CLEANING SYSTEM WITH A SELF-DRIVING CLEANING ROBOT AND A CHARGING STATION AND METHOD FOR STARTING THE CLEANING ROBOT AT THE CHARGING STATION

Description

[0001] The present invention relates to a cleaning system with a self-propelled cleaning robot and a charging station according to the preamble of claim 1. The invention also relates to a method for driving a self-propelled cleaning robot to a charging station of such a cleaning system.

[0002] Known cleaning systems usually have a self-driving, i.e. autonomous, cleaning robot, for example a vacuum robot, and a charging station into which the cleaning robot temporarily drives or docks to charge its battery.

[0003] Such cleaning systems are known, for example, from the documents US 2014 / 0100693 A1 and US 2016 / 0318415 A1.

[0004] US 2016 / 0091899 A1 discloses a generic cleaning system comprising a self-propelled cleaning robot and a charging station for charging a battery of the cleaning robot. The charging station has visibly arranged reference markings, and the cleaning robot has an optical detection device via which it can detect these reference markings. The cleaning robot also has a computer device that is communicatively connected to the detection device and a control device for controlling a driving process of the cleaning robot, for example a drive, and that is capable of receiving an image of two reference markings provided by the optical detection device and transmitting it to the control device orto control a drive system of the cleaning robot such that it moves to the charging station based on a difference between the representation of two reference markings in the received image and a predetermined relative distance between the first and second reference markings. The charging station is thus controlled via a camera-based detection device for detecting two reference markings.

[0005] EP 3 185 096 A1 discloses a charging station for an autonomous cleaning device, comprising a station body with a power supply interface. The power supply interface is arranged on a side surface of the station body to supply the autonomous cleaning device with power. Furthermore, an identification structure is provided, which is also arranged on the side surface of the station body and which enables a position of the power supply interface to be detected by a detection device on the autonomous cleaning device and to control the autonomous cleaning device accordingly. The identification structure comprises at least one protrusion and at least one depression, which form a surface structure that, when light is incident, enables a different intensity of reflected light and thus enables identification.This also makes it possible to control the autonomous cleaning device, for example a self-driving cleaning robot, to the charging station.

[0006] US Pat. No. 8,749,196 B2 discloses a combination of an autonomous robot, a charging station, and a navigation control system for controlling the movement and operation of the robot. A signal emitter for projecting a signal to the robot is provided at the charging station. The latter has at least one signal detector designed to detect at least one infrared docking signal projected by the charging station's emitter, using spatial mapping programs using the charging station and / or walls and large objects as reference points.

[0007] The disadvantage of the cleaning systems or charging stations known from the state of the art is that they are not only comparatively large and technically complex and therefore expensive, but also that it is difficult for a cleaning robot to find the charging station if there is no line of sight or signal connection between the cleaning robot and the charging station.

[0008] The present invention therefore addresses the problem of providing an improved or at least an alternative embodiment for a cleaning system of the generic type, which in particular overcomes the disadvantages known from the prior art and in particular enables an improved locating of a charging station by an autonomous cleaning robot.

[0009] This problem is solved according to the invention by the subject matter of independent claim 1. Advantageous embodiments are the subject matter of the dependent claims.

[0010] The present invention is based on the general idea of ​​providing a cleaning system with a charging station and an autonomous, i.e., self-driving, cleaning robot that uses two different methods to automatically locate the charging station: a so-called HOG method (histogram of oriented gradients) and a reference marker method (fiducial marker). The cleaning system according to the invention thus combines these two methods for the first time, resulting in the significant advantage that the charging station itself can be constructed more compactly and less conspicuously, and the cleaning robot can find the charging station even when it is so far away from it, for example, in another room of an apartment to be cleaned, that there is no direct visual contact with the charging station and thus no direct signal connection.The cleaning system according to the invention thus comprises the previously described self-propelled cleaning robot and the charging station for charging a battery of the cleaning robot, wherein the charging station has a visibly arranged reference marking and the cleaning robot has an optical detection device via which it can detect the reference marking on the charging station side as well as an environment. Furthermore, the cleaning robot according to the invention has a computer device, for example a microprocessor, which is communicatively connected to the detection device and a control device for controlling a driving process of the cleaning robot. The control device preferably also comprises a drive device via which the cleaning robot is driven. Alternatively, it is of course also conceivable for the control device to be merely communicatively connected to the drive device.According to the invention, the computer device is designed such that it controls a process of the cleaning robot approaching the charging station using the previously described HOG method (histogram of oriented gradients), provided that the distance between the cleaning robot and the charging station is greater than a predefined distance or the detection device of the cleaning robot does not detect the reference marking. A feature classifier is created using the HOG method, which is designed to detect objects in image processing and thereby generates a so-called region of interest (ROI) containing the reference marking. This is the case, for example, if the cleaning robot is located so far away from its charging station that its optical detection device cannot clearly detect the reference marking on the charging station side.This situation is conceivable for distances greater than 3 meters between the charging station and the cleaning robot, since in this case, detecting the reference marking is difficult for the optical detection device on the cleaning robot. Furthermore, the computer device according to the invention is designed such that it controls the approach of the cleaning robot to the charging station via the reference marking, provided that the detection device detects the reference marking and the distance between the cleaning robot and the charging station is less than or equal to the predefined distance, and the detection device of the cleaning robot detects the reference marking.If the cleaning robot is located in close proximity to the charging station, it can be approached exclusively using the reference marking method. Whereas if the distance between the charging station and the cleaning robot is greater, the cleaning system according to the invention initially uses the HOG method until the cleaning robot has moved close enough to the charging station to allow further navigation using the reference marking on the charging station. The cleaning system according to the invention thus combines, for the first time, two different navigation methods for approaching a cleaning robot to an associated charging station, thereby enabling a smaller and less complex charging station. The HOG method also makes it possible to implement a self-learning effect, which enables the charging station to be found at different angles, in different lighting conditions, and at different distances.According to the invention, the computer device calculates the distance d, coordinates and an angle α of the cleaning robot to a center point of the reference marking, provided that the detection device detects the reference marking.

[0011] The reference marking expediently has at least two images and / or shapes. The two images or shapes can be arranged at a predefined distance from one another and form edges, lines, or points that are clearly identifiable by the optical detection device of the cleaning robot and can be converted by the computer device into corresponding control commands to the control device or a drive device of the cleaning robot. The reference marking or its images and / or shapes are designed such that the fixed points or lines required for navigation can be clearly and easily detected by the detection device. The reference marks with their individual points are then overlaid with an image captured by the detection device and used as a unit of measurement or reference point.

[0012] The reference marker expediently comprises three rectangles arranged next to one another, the middle rectangle of which has a different color and / or shape than the two side rectangles. In particular, it is conceivable for the middle rectangle to be 7.5 cm high and 8 cm wide, while the two side rectangles are squares with sides of 7.5 cm each. This means that a total of eight fixed points on the reference marker, namely the respective corners of the square rectangles, can be used for navigation. To make it easier to find or record these reference markers, the middle rectangle can also be white, while the two side rectangles are red. This highlights the respective corners of the side rectangles in color.

[0013] In a further advantageous embodiment of the solution according to the invention, the predefined distance d is approximately 3 meters. During the development of the invention, it has been shown that a maximum distance for reliable detection of the reference marking or for reliable navigation of the cleaning robot to the charging station is approximately 3 meters. Within this distance d, it is easily possible for the optical detection device, for example a USB camera, to detect the individual fixed points or lines / edges of the reference marking and to approach the charging station solely via these. At a distance d of greater than 3 meters, difficulties may arise under certain circumstances, so that in this case the HOG method according to the invention is used in the present invention.Using the HOG method, a feature classifier can be created that is trained to detect objects in image processing and thereby generates a so-called region of interest (ROI) containing the reference marker. The biggest challenge in creating such a feature classifier is that it must cope with variations in the lighting, position, and occlusion of the charging station in the image. To achieve this, the classifier must be trained, for example, with different orientations, backgrounds, lighting situations, and different occlusions of the charging station.

[0014] The present invention is further based on the general idea of ​​specifying a method for approaching a self-propelled cleaning robot to a charging station of a cleaning system described in the previous paragraphs, in which the computer device controls an approach process of the cleaning robot to the charging station via a histogram of oriented gradients (HOG) if a distance of the cleaning robot to the charging station is greater than a predefined distance d or the detection device of the cleaning robot does not detect the reference marking, by creating a feature classifier via the HOG method, which is designed to recognize objects in image processing and thereby generates a so-called region of interest (ROI) containing the reference marking.If the detection device detects the reference marking and the distance between the cleaning robot and the charging station is less than or equal to the predefined distance d, and the detection device of the cleaning robot detects the reference marking using the optical detection device, the approach process of the cleaning robot to the charging station is controlled via the reference marking in the detection device method. The method according to the invention thus uses different navigation methods depending on the detection and distance, namely the reference marking method and the HOG method. Using the method according to the invention, it is thus possible to specify a precise and reliable method for locating the charging station, which is also cost-effective and enables a space-optimized and cost-effective design of the charging station.

[0015] According to the invention, the computer device calculates the distance d, coordinates, and an angle α of the cleaning robot to a center point of the reference marking, provided the detection device detects the reference marking. This can be done, for example, by overlaying the image captured by the detection device with the reference marking contained therein. Using the distance d, the coordinates, the angle α, and the resulting calculated center point of the reference marking, the cleaning robot can be precisely guided to the charging station via the computer device and the control device communicating with it.

[0016] For practical purposes, the computer system steers the cleaning robot directly toward the charging station at an angle α ≤ 10 degrees, since this range represents an optimum in terms of d between the cleaning robot and the charging station. A controller, e.g., a proportional controller, ensures that any deviations from the center of the reference mark determined in the previous paragraph are corrected during approach to the charging station. This reacts to any inaccuracies in the robot's position estimation resulting from measurement errors in the odometry sensors. This enables precise control of the cleaning robot to the charging station.

[0017] For angles α greater than 10 degrees, the following further advantageous embodiment of the method according to the invention has been found for locating the charging station: First, at an angle of α ≥ 10 degrees, the computer device initiates a counterclockwise rotation of the cleaning robot by an angle β = (90 degrees - α) and then moves it a distance d 1 = dx cos(β). Once the cleaning robot has moved the distance d 1, the computer device initiates a further rotation of the cleaning robot by an angle of 90 degrees clockwise via the control device and then moves it a distance d 2 = dx sin(β). After traveling the distance d 2, the cleaning robot has generally reached a more advantageous position with regard to the charging station, so that this can now be found more easily using the methods described above.

[0018] In a further advantageous embodiment of the method according to the invention, the computer device moves the cleaning robot back if its distance d from the charging station is less than 100 cm and, at the same time, the angle α is greater than 10 degrees. If the distance d is less than 100 cm and the angle α is less than 10 degrees, the cleaning robot can move directly to the charging station, whereas if the angle α is > 10 degrees, it first rotates counterclockwise, moves along the distance d 1 , and then rotates clockwise and moves along the distance d 2 according to the previous paragraph. All in all, the method according to the invention can achieve a comparatively simple yet reliable way of locating the charging station.

[0019] Further important features and advantages of the invention emerge from the subclaims, from the drawings and from the associated description of the figures based on the drawings.

[0020] It is understood that the features mentioned above and those to be explained below can be used not only in the combination specified in each case, but also in other combinations or on their own, without departing from the scope of the present invention.

[0021] Preferred embodiments of the invention are illustrated in the drawings and are explained in more detail in the following description, wherein the same reference numerals refer to the same or similar or functionally identical components.

[0022] They show, schematically Fig. 1 shows a cleaning system according to the invention, Fig. 2 shows a possible embodiment of a detection device with reference marking, Fig. 3 shows a flow chart for carrying out a method according to the invention for approaching a self-propelled cleaning robot to a charging station.

[0023] According to the Figure 1 , a cleaning system 1 according to the invention comprises a self-propelled, i.e. autonomous, cleaning robot 2 and a charging station 3. In the charging station 3, a battery of the cleaning robot 2 can be charged, provided it docks there and makes electrical contact. The charging station 3 has a visibly arranged reference marking 4 (see also Figure 2), while the cleaning robot 2 also has an optical detection device 5, for example a camera. Using the detection device 5, the cleaning robot 2 can detect both the reference marking 4 and the surroundings of the cleaning robot 2, for example to avoid obstacles. In addition to the detection device 5, the cleaning robot 2 also has a computing device 6, for example a computer or a microprocessor, which is communicatively connected to the detection device 5 and a control device 7 for controlling a driving process of the cleaning robot 2. Using the control device 7, control commands are transmitted, for example, to a drive device via which the cleaning robot 2 is driven.The detection device, the computer device 6, is now designed such that it carries out a process of moving the cleaning robot 2 to the charging station 3 in a distance-dependent and detection-dependent manner using a HOG method (histogram of oriented gradients) and a reference marking method. According to the invention, the computer device 6 thus controls the process of moving the cleaning robot 2 to the charging station 3 using the HOG method if a distance d of the cleaning robot 2 to the charging station 3 is greater than a predefined distance, for example, greater than 3 meters, or if the detection device 5 of the cleaning robot 2 does not recognize or detect the reference marking 4 on the charging station 3.If a distance of the cleaning robot 2 to the charging station 3 is less than or equal to the predefined distance, for example less than or equal to 3 meters, and the detection device 5 of the cleaning robot 2 detects the reference marking 4, the computer device 6 controls the start-up process of the cleaning robot 2 to the charging station 3 via the reference marking 4. The cleaning system 1 according to the invention thus uses a method combined from an HOG method and a reference marking method to control the start-up process.

[0024] The reference mark 4 can, for example, represent two images 8, 8' (compare Figure 2 ) and / or two molds 9, 9' which are arranged at a distance from each other. In the case of the Figure 2The reference marking 4 shown has three rectangles 10a, 10b and 10c arranged next to each other, of which the middle rectangle 10b has a different color and a different shape than the two side rectangles 10a and 10c. In the specific case of the Figure 2 The middle rectangle 10b is 7.5 cm high and 8 cm wide, while the two side rectangles 10a and 10c are each squares with sides of 7.5 cm. Furthermore, in the illustrated case, the middle rectangle 10b is white, while the two side rectangles 10a and 10c are red. Figure 2 The reference marking 4 shown thus has four corner points and four edges on each of the two lateral rectangles 10a and 10c, which enable a clear assignment in a reference marking process.

[0025] Such a method according to the invention for driving the self-propelled cleaning robot 2 to the charging station 3 is, for example, according to the Figure 3shown and described below.

[0026] The method according to the invention is started, whereupon the detection device 5 searches for the reference marking 4. If it does not detect it, for example because a distance d to the reference marking 4, i.e., between the cleaning robot 2 and the charging station 3, is too large, the charging station 3 is detected using the HOG method. If the charging station 3 is detected using the HOG method, a region of interest (ROI) is determined and traversed, after which another attempt is made to detect the reference marking 4 using the optical detection device 5. If the charging station 3 cannot be detected using the HOG method, the cleaning robot 2 rotates back and forth by plus / minus 120 degrees and another attempt is made to detect the reference marking 4 using the detection device 5.

[0027] The last-mentioned process steps can occur in particular when the cleaning robot 2 is located at a large distance from the charging station 3, for example greater than 3 meters.

[0028] If the reference marking 4 is detected by the detection device 5 of the cleaning robot 2, the computer device 6 controls the approach process of the cleaning robot 2 to the charging station 3 via the reference marking 4. This is the case if, for example, a distance of the cleaning robot 2 to the charging station 3 is less than or equal to the predefined distance or the detection device 5 detects the reference marking 4.

[0029] After detecting the reference marking 4, the computer device 6 calculates the distance d, coordinates of the cleaning robot 2 and an angle α of the cleaning robot 2 to a center point M of the reference marking 4. If the angle α is ≤ 10 degrees, the control device 7 controls the cleaning robot 2 directly towards the charging station 3 using a controller, e.g. a proportional controller, in that the (proportional) controller moves the cleaning robot 2 such that the center point M of the reference marking represents a center point of a predetermined frame of a camera image. If the center, or the center point M of the reference marking 4, is not in the center of the frame of the image or image frame detected by the detection device 5, the (proportional) controller controls the center of the reference marking 4 to the center of the frame by correcting the direction of movement of the cleaning robot 2.

[0030] If the angle α calculated by the computer device 6 and detected by the detection device 5 is above 10 degrees, the cleaning robot is first rotated counterclockwise by an angle β = (90 degrees - α) and then moved by a distance d 1 = dx cos(β). This corresponds to point B in Figure 1 The cleaning robot 2 is then rotated by an angle of 90 degrees via the computer device 5, the control device 7, and a drive device, and then moved by a distance d 2 = dx sin(β). Afterward, the cleaning robot 2 should be able to be docked to the charging station 3. If this is not the case, the cleaning robot 2 would have to be realigned or moved again.

[0031] If, for example, the angle α is > 10 degrees and at the same time a distance d is less than 100 cm to the charging station 3, the cleaning robot 2 first moves back in order to be able to reliably carry out the previously mentioned process steps and thus a start-up process.

[0032] With the method according to the invention and the cleaning system 1 according to the invention, it is possible for the first time to achieve precise and reliable approach of the cleaning robot 2 to the charging station 3 by combining an HOG method and a reference marking method, whereby the charging station 3 can be constructed more compactly and cost-effectively overall. Furthermore, reliable approach can be ensured even with greater distances or when the reference marking 4 is not directly detected by the detection device 5. It is of course clear that the HOG method increases a self-learning effect by specifying or playing back different images of the charging station 3 with different backgrounds and lighting situations or positions, thus enabling the cleaning robot 2 to find the charging station 3 more effectively. List of reference symbols

[0033] 1Cleaning system 2Cleaning robot 3Charging station 4Reference marking 5Optical detection device 6Computer device 7Control device 8Image 9Shape 10Rectangle

Claims

1. Cleaning system (1) having a self-driving cleaning robot (2) and a charging station (3) for charging a rechargeable battery of the cleaning robot (2), wherein the charging station (3) has a visibly arranged reference marker (4) and the cleaning robot (2) has an optical detecting facility (5), by way of which it detects the reference marker (4) and an environment, wherein the cleaning robot (2) has a computing facility (6), which is connected in a communicating manner to the detecting facility (5) and a control facility (7) so as to control a moving process of the cleaning robot (2), wherein the computing facility (6) is designed in such a manner - that said computing facility controls a moving process of the cleaning robot (2) towards the charging station (3) by way of a histogram of oriented gradients (HOG) if a distance d between the cleaning robot (2) and the charging station (3) is greater than a pre-defined distance or the detecting facility (5) does not detect the reference marker (4), in that, by way of the HOG method, a feature classifier is created which is designed so as to recognise objects in the image processing and as a result generates a so-called region of interest (ROI) that comprises the reference marker, - that said computing facility controls a moving process of the cleaning robot (2) towards the charging station (3) by way of the reference marker (4) if the detecting facility (5) recognises the reference marker (4) and if a distance d between the cleaning robot (2) and the charging station (3) is less than or equal to the pre-defined distance, and the detecting facility (5) detects the reference marker (4), characterised in that the computing facility (6) calculates a distance d, coordinates and an angle α of the cleaning robot (2) with respect to a middle point (M) of the reference marker (4) if the detecting facility (5) detects the reference marker (4).

2. Cleaning system according to claim 1, characterised in that the reference marker (4) has at least two images (8, 8') and / or shapes (9,9').

3. Cleaning system according to claim 2, characterised in that the reference marker (4) has three rectangles (10a, 10b, 10c) which are arranged adjacent to one another and of which a middle rectangle (10b) has a different colour and / or shape than the two lateral rectangles (10a, 10c).

4. Cleaning system according to claim 3, characterised in that the middle rectangle (10b) is 7.5 cm high and 8 cm wide, whereas the two lateral rectangles (10a, 10c) are squares with sides 7.5 cm long and / or that the middle rectangle (10b) is white, whereas the two lateral rectangles (10a, 10c) are red.

5. Cleaning system according to one of the preceding claims, characterised in that the pre-defined distance is 3 m.

6. Method for moving a self-driving cleaning robot (2) to a charging station (3) of a cleaning system (1) according to one of the preceding claims, in which the computing facility (6) - controls a moving process of the cleaning robot (2) towards the charging station (3) by way of a histogram of oriented gradients (HOG) if a distance d between the cleaning robot (2) and the charging station (3) is greater than a pre-defined distance or the detecting facility (5) of the cleaning robot (2) does not detect the reference marker (4), in that, by way of the HOG method, a feature classifier is created which is designed so as to recognise objects in the image processing and as a result generates a so-called region of interest (ROI) that comprises the reference marker, - controls a moving process of the cleaning robot (2) towards the charging station (3) by way of the reference marker (4) if the detecting facility (5) recognises the reference marker (4) and if a distance d between the cleaning robot (2) and the charging station (3) is less than or equal to the pre-defined distance and the detecting facility (5) of the cleaning robot (2) detects the reference marker (4), characterised in that the computing facility (6) calculates a distance d, coordinates and an angle α of the cleaning robot (2) with respect to a middle point (M) of the reference marker (4) if the detecting facility (5) detects the reference marker (4).

7. Method according to claim 6, characterised in that in the case of an angle α ≤ 10° the computing facility (6) steers the cleaning robot (2) directly towards the charging station (3) by means of a proportional regulator, in that the proportional regulator moves the cleaning robot (2) in such a manner that the middle point (M) of the reference marker (4) illustrates a middle point of a specified frame of the detecting facility (5).

8. Method according to claim 6 or 7, characterised in that in the case of an angle α > 10° the computing facility (6) - rotates the cleaning robot (2) in an anti-clockwise direction by an angle β = (90° - α) and subsequently moves said cleaning robot over a distance d1 = d · cos(β). - rotates the cleaning robot in a clockwise direction by an angle of 90° and subsequently moves said cleaning robot over a distance d2 = d · sin(β).

9. Method according to one of claims 6 to 8, characterised in that in the case of an angle α > 10° and a distance d < 100 cm the computing facility (6) moves the cleaning robot (2) back towards the charging station (3).