Device and method for predictively estimating a friction coefficient network of a road surface, and means of transportation

Abstract

The invention relates to a device (100) for predictively determining a friction coefficient network (10) of a surface (20), the device having the following features: a capturing device (30) for capturing at least one group of pixels (31), which depict the surface (20) at least in part, in order to determine a first network of first values from the at least one group of pixels (31) in a region of the captured surface (20), wherein the first network is superimposed on the region; and a determination unit (40) for determining friction coefficient values which form the friction coefficient network (10) on the basis of the first values of the first network, wherein the friction coefficient network (10) represents a distribution of the determined friction coefficient values. The invention also relates to a means of transport (1000) having such a device, and to a method for predictively determining a friction coefficient network (10) of a surface.

Description

[0001] Device and method for predictively estimating a friction coefficient network of a road surface, and means of transport

[0002] A device and a method for predictively estimating a friction coefficient network of a road surface are described. The present invention also relates to a means of transport equipped with such a device.

[0003] When a vehicle brakes or accelerates, a corresponding force is transmitted to the road via the tires. Fig. 1 shows, for example, a braking process.

[0004] The braking force applied by the vehicle (indicated by a tire) does not directly affect the vehicle's motion, but rather opposes the wheel rotation. This braking force is then transferred to the road via the tires in the form of static or kinetic friction. The maximum transmissible frictional force depends on the normal force (i.e., the vehicle's weight and the road gradient) and the coefficient of friction between the tires and the road. If the braking force exceeds the maximum transmissible frictional force, the tires will slip (or spin) when the wheel rotation no longer corresponds to the vehicle's motion.

[0005] Figure 2 illustrates the curve of the resulting frictional force, normalized to the normal force, when attempting to move a tire across the road surface without rotation under various road conditions such as dry, wet, snowy, or icy. Initially, the tire is stationary and resists movement due to static friction. As the applied force increases, the frictional force rises until it reaches a maximum value—the maximum of each curve shown in Figure 2. This point is called the limit of adhesion—the moment when the applied force exceeds the maximum static friction and the motion transitions from static to sliding. After this transition to sliding, the measured frictional force decreases significantly. This means that less force can be transferred to the vehicle's motion once the tire begins to slip.The greatest effect on vehicle motion during acceleration or braking is achieved when the tire does not transition into sliding motion and static friction persists. Several approaches exist for estimating the coefficient of friction in real time. A general overview of coefficient of friction estimation is given in reference [1]. An approach using a forward-looking camera is described in reference [2].

[0006] FH251202PCT-2025396128.DOCX ex US 2017 0 550 252 A1 and US 2018 0 108 792 describe a forward-looking camera connected to a local sensor for referencing. DE 2020 10203293 A1 describes a system suitable for determining weather-related road conditions.

[0007] One object of the present invention is to provide a device and a method, as well as a means of transport, with which the braking or acceleration force can be limited in such a way as to prevent the tires of the means of transport from slipping and to ensure the maximum transmission of force to the road surface.

[0008] This problem is solved by the subject matter of the attached independent claims.

[0009] The proposed device for predictively determining a friction coefficient network of a surface comprises a detection device for capturing at least one group of pixels that at least partially represent the surface in order to determine a first network of initial values ​​from the at least one group of pixels in a region of the captured surface, wherein the first network is superimposed on the region. The device also comprises a determination unit for determining friction coefficient values ​​that form the friction coefficient network based on the initial values ​​of the first network, wherein the friction coefficient network represents a distribution of the determined friction coefficient values. The surface can, for example, be the roadway of a car, an aircraft, or a railway. The surface can also, for example, be the wings of an aircraft.In general, a surface can be any area whose coefficient of friction is of interest for an action. The determining unit can evaluate the detected group of pixels with respect to information about the surface state of each pixel and determine the current coefficient of friction pertaining to the surface. The determining unit can evaluate the pixels, in particular, with respect to the current state of water (H₂O). The determining unit could be an artificial intelligence, especially a neural network.

[0010] FH251202PCT-2025396128.DOCX ex According to a further aspect of the present invention, a method for predictively determining a friction coefficient network of a surface is proposed. The method comprises acquiring at least one group of pixels to determine a region of a surface from the at least one group of pixels and to determine a first network of first values ​​within that region. Furthermore, the method comprises determining a region of the surface based on information contained in the at least one group of pixels and subsequently determining friction coefficient values ​​in a friction coefficient network corresponding to the region, based on the first values ​​of the first network, in order to obtain a distribution of the determined friction coefficient values.

[0011] Another aspect of the present invention relates to a means of locomotion with the device described above, wherein the means of locomotion includes an additional sensor. This sensor is configured to provide the target unit with information regarding at least one reference value and / or motion data of the means of locomotion. Various approaches can be taken to provide the target unit with in-situ reference values, in particular:

[0012] Contact-based approaches utilize the interaction between the vehicle and the road surface. The measuring unit is designed to evaluate the forces transmitted to the vehicle's wheels, such as the torque resulting from acceleration or braking. Additionally, tire slip can be determined, either by measuring motion data or using sensors such as acoustic sensors, gyroscopes, or tire pressure monitoring systems (TPMS) that can detect tire slippage. By analyzing the relationship between the applied forces and the detected tire slip, the actual limit of adhesion can be approximated.

[0013] The determining unit can also be designed to execute procedures that operate independently of the interaction with the means of transport. Such procedures can estimate the coefficient of friction based on information about the road surface properties. This involves the use of sensors that detect surface properties by measuring their interaction with light or other electromagnetic waves. Examples of such sensors include radar, lidar, and cameras.

[0014] The determination unit is designed to use the reference information provided by the additional sensor to determine the, in particular the actual, adhesion limit.

[0015] FH251202PCT-2025396128.DOCX to calculate the friction coefficient between the tires and the road surface using the normalized coefficient of friction network and the reference value.

[0016] Furthermore, the control unit is designed to influence the vehicle's steering based on the determined coefficient of friction. If the coefficient of friction indicates a risk of loss of traction and that the vehicle could begin to slide during the initiated action, appropriate adjustments can be made to the drive or braking systems to ensure the vehicle's stability and safety.

[0017] Another aspect of the present invention relates to a computer program for carrying out the method described herein when the computer program runs on a computer.

[0018] The technical teaching proposed herein allows the coefficients of friction to be predictively determined and localized, i.e., assigned to a section of roadway that has been photographed.

[0019] If the coefficients of friction of the road surface are known, ideal braking can be planned in advance. This means that wheel slippage can be reduced or prevented by always limiting the braking force to the maximum frictional force. Preferably, the braking or acceleration force can be limited to the maximum frictional force to make the vehicle safer. The resulting braking performance is significantly higher than achievable with ABS. The proposed device could potentially replace ABS altogether. The system also functions during acceleration: power can be automatically limited to prevent wheel spin. Any slippage not only reduces efficiency but also increases tire wear, which impacts costs and environmental pollution. Therefore, the proposed technology can also reduce the vehicle's operating costs.

[0020] Preferred embodiments of the present invention are explained in detail below with reference to the accompanying drawings. These show:

[0021] FH251202PCT-2025396128.DOCX ex Fig. 1 a schematic braking process of a means of transport;

[0022] Fig. 2 shows the course of the resulting frictional force, normalized to the normal force, during continuous tire slippage under different road surface conditions;

[0023] Fig. 3 shows a false-color representation of in-situ determined coefficient of friction values ​​on a roadway;

[0024] Fig. 4 shows a schematic representation of the device described herein;

[0025] Fig. 5 shows a further schematic representation of the device described herein;

[0026] Fig. 6 shows a further schematic representation of the device described herein;

[0027] Fig. 7 shows a further schematic representation of the device described herein;

[0028] Fig. 8a, b shows a schematic representation of a means of transport with the device described herein; and

[0029] Fig. 9 shows a flowchart of the proposed procedure.

[0030] Individual aspects of the invention described herein are described below in Figures 1 to 9. In the present application, identical reference numerals refer to identical or equivalent elements, and it is not necessary for all reference numerals to be repeated in all drawings.

[0031] When a component is described herein as "designed to do something", it means that the component has been designed structurally and physically in such a way that the component performs what it is supposed to do.

[0032] Figures 1 and 2 were already discussed in the introductory section. Figures 4 to 7 each show a proposed device 100, which is described below with reference to Figures 4 to 7. The device 100 for predictively determining a friction coefficient network 10 of a surface 20 has a detection device 30 for detecting at least one group of pixels 31, which define the surface 20 (see Figure 1).

[0033] FH251202PCT-2025396128.DOCX ex Figs. 1, 3 or 8) at least partially reproduce in order to determine a first network of first values ​​from the at least one group of pixels 31 in a region of the detected surface 20, wherein the first network is superimposed on the region. The detection device 30 comprises at least one sensor 35, preferably several sensors 25, to detect different aspects regarding a state of matter of water on the surface 20, particularly in situ, i.e., in real time. The proposed device 100 further comprises a determination unit 40 for determining coefficient of friction values, which form the coefficient of friction network 10 based on the first values ​​of the first network, wherein the coefficient of friction network 10 represents a distribution of the determined coefficient of friction values. In Figures 4 to 7, the determining unit 40 is represented as a neural network (CNN).Alternatively, the determination unit can also be implemented as a method or algorithm designed to execute a defined sequence of operations. The determination unit 40 can be implemented in various configurations. In particular, it can be implemented as a processor, a field-programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). Figures 4 and 5, for example, show a friction coefficient network 10, which can be determined by the device 100 based on the detected pixels 31.

[0034] The determination unit 40 is configured to superimpose the distribution of the determined coefficient of friction values ​​onto the area of ​​the measured surface 20 and to determine a one-dimensional or multi-dimensional distribution. The one-dimensional or multi-dimensional distribution forms, for example, the coefficient of friction network 10, which is shown in Figures 4 and 5 as Figure 70. The multi-dimensional distribution is preferably a two-dimensional to n-dimensional distribution, in which each value is included in the calculation of the coefficient of friction network. In particular, the distribution can be given in the form of a matrix with dimensions N, M, C, and T, where: a) N represents a horizontal distribution of pixels, b) M represents a vertical distribution of pixels, c) C represents a number of different channels, such as RGB, NIR, 970 nm, polarization 0°, polarization 90°, etc., d) T represents different time points.

[0035] FH251202PCT-2025396128.DOCX ex Where N, M, C and T are natural numbers and independent of each other and can be used individually or in any combination as input for the algorithm to calculate the friction coefficient network.

[0036] The friction coefficient grid 10 is calculated using the acquired pixel matrix and is defined as a normalized grid with numerical values ​​from 0 (lowest friction) to 1 (highest friction). When referenced to at least one in-situ acquired absolute value or similar, the calculated adhesion limits of the areas represented in the pixels are obtained. The matrix can be used directly for superimposing the front view of the surface 20 (see Fig. 3). In Fig. 3, approximately 0 (low adhesion, bright red), 1 (sufficient adhesion, green), and a mix in between are represented. Typically, the matrix is ​​transmitted to a control system 160 of a means of transport 1000, in particular the vehicle's central computer, which can use the data for further actions such as warning the driver by displaying a warning on a screen or even directly intervening in the vehicle's driving behavior.

[0037] Figure 3 shows a superposition of the calculated friction coefficient network 10 and the surface 20. Figure 3 depicts various areas, each corresponding to a specific friction coefficient. The device 100 determines significantly reduced friction coefficients for the water-filled ruts based on absorption and polarization information. The device 100 can be used to calculate friction coefficients of the surface 20 based on the acquired polarization and / or absorption values. Furthermore, this is not merely a point measurement or an average value of the surface condition, but an imaging method that can be used to locate areas with significantly reduced friction coefficients, thus enabling dynamic responses to slippery areas on the roadway.

[0038] The at least one group of pixels 31 preferably comprises a plurality of groups of pixels 31 which can be grouped with regard to their evaluation in respect of the coefficients of friction.

[0039] The at least one group of pixels 31 includes at least one piece of information about the light intensity 52 reflected from the surface 20. The information about the light intensity can then be used for further evaluation by means of the determination unit 40 (see Fig. 6). In particular, the at least one piece of information includes information about

[0040] FH251202PCT-2025396128.DOCX ex a spectral distribution of the reflected light, such as an absorption rate 53, in particular defined as the ratio of the light intensities in two different wavelength ranges, and / or polarization information, such as a polarization rate 51, in particular defined as the ratio of the light intensities in two different polarization planes, of the reflected light and / or information on an intensity 52 (the light intensity) of the reflected light. By evaluating this information, for example, a conclusion can be drawn about a surface roughness 61 and / or the presence of water 62 on the surface 20. Finally, estimations 63 of the coefficients of friction can be made from this, which ultimately form the coefficient of friction network 10 (see Fig. 7).

[0041] The detection device 30 has a sensor 35 configured in one of the following embodiments: a. as a partial sensor designed to detect p-polarized light; or b. as a), but with an additional partial sensor designed to detect s-polarized light; or c. as b), but with an additional partial sensor designed to detect light in another polarization plane, in particular a 45° plane that corresponds neither to p- nor s-polarization; or d. as a), but with an additional partial sensor that does not have a preferred polarization plane.

[0042] Figures 4, 5, 6, and 7 each show 35° polarizing filters with degree markings. In technical terms, it is more common to speak of p-polarized (p for parallel, i.e., 0° polarization) and s-polarized (s for perpendicular, i.e., 90° polarization) light. Here, p-polarized light is the more important, and s-polarized light can alternatively be replaced by the unfiltered image. With a 45° polarizing filter, the angle bisector, which is used in addition to the other two axes, indicates a rotation of the polar axes. Using a 45° polarizing filter provides further information.

[0043] FH251202PCT-2025396128.DOCX The topography and the material influence the polarization behavior. This is used to estimate the surface roughness using the acquired polarization information. Since water and its states of matter can significantly alter the coefficients of friction as well as the optical behavior of the surface 20, the parameters surface roughness 61 and the presence of water 62 are determined (see Fig. 6).

[0044] Alternatively or additionally, the detection device 30 has a sensor 35 configured in one of the following embodiments: a. as a partial sensor designed to detect a wavelength range of an absorption maximum of water or its states of matter. In particular, the detected wavelength range includes

[0045] 980 nm, 1190 nm and / or 1450 nm; or b. as a), but with an additional sub-sensor designed to detect a wavelength range outside the absorption maxima of water; or c. as a), but with an additional sub-sensor without a restricted wavelength range.

[0046] The proposed sensor 35 can in particular be designed such that an aspect of the current state of matter of the water on the surface 20 can be detected in order to be able to draw a conclusion about the current coefficient of friction of the surface 20.

[0047] Figures 4 to 7 show that the detection device 30 is coupled to the determination unit 40. The determination unit 40 is configured to determine the area of ​​the surface 20 from the at least one group of pixels 31 by image recognition. Preferably, several groups of pixels 31 can be detected, each of which represents a different area of ​​the surface 20 by means of image recognition. The different groups of pixels 31 can be combined to form an image of pixels 31 to represent the area of ​​the surface 20. The area of ​​the surface is an area that has been detected by the sensor 35, in particular with all its sub-sensors.

[0048] Optionally, the device can include a signal processing device 45 between the sensing device 30 and the destination unit 40 to perform preprocessing of the signals. In Figures 4 and 5, the optional signal processing device 45 is indicated by dashed lines. By preprocessing the signals, the destination unit 40, which would receive the preprocessed signals from the signal processing device 45, can process them faster than if the acquired signals from the sensing device 30 were directly transmitted to the destination unit 40. The signal processing device 45 can be implemented as an algorithm on a processor unit or on specialized electronics such as an ASIC or FPGA.

[0049] Figures 4 to 7 show that the proposed device uses 100 polarization filters Pol O° (p-polarization), Pol 90° (s-polarization) and a bandpass BP 970 nm, and that the “intensity” required as a brightness reference is calculated from the mean of Pol 0° and Pol 90°.

[0050] If the signal processing device 45 is present, it is configured to perform preprocessing of the data, in particular the preprocessing comprising: a. normalizing sensor information with the intensity (52), and / or b. calculating a degree of polarization by comparing p-polarized and s-polarized light, and / or c. calculating absorption behavior by comparing different spectra of the reflected light.

[0051] In this context, the terms data and signal are used synonymously. The term intensity refers to light intensity.

[0052] The detection unit 40 or the signal processing device 45 has an interface for receiving at least one group of pixels 31 from an external sensor 50 or for receiving known data, in particular from a database such as a cloud or similar. The external sensor 50 is therefore configured to communicate with the detection unit 40 or with the signal processing device 45. Furthermore, the detection unit 40 or the signal processing device-

[0053] FH251202PCT-2025396128.DOCX ex tung 45 each designed to communicate with the external sensor 50, in particular to query currently recorded data or known data relating to a state of matter of the water on the surface 20.

[0054] The determination unit 40 is configured to divide the determined area into segments 61, 62, 63, 64, each consisting of one or more pixels, in order to determine a coefficient of friction value for each segment 61, 62, 63, 64. Figures 4 and 5, for example, graphically represent the segments 61, 62, 63, 64, each of which represents a single coefficient of friction value. The size of the segments 61, 62, 63, 64 can correspond to the size of the respective group of pixels 31.

[0055] Fig. 4 shows, for example, an embodiment of the device 100 based on camera inputs for capturing at least one group of pixels 31. The neural network, in particular a Convolutional Neural Network (CNN) block, can be considered a black box that learns the best method for calculating or estimating the friction coefficients from training data.

[0056] As previously described, the process can alternatively be supported by data preparation by placing the signal processing device 45 upstream of the determination unit 40. This allows the polarization rate 51, intensity 52, and absorption rate 53 to be calculated directly, thus reducing the input parameters. The primary reason these parameters are particularly well-suited for classifying the surface condition and calculating or estimating the coefficient of friction is that polarization is especially effective for determining surface roughness, and absorption, particularly at 970 nm or other absorption peaks of water, is well-suited for detecting water in all states of matter.

[0057] The signal processing device 45 does not calculate the coefficients of friction, but merely preprocesses the acquired signals. This involves, for example, calculating the polarization rate 51 or absorption rate 53 from the intensities of the various sensors 35a, 35b, 35c, 35d, and then passing these on to the detection unit 40 as a combined group of pixels. If no signal processing device 45 is present, the detection unit performs all these tasks, as shown in Fig. 6. Preferably, an acquired group of at least one pixel 31 is combined into a segment 61, 62, 63, 64.

[0058] FH251202PCT-2025396128.DOCX ex The determination unit 40 is designed to determine adhesion limits 101 (see Figs. 2 and 3) from the coefficient of friction values, which are assigned to different positions in the area in order to determine an adhesion limit network in the area, where the adhesion limit 101 indicates a maximum transmissible frictional force. In Fig. 2, for example, the adhesion limits are given by the maxima of the curves shown in Fig. 2. In Fig. 3, the different adhesion limits 101 are shown as an adhesion limit network superimposed on the surface 20. Fig. 3 shows the adhesion limit network in a false-color representation.

[0059] The determination unit 40 is configured to store reference values ​​depending on external environmental conditions and / or the acting standard force in order to determine the adhesion limits 101 upon re-acquisition of at least one group of pixels 31. Alternatively or additionally, the determination unit 40 is configured to receive data from an in-situ measurement in order to determine the coefficients of friction and / or adhesion limits 10 in the area in-situ and to store the in-situ determined values. The in-situ determined values ​​can be stored to be subsequently retrieved as reference values. In particular, in-situ determined data and the values ​​derived from them can be stored by the determination unit 40 to serve as a reference value for a future event.

[0060] An in-situ measurement can, for example, be contact-based, laser-based, LiDAR-based, radar-based, or include measurement data stored and queried in a cloud. For this purpose, 100 different sensors or measuring devices can be in communication with the device. In particular, a controller associated with the device can initiate an in-situ measurement.

[0061] The signal processing device 45 and the determination unit 40 are designed separately from each other, or the determination unit 40 includes the signal processing device 45, or, as already noted, there is no signal processing device 45 and the determination unit 40 takes over the functions of the signal processing device 45. Depending on the workload of the determination unit 40, the signal processing device 45 may be useful or unnecessary.

[0062] FH251202PCT-2025396128.DOCX ex In any case, the determining unit 40 for determining coefficients of friction is designed to perform a mathematical calculation or to perform the determination based on artificial intelligence, in particular a convolutional neural network (CNN). In particular, the determining unit 40 is a processor or specialized hardware, often called AI accelerators, such as GPUs, TPUs, ASICs or FPGAs.

[0063] The determination unit 40 is configured to calculate, as a first value, a polarization rate 51 and / or a light intensity 52 and / or an absorption rate 53 for a point in the area, in order to determine the coefficient of friction for that point. Here, a point position is defined by a segment 61, 62, 63, 64. In other words, a single segment 61, 62, 63, 64 defines which data from which group of pixels 31 are used to calculate the coefficient of friction in that segment.

[0064] The determination unit 40 is configured to receive data acquired by the sub-sensors 35a, 35b, 35c, 35d at an output that is not coupled to any filter and / or bandpass filter and / or to one of the filters and / or to one of the bandpass filters, in order to calculate the polarization rate 51 and / or the light intensity 52 and / or the absorption rate 53 (see Fig. 6). Polarization data and / or absorption ranges and / or specific wavelength ranges, which are of interest with regard to one of the states of matter of water, can be acquired using the sub-sensors 35a, 35b, 35c, 35d.

[0065] A bandpass filter reflects or absorbs light outside its specified wavelength range, also called the band. In particular, light from approximately 400 to 950 nm should be blocked for this application, as light in this wavelength range would significantly obscure the relevant information at 970 to 980 nm. The bandpass filter allows light to pass through within a band with a fixed full width at half maximum (FWHM) and a fixed central wavelength (CWL). For example, a bandpass filter with a CWL of 980 nm and an FWHM of 50 nm would allow light between 955 nm and 1005 nm (980 nm ± 25 nm => 50 nm width) to pass through. The bandpass filter used here is designed to allow only light at the known absorption peak of water at approximately 980 nm to pass through. There are further absorption peaks of water at approximately 1190 nm and 1450 nm. Corresponding

[0066] FH251202PCT-2025396128.DOCX ex bandpass filters are designed analogously to those just described. Bandpass filters that allow light to pass through near the absorption peaks of water are preferred.

[0067] The detection unit 40 is configured to compare at least one group of pixels 31 with a reference group of pixels. The reference group of pixels can be stored in the detection unit 40. The reference group of pixels can consist of a previously acquired at least one group of pixels 31 or of several groups of pixels 31 that were previously acquired and stored in the detection unit 40 as a reference group of pixels. The source of the reference group can be different. The source of the reference group could be an absorption with a different wavelength, a polarization in a different plane, or an unfiltered image (group of pixels).

[0068] For example, the signal processing device 45 or the determination unit 40 may be configured to apply an average of one of the previously determined quantities or similar aggregating procedures to the respective segments 61, 62, 63, 64 of the groups of pixels 31 in order to obtain a further value of interest associated with the corresponding segment 61, 62, 63, 64.

[0069] For example, the device is configured, after learning how to determine the adhesion values, to determine the adhesion values ​​in-situ based on the stored reference values ​​after re-acquiring the at least one group of pixels 31, or after re-acquiring the at least one group of pixels 31 and the at least one group of pixels 31 acquired for the second time. It is possible for the at least one group of pixels 31 to be acquired multiple times, in particular two, three... n times, where n is a natural number, in succession, in order to determine and derive the quantities mentioned herein. The at least one group of pixels 31 can have the same number of pixels at the same location on the surface, or the same number of pixels at different locations on the surface, or a different number of pixels at the same location on the surface 20.In other words, the resolution of the at least one group of pixels 31 can be constant or vary at the same location on the surface 20, or the resolution of the at least one group of pixels 31 is constant, but the location on the surface 20 being detected is changed.

[0070] FH251202PCT-2025396128.DOCX ex The at least one group of pixels 31 and a second time acquired at least one group of pixels 31 are preferably identical with respect to the position or the number and position of the pixels, however, the first time the at least one group of pixels 31 is acquired with no filter or different filters and / or bandpass filters or no bandpass filter. In other words, the first time the at least one group of pixels 31 is acquired with different sub-sensors 35a, 35b, 35c, 35d of the sensor 25 than the second time the at least one group of pixels 31 is acquired.

[0071] The determination unit 40 is configured to divide the coefficient of friction network 10 into different sub-areas, each sub-area being assigned a single specific coefficient of friction, and in particular, the device is configured to visually represent the different sub-areas in a false-color representation. Fig. 3, for example, shows a false-color representation in which the different determined adhesion limits 101 of the surface 20, a road surface, are superimposed. In Fig. 3, it can be seen that at a normalized value of 1, a maximum adhesion limit 101 is reached by an object moving on the surface 20, while in areas where the normalized adhesion limit is 0.1, the object moving on the surface 20 is at risk of transitioning into a sliding motion, since the adhesion limit 101 is not at its maximum.

[0072] Another aspect of the present invention relates to a means of locomotion 1000 with the device 100 described herein, wherein the means of locomotion 1000 comprises an additional sensor 150 configured to provide the measuring unit 40 with information regarding at least one reference value and / or information regarding a movement of the means of locomotion 1000. A means of locomotion 1000 is shown in Figures 8a and 8b. The measuring unit 40 is configured to determine the coefficients of friction of the surface 20. When the means of locomotion 1000 accelerates or decelerates, it enters a sliding motion, i.e., it begins to slip, if the force used for braking or acceleration is higher than the adhesion limit 101.Therefore, the determining unit 40 is designed to influence a control 160 of the means of locomotion 1000 if the determining unit 40 determines adhesion limits 101 based on the calculated coefficients of friction that are below the applied braking or acceleration force, which would mean a transition of the means of locomotion 1000 into a sliding motion. If the means of locomotion 1000 appears to be transitioning into a sliding motion, the control 160 is

[0073] FH251202PCT-2025396128.DOCX, for example, is designed to reduce the speed of the vehicle 100 less drastically. This is because reducing speed requires braking, which means increasing the braking force. If the force used for braking is already at the limit of adhesion, further increasing the braking force would inevitably lead to skidding. The proposed device 100 is designed to prevent precisely this. Therefore, the control unit 160, through communication with the device 100, reduces the braking force on the vehicle 100 when it appears to be about to slide. Less forceful braking reduces the braking distance by preventing the vehicle from sliding. The control unit thus limits the braking / acceleration force to prevent skidding.The control unit 160 is preferably in communication contact with the device 100, such that the device 100 can communicate with the control unit and vice versa. The communication between the device 100 and the control unit 160 can be bidirectional. As can be seen in Figures 8a and 8b, the means of locomotion 1000 has a field of view 200, which is continuously scanned by the device 100 or at predetermined time intervals in order to allow the adhesion limits 101 to be determined by the determination unit 40.

[0074] The determination unit 40 can communicate with a cloud or server via GPS to request current weather data. The determination unit 40 can be configured to store reference values ​​depending on external environmental conditions and / or the acting normal force of a means of transport 1000 in order to determine the adhesion limits upon re-acquisition of at least one group of pixels, or upon re-acquisition of at least one group of pixels and the second re-acquisition of at least one group of pixels. The normal force acting on the surface 20 depends on the weight of the means of transport 1000 and its load, and on the gradient of the surface 20, which can be, in particular, a road surface.

[0075] The vehicle 1000 has an interface designed to alert a driver of the vehicle 100 if the vehicle 1000 appears to be entering a skidding motion. The interface may include an LED indicator that illuminates when the vehicle is about to enter a skidding motion. The interface may include a display indicating

[0076] FH251202PCT-2025396128.DOCX ex, which, for example, presents a false color display to the driver of the means of transport 1000, as shown in Fig. 3.

[0077] The interface is therefore designed to visually display the friction coefficient network 10 with its various segments 61, 62, 63, 64 to the driver as an image 70 in a false-color representation. The friction coefficient network is created based on visible and predominantly invisible wavelengths, which are converted into a false-color image in the visible wavelength range for the driver. The false-color representation means that the colors do not represent light / wavelengths / colors, but rather other information, in particular the adhesion limits 101 and / or friction values ​​of the friction coefficient network 10.

[0078] Another aspect of the present invention relates to a method 900 for predictively determining a coefficient of friction network 10 of a surface 20. In step 910, the method 900 comprises acquiring at least one group of pixels 31 in order to determine a region of a surface 20 from the at least one group of pixels 31 and to determine a first network of first values ​​within that region from the at least one group of pixels 31. Furthermore, in step 920, the method 900 comprises determining a region of the surface 20 based on information contained in the at least one group of pixels 31, and in step 930, determining coefficient of friction values ​​in a coefficient of friction network 10 corresponding to the region, based on the first values ​​of the first network, in order to obtain a distribution of the determined coefficient of friction values. Steps 910, 920, and 930 are performed in this order. Fig.Figure 9 shows a flowchart of the proposed procedure 900.

[0079] Method 900 can also be used to calibrate Method 900, in which case Method 900 includes obtaining data from an in-situ measurement to determine adhesion limits 101 in the area from the previously determined coefficient of friction values, and storing the determined adhesion limits 101 as reference values. For example, adhesion limits 101 determined in-situ can be stored as reference values ​​with respect to the environmental conditions from which the adhesion limits 101 were determined in the determination unit 40, in particular in the neural network. The determination unit 40 can be configured to receive data from an in-situ measurement in order to determine the coefficient of friction values ​​in-situ.

[0080] FH251202PCT-2025396128.DOCX to determine the adhesion limits 101 ex and / or in-situ in the area, wherein the determination unit 40 is configured to store the in-situ determined coefficient of friction values ​​and / or the adhesion limits 101. In-situ determined adhesion limits 101 can be stored in the neural network as a reference value with respect to the environmental conditions on the basis of which they were determined. This allows the device 100 to be calibrated.

[0081] Another aspect of the present invention relates to a computer program for carrying out the method described herein when the computer program runs on a computer.

[0082] In the acquisition of the at least one group of pixels 31 described herein, cameras with absorption and polarization filters can be used that can detect the specific absorption behavior of water and the roughness of the surface in order to distinguish between the states dry, wet, water, ice, and black ice. As can be seen in Fig. 2, knowledge of the surface state or road condition is fundamental for estimating the coefficient of friction. Instead of using the proposed device only for classifying the surface state or road condition, it is proposed here to use a neural network to directly determine or estimate a coefficient of friction for each point on the road surface in advance.

[0083] If a local or in-situ measurement is available for an area that was previously optically detected, the measurement can be used as an absolute reference. Common local measurements typically require slippage due to acceleration or similar dynamic motion to estimate the coefficient of friction. Therefore, these measurements are event-driven and not available at all times. In the proposed device and method, the local or in-situ measurements serve only as a correction value for the always available predictive friction coefficient calculation or estimation. It is possible to operate the device 100 completely without in-situ measurements for correction / calibration.

[0084] With the proposed device 100 and the proposed method 900, the friction coefficients can be predictively determined and localized.

[0085] FH251202PCT-2025396128.DOCX ex If the surface or road surface properties are known, ideal braking can be planned in advance. This means that wheel slippage can be completely prevented if the braking force is always limited to the maximum frictional force. The resulting braking performance is significantly higher than achievable with ABS. It might even be possible to completely replace ABS. The system also works when starting off: power can be automatically limited to prevent wheel spin.

[0086] Any slippage not only reduces efficiency but also increases tire wear, which affects costs and environmental impact.

[0087] Other filters besides those described here can be used to achieve similar results. It is reasonable to assume that there will be an unlimited number of machine learning models capable of producing similar results.

[0088] The proposed device 100 and the proposed method 900 can be used, for example, in the automotive, long-distance transport, or aviation sectors. The increased efficiency and improved braking performance in hazardous situations make the device 100 and the method 900 a useful tool for any means of transport 1000.

[0089] Autonomous vehicles will be able to use Device 100 and Procedure 900 particularly effectively, as these plan all actions in advance on an internal map. The additional information can further improve action planning.

[0090] Although some aspects have been described in connection with a device or a process, it is understood that these aspects also represent a description of a corresponding process or device, so that a block or component of a device or system is also to be understood as a corresponding process step or as a feature of a process step and differently.

[0091] FH251202PCT-2025396128.DOCX ex around. A complete description of the present invention in the form of process steps or in the form of device features is omitted here for reasons of redundancy.

[0092] In the preceding detailed description, various features were sometimes grouped together in examples to streamline the disclosure. This type of disclosure should not be interpreted as indicating that the claimed examples have more features than are expressly stated in each claim. Rather, as the following claims reflect, the subject matter may consist of fewer than all the features of a single disclosed example. Consequently, the following claims are hereby incorporated into the detailed description, with each claim potentially representing a separate, independent example.While each claim can stand as a separate example, it should be noted that, although dependent claims refer back to a specific combination with one or more other claims, other examples also include a combination of dependent claims with the subject matter of any other dependent claim, or a combination of any feature with other dependent or independent claims. Such combinations are included unless it is stated that a specific combination is not intended. Furthermore, it is intended that a combination of features of a claim with any other independent claim is also included, even if that claim is not directly dependent on the independent claim.

[0093] FH251202PCT-2025396128.DOCX ex

Claims

Patent claims 1. Device (100) for predictively determining a coefficient of friction network (10) of a surface (20), comprising: a detection device (30) for detecting at least one group of pixels (31) which at least partially represent the surface (20) in order to determine a first network of first values ​​from the at least one group of pixels (31) in an area of ​​the detected surface (20), wherein the first network is superimposed on the area, and a determination unit (40) for determining coefficient of friction values ​​which form the coefficient of friction network (10) based on the first values ​​of the first network, wherein the coefficient of friction network (10) represents a distribution of the determined coefficient of friction values.

2. Device (100) according to claim 1 wherein the determination unit (40) is configured to superimpose the distribution of the determined coefficient of friction values ​​onto the area of ​​the detected surface (20) and to determine a one-dimensional or multi-dimensional distribution.

3. Device (100) according to one of claims 1 to 2, wherein the at least one group of pixels (31) comprises at least one piece of information about a light intensity (52) reflected from the surface (20), in particular: a. information about a spectral distribution of the reflected light, and / or b. polarization information of the reflected light and / or c. information about an intensity (52) of the reflected light.

4. Device (100) according to any one of claims 1 to 3, wherein the detection device (30) comprises a sensor (35) configured in one of the following embodiments: a. as a partial sensor designed to detect p-polarized light; or b. as a), but with an additional partial sensor designed to detect s-polarized light; or FH251202PCT-2025396128.DOCX c. as b), but with an additional sub-sensor designed to detect light in a further polarization plane that is neither p- nor s-polarization; or d. as a), but with an additional sub-sensor that does not have a preferred polarization plane.

5. Device (100) according to any one of claims 1 to 4, wherein the detection device (30) comprises a sensor (35) configured in one of the following embodiments: a. as a partial sensor designed to detect a wavelength range of an absorption maximum of water or its states of matter; or b. as a), but with an additional partial sensor designed to detect a wavelength range outside the absorption maxima of water; or c. as a), but with an additional partial sensor without a restricted wavelength range 6. Device (100) according to one of claims 1 to 5, wherein the detection device (30) is coupled to the determination unit (40), and wherein the determination unit (40) is configured to determine the area of ​​the surface (20) from the at least one group of pixels (31) by image recognition.

7. Device (100) according to one of claims 1 to 6, wherein a signal processing device (45) is provided between the detection device (30) and the determination unit (40) to perform preprocessing of the signals.

8. Device (100) according to claim 7, wherein the signal processing device (45) is configured to perform preprocessing of the data, in particular wherein the preprocessing comprises: a. normalizing sensor information with the intensity (52), and / or b. calculating a degree of polarization by comparing p-polarized and s-polarized light, and / or c. calculating absorption behavior by comparing different spectra of the reflected light. FH251202PCT-2025396128. DOCX 9. Device (100) according to any one of claims 1 to 8, wherein the determination unit (40) or the signal processing device (45) has an interface for receiving the at least one group of pixels (31) from an external sensor (50) or known data.

10. Device (100) according to one of claims 1 to 9, wherein the determining unit (40) is configured to divide the determined area into segments (61, 62, 63, 64) consisting of one or more pixels in order to determine a coefficient of friction value for each segment (61, 62, 63, 64).

11. Device (100) according to one of claims 1 to 10, wherein the determining unit (40) is configured to determine adhesion limit values ​​(101) from the coefficient of friction values, which are assigned to different positions in the area in order to determine an adhesion limit value network in the area, wherein the adhesion limit value (101) indicates a maximum transmissible frictional force.

12. Device (100) according to any one of claims 1 to 11, wherein the determination unit (40) is configured to: a. store reference values ​​depending on external environmental conditions and / or the acting normal force in order to determine the adhesion limits (101) upon re-acquisition of the at least one group of pixels (31); b. and / or receive data from an in-situ measurement in order to determine in-situ the coefficient of friction values ​​and / or adhesion limits (101) in the area in order to store the values ​​determined in-situ.

13. Device (100) according to claim 11, wherein an in-situ measurement comprises one or more of the following: a. contact-based measurement and / or b. a laser-based measurement and / or c. a LIDAR-based measurement and / or d. a RADAR-based measurement and / or e. measurement data stored and queried in a cloud. FH251202PCT-2025396128. DOCX 14. Device (100) according to any one of claims 7 to 13, wherein the signal processing device (45) and the determination unit (40) are separate from each other, or wherein the determination unit (40) comprises the signal processing device (45).

15. Device (100) according to one of the preceding claims, wherein the determination unit (40) for determining coefficient of friction values ​​is configured to perform a mathematical calculation or to perform the determination on the basis of artificial intelligence.

16. Device (100) according to one of the preceding claims, wherein the determining unit (40) is configured to calculate a polarization rate (51) and / or a light intensity (52) and / or an absorption rate (53) as a first value for a point in the area in order to determine the coefficient of friction for that point.

17. Device (100) according to claim 4 or 5, wherein the determination unit (40) is configured to receive data acquired by the partial sensors at an output which is not coupled to any filter and / or bandpass and / or to one of the filters and / or to one of the bandpasses, in order to calculate the polarization rate (51) and / or the light intensity (52) and / or the absorption rate (53).

18. Device (100) according to any one of the preceding claims 1 to 17, wherein the determining unit (40) is configured to compare the at least one group of pixels (31) with a reference group of pixels, 19. Device (100) according to one of the preceding claims, wherein the at least one group of pixels (31) and a second time detected at least one group of pixels (31) are identical with respect to the position or the number and position of the pixels, but are detected with no filter or different filters and / or bandpass filters or no bandpass filter.

20. Device (100) according to one of the preceding claims, wherein the determining unit (40) is configured to divide the coefficient of friction network (10) into different sub-areas, each sub-area being assigned to a single FH251202PCT-2025396128. DOCX is assigned to specific coefficients of friction, in particular wherein the device is designed to visually represent the different sub-areas in a false color representation.

21. Means of locomotion (1000) with the device (100) according to one of the preceding claims, wherein the means of locomotion (1000) comprises an additional sensor (150) configured to provide the determining unit (40) with information regarding at least one reference value and / or regarding a movement of the means of locomotion (1000).

22. Means of transport (1000) according to claim 21, which has an interface configured to alert a driver of the means of transport (1000) when the means of transport (1000) appears to be entering a sliding motion.

23. Means of transport (1000) according to claim 22, wherein the interface is configured to visually display to the driver the coefficient of friction network (10) with its various segments (61 , 62, 63, 64) as an image (70) in a false color representation.

24. Means of transport (1000) according to one of claims 21 to 23, wherein the determining unit (40) is configured to influence the control (160) of the means of transport (1000) in such a way as to limit a braking force or an acceleration force of the means of transport (1000) in such a way that the in-situ determined coefficient of friction values ​​do not exceed a limit of adhesion (101) in order to prevent the means of transport (1000) from slipping and to ensure a maximum transmission of force to a roadway of the means of transport (1000).

25. Method for predictively determining a friction coefficient network (10) of a surface (20), wherein the method comprises: Capturing at least one group of pixels (31) in order to determine from the at least one group of pixels (31) an area of ​​a surface (20) and in order to determine in the area from the at least one group of pixels (31) a first network of first values, FH251202PCT-2025396128. DOCX Determining an area of ​​the surface (20) based on information contained in the at least one group of pixels (31), and Determining coefficient of friction values ​​in a coefficient of friction network (10) corresponding to the area, based on the first values ​​of the first network, in order to obtain a distribution of the determined coefficient of friction values.

26. Method for calibrating the method according to claim 25, wherein the method comprises: obtaining data from an in-situ measurement to determine adhesion limits (1001) in the area from the previously determined coefficient of friction values, and storing the determined adhesion limits (101) as reference values.

27. Computer program for carrying out the method according to claim 25 or 26, if the computer program runs on a computer. FH251202PCT-2025396128.DOCX

Images