundaries of a drivable section of the pavement 110. Reference sign

[0056] 100 Processing machine 105 Floor area 110 Covering 115 Processing device 120 Chassis 125 Storage device 130 Processing device 135 Suction unit 140 Suction nozzle 145 Bristle roller 150 Wiping device 155 Liquid tank 200 Procedure 205 Scan floor area 210 Determine condition 215 Mark sections of predetermined condition 220 Apply binary morphological filter 225 Mark sections where the condition changes 230 Determine outline 235 Cut object sets 240 Visualize object 245 Control floor cleaner 300 Occupancy card 305 First mask 310 Second mask 315 Third mask 405Path 410Track 415Point 420Center 425Outline 605Obstacle 610First boundaries 610'Dilated first boundaries 615Second boundaries

Claims

1. Method (200) for mapping a floor area (105) in a household, wherein the method (200) comprises the following steps: - scanning (210) the floor area (105) and determining sections of the floor area (105), on which a predetermined first quality of the floor area (105) is present, wherein the floor area (105) is scanned by means of a treatment machine (100) with a drivable bristle roller (145) for cleaning the floor area (105); wherein the quality of the floor area (105) is determined on the basis of a current consumption of a drive of the bristle roller (145), by an electrical current being determined, which flows through the drive of the bristle roller (145) and varies as a function of the length of a pile of a subsurface; - entering (215) the determined sections into an occupancy map (300) of the floor area (105); - applying (220) a binary morphological filter to the occupancy map (300), wherein the bristle roller (145) has a predetermined width; and - determining (235) first limits of a covering (110) lying on the floor area (105), on the basis of the occupancy map (300); - choosing a size of a filter mask (305-315) of the binary morphological filter as a function of the width of the bristle roller (145), so that the filter reflects circumstances of the scanning.

2. Method (200) according to claim 1, wherein the filter operation comprises a dilation, followed by an erosion.

3. Method (200) according to claim 1 or 2, wherein a circular filter mask (305-315) is used, the radius of which lies in a range from half the width to twice the width of the bristle roller (145).

4. Method (200) according to claim 3, wherein the radius of the filter mask (305-315) corresponds to the width of the bristle roller (145).

5. Method (200) according to one of the preceding claims, wherein sections are determined on regularly offset, antiparallel paths (410) over the floor area (105) in each case, on which sections the quality of the floor area (105) changes; wherein second limits of the covering (110) are determined on the basis of a large number of such sections; and wherein an overlap of objects is determined, which are determined by the first and the second limits.

6. Method (200) according to claim 5, wherein a centre point (420) is determined between sections on two antiparallel paths (410), on which the quality changes in an opposing manner; and wherein the second limits of the covering (110) are determined on the basis of a large number of such centre points (420).

7. Method (200) according to claim 5 or 6, wherein a dilation of the first limits takes place before the overlap is formed.

8. Method (200) according to one of the preceding claims, wherein a treatment machine (100) for cleaning the floor area (105) is controlled as a function of determined limits of the covering (110).

9. Method (200) according to one of the preceding claims, wherein a graphic representation of the floor area (105) is provided (240), which comprises a note regarding determined limits of the covering (110).

10. Treatment machine (100) for a floor area (105), wherein the treatment machine (100) comprises the following: - a chassis (120) for driving the treatment machine (100) over the floor area (105); - a facility (130) for determining whether a travelled section of the floor area (105) has a predetermined quality, wherein the floor area (105) is scanned by means of a treatment machine (100) with a drivable bristle roller (145) for cleaning the floor area (105); wherein the quality of the floor area (105) is determined on the basis of a current consumption of a drive of the bristle roller (145); - a storage facility (125) for an occupancy map (300) of the floor area (105); - a processing facility (115), which is configured to enter particular sections, on which the predetermined quality is present, into the occupancy map (300); to apply a binary morphological filter to the occupancy map (300), wherein the bristle roller (145) has a predetermined width; and to determine first limits of a covering (110) lying on the floor area (105) on the basis of an occupancy map (300); - choosing a size of a filter mask (305-315) of the binary morphological filter as a function of the width of the bristle roller (145), so that the filter reflects circumstances of the scanning.