Method for generating a resulting friction coefficient distribution of the maximum friction coefficient of a vehicle

Abstract

Method for generating a resulting friction coefficient distribution of the maximum friction coefficient of a vehicle at a current and / or future waypoint on a road with a given road surface, the method comprising the steps: - Providing all known conditional probability distributions that estimate a coefficient of friction under the condition of a specific road surface condition, - Providing all possible current road weather information in relation to the current and / or future waypoint of the road as a probability vector for the road surface condition, wherein the possible current road weather information in relation to the current and / or future waypoint of the road is generated by a dedicated vehicle sensor system, and wherein a road weather model is used to generate the recorded current road weather information as a probability vector for the road surface condition, - Generating the resulting friction coefficient distribution of the friction coefficient at a current and / or future waypoint by determining the total probability distribution based on the conditional probability distributions in conjunction with the probability vector.

Description

[0001] The invention relates to a method for generating a resulting friction coefficient distribution of the maximum friction coefficient of a vehicle at a current and / or future waypoint of a road, as well as a driver assistance system.

[0002] To increase the safety of a motor vehicle, precise knowledge of the road surface condition and the resulting friction potential, i.e., the maximum coefficient of friction between tires and road surface, is essential. This applies particularly to existing driver assistance systems (e.g., friction-adaptive emergency braking or adaptive cruise control) as well as to future autonomous vehicles.

[0003] The maximum coefficient of friction determines how strongly a motor vehicle can be accelerated or decelerated and at what lateral acceleration the motor vehicle will break away sideways.

[0004] The coefficient of friction is also required for controlling vehicle dynamics control systems and driver assistance systems. With precise knowledge of the coefficient of friction, anti-lock braking systems, electronic stability systems, and traction control systems can be controlled with exceptional accuracy. Known methods for determining the coefficient of friction between the vehicle and the road surface are based on an estimation that considers either the lateral or longitudinal dynamics of the vehicle.

[0005] The coefficient of friction depends on a variety of friction parameters. In particular, the road surface condition, i.e., whether the road surface is dry, wet, snowy, or icy, is an important influencing factor.

[0006] Sudden changes in friction coefficients, such as those caused by altered environmental conditions, can lead to unstable driving situations and thus increase the risk of accidents. It becomes particularly dangerous if the driver approaches a curve too fast due to a misjudgment of the available friction coefficient.

[0007] When a vehicle is operating semi- or fully autonomously, assessing the current coefficient of friction of a road surface section is more difficult because the information from the human driver is at least partially lost.

[0008] WO 2016 / 120 092 A1 discloses a method for operating a database-supported friction coefficient map, in which information transmitted by sending vehicles is received and stored in the database, wherein the information includes at least friction coefficient data describing the specific friction coefficient potential of a road segment, location data describing the geometric position of this road segment, and time data describing the time of determination of the friction coefficient data, and the data stored in the database can be retrieved by receiving vehicles.

[0009] DE 10 2016 209 984 A1 discloses a method for estimating a conditional probability distribution of the maximum coefficient of friction at a current and / or future waypoint of a vehicle. This method uses an initial set of information, determined for a waypoint of the vehicle and characterizing the maximum coefficient of friction at that waypoint, to determine a first probability distribution for the maximum coefficient of friction at that waypoint. This first probability distribution is determined using a Bayesian network. The Bayesian network is a directed acyclic graph whose nodes are random variables, each node having a conditional probability distribution that depends on the probability distribution of its predecessor nodes.

[0010] Reference is also made to DE 10 2017 214 030 A1, US 2010 / 0 250 086 A1 and DE 10 2019 213 471 A1, which each deal with methods in the field of the invention.

[0011] Although this already provides the vehicle with helpful information about the current and / or future coefficient of friction on a road, there is a desire and, especially in the case of automated vehicles, a need to provide the vehicle with even more precise coefficient of friction data.

[0012] It is therefore an object of the invention to provide a method for generating a resulting friction coefficient distribution of the maximum friction coefficient as well as a driver assistance system.

[0013] The problem is solved by a method having the features of claim 1 and a driver assistance system having the features of claim 13.

[0014] Further details of the invention and combinations of different embodiments will become apparent from the features of the dependent claims.

[0015] The problem is solved by a method for generating a resulting friction coefficient distribution of the maximum friction coefficient of a vehicle at a current and / or future waypoint of a road, comprising the following steps: - Providing all known conditional probability distributions that estimate a maximum coefficient of friction under the condition of a specific road surface condition, - Providing all possible current road weather information in relation to the current and / or future waypoint of the road as a probability vector for the road surface condition, wherein the possible current road weather information in relation to the current and / or future waypoint of the road is generated by a dedicated vehicle sensor system, and wherein a road weather model is used to generate the recorded current road weather information as a probability vector for the road surface condition, - Generating the resulting friction coefficient distribution of the maximum friction coefficient at a current and / or future waypoint by determining the total probability distribution based on the known conditional probability distributions in conjunction with the probability vector.

[0016] For example, a coefficient of friction distribution of the maximum coefficient of friction for a given road surface condition is P(µ_max|dry), which corresponds to a probability distribution of the coefficient of friction in dry weather / road conditions.

[0017] Accordingly, P(µ_max|ice) corresponds, for example, to a conditional probability distribution of the maximum coefficient of friction on an icy road surface.

[0018] Current road weather information in the form of probabilities for road conditions can be purchased, for example, via third-party providers (weather portals).

[0019] Such weather portals generate current weather data for operational planning, such as setting up a construction site on the highway. This data provides, for example, a forecast of road conditions (frozen / dry), as well as the type and intensity of precipitation for a specific road section as a probability value. For instance, the weather data for a road might indicate 70% chance of wet conditions, 20% chance of very wet conditions, and 10% chance of being partially frozen.

[0020] According to the invention, the resulting friction coefficient distribution of the maximum friction coefficient at a current and / or future waypoint is generated by determining the total probability distribution from the known conditional probability distributions in conjunction with the probability vector. A precise, actual, current friction coefficient can be determined from such a generated resulting friction coefficient distribution.

[0021] The resulting friction coefficient distribution thus more accurately reflects the actual probability distribution of the maximum friction coefficient.

[0022] In further training, the conditional probability distribution takes into account not only the road surface condition but also the given road surface, or the respective conditional probability distribution is adapted with respect to the given road surface. For example, the conditional probability distribution of the maximum coefficient of friction for an icy road surface P(µ_max|ice) on cobblestones differs from that on a road surface such as tarmac.

[0023] This allows for the generation of a more accurate resulting friction coefficient distribution. Furthermore, in a subsequent development, the total probability distribution is determined by weighting the conditional probabilities using the probability vector and then summing them.

[0024] To calculate the total probability distribution, the conditional probabilities are weighted and summed, taking into account the road weather information, specified as a probability vector for the road surface condition. Examples of road weather information as a probability vector are: the probability that the road surface is dry (p(dry)), the probability that the road surface is wet (p(wet)), the probability that there is snow on the road surface (p(snow)), etc.

[0025] Thus, all road weather information relating to the current and / or future waypoint of the road can be linked as a probability vector with the conditional probabilities, resulting in a more accurate friction coefficient distribution.

[0026] In further training, the probability vector for the road surface condition is estimated using an in-house sensor system and stored information. This could, for example, be a camera / LiDAR and / or radar sensor system. This eliminates the need for a connection to, for example, an external server that provides road weather information.

[0027] In further training, the conditional probability distributions, which specify the maximum coefficient of friction with respect to all existing road surface conditions, are stored in an internal database and / or can be loaded from an external server.

[0028] This tire information, in the form of (conditional) probability distributions of the coefficient of friction for a given road surface condition, is available from public sources for general tire classes (summer, winter tires). For example, over 4,200 data sets have already been generated and stored in a comprehensive database on maximum coefficient of friction as part of research projects. The database includes data sets for conditions such as "dry," "damp," "wet," "snowy," and "icy." For specific tires, this information can be obtained from statistical models provided by the tire manufacturer or estimated.

[0029] In further training, conditional probability distributions are determined or adjusted based on the current operating parameters. These current operating parameters can include at least the vehicle's speed and / or the condition of the tires, such as pressure, temperature, tread depth, and contact pressure, which can be easily recorded by an internal sensor system. Such an adjustment can be easily performed, for example, using regression functions.

[0030] Further training involves using a road weather model to generate a probability vector from the recorded current road weather information. The road weather model can generate the probability vector taking into account, for example, the current outside temperature, humidity, wiper activity, camera information, acoustic information, etc.

[0031] This road weather information can also be obtained in the form of probability vectors, such as class probabilities, from external companies, for example commercial weather portals.

[0032] Further training involves defining a percentage risk value that indicates the risk of overestimating the actual maximum coefficient of friction at a current and / or future waypoint. This risk value is preferably set at or around 10%, which is within an acceptable range for vehicle safety.

[0033] In further training, a scalar quantile value of the resulting friction coefficient distribution is determined according to the percentage risk value, in particular 10% in this case, which is then set as the risk-adjusted friction coefficient at a current and / or future waypoint.

[0034] In further training, the risk-adjusted coefficient of friction determined in this way can be displayed to the driver as information. This allows the driver to adjust their driving style, especially their speed, to this coefficient of friction.

[0035] The friction coefficient determined in this way can also be transmitted as an input value to existing driver assistance systems.

[0036] Furthermore, the task is solved by a driver assistance system using a method as described above, wherein the driver assistance system is designed to use the friction coefficient transmitted as an input value to adjust various operating parameters.

[0037] The driver assistance system primarily functions as a braking system. This can increase driving safety, especially in wet and dark conditions, such as during the winter months.

[0038] Further features and advantages of the present invention will become apparent from the following description with reference to the accompanying figures. Variations thereof can be derived by a person skilled in the art without departing from the scope of protection of the invention as defined by the following patent claims.

[0039] The figures show schematically: Fig. 1: A method according to the invention illustrated in a first embodiment, Fig. 2: the procedure in a second form, Fig. 3: a driver assistance system according to the invention.

[0040] Fig. Figure 1 shows the inventive method for generating a resulting friction coefficient distribution at a current and / or future waypoint on a road with a given road surface.

[0041] The current waypoint represents the current vehicle position.

[0042] First, the conditional probability distributions, which specify a coefficient of friction under the condition of a certain road surface state, are provided.

[0043] A conditional probability distribution of the maximum coefficient of friction for a given road surface condition is, for example, P(µ_max|dry), which corresponds to a probability distribution of the coefficient of friction in dry weather / road conditions. Similarly, P(µ_max|icy) corresponds to a conditional probability distribution of the maximum coefficient of friction on an icy road surface.

[0044] For example, P(u_max|snow) corresponds to a conditional probability distribution of the maximum coefficient of friction on a snow-covered road.

[0045] In Fig. Figure 1 shows such conditional probability distributions as a diagram.

[0046] Alternatively, the conditional probability distributions are stored in an internal database. This database can be accessed from public sources for general tire classes (summer, winter tires), for example. For specific tires, this information may be stored in the tire manufacturer's statistical models. For instance, over 4,200 data records have already been generated and stored in a comprehensive database on maximum friction coefficients as part of research projects. This database includes data records for road surfaces such as "dry," "damp," "wet," "snowy," and "icy." Therefore, the conditional probability distribution for virtually all weather conditions and road surface conditions for the various road surfaces can be derived or extracted from the database.

[0047] The diagram shows the different conditional probability distributions P(µ_max|snow), P(µ_max|ice), P(µ_max|dry) P(µ_max|wet) plotted against the maximum coefficient of friction µ_max.

[0048] The conditional probability distribution can be a discrete probability distribution or a continuous probability distribution.

[0049] Subsequently, all possible current road weather information relating to the current and / or future waypoint of the road is provided as a probability vector for the road surface condition. This probability vector can consist of class probabilities, indicating the probability that the current road surface condition belongs to a specific class (e.g., wet).

[0050] Such road weather information in the form of probability vectors such as p(dry), p(wet), p(snow), p(ice) in relation to the current and / or future waypoint of the road can be loaded into the vehicle via app or other external data connection from third-party providers.

[0051] These third-party providers can be, for example, commercial weather portals for summer / winter services.

[0052] The probability of a dry road surface during heavy rain can also be set to zero.

[0053] Alternatively or additionally, for example to validate the loaded road weather information, a road weather model may be available to generate the probability vector.

[0054] This road weather model can be a mathematical road weather model. Subsequently, the road weather model can generate the probability vector, taking into account, for example, the current outside temperature, humidity, wiper activity, camera information, acoustic information, etc. This information can be acquired by a dedicated vehicle sensor system.

[0055] The resulting friction coefficient distribution at a current and / or future waypoint is then generated by determining the total probability distribution based on the conditional probability distributions in conjunction with the probability vector.

[0056] To calculate the total probability distribution, the conditional probabilities are weighted and added together based on the road weather information.

[0057] The resulting coefficient of friction distribution P(µ_max), forming the total probability distribution, can thus be given by: P(μ_max)=p(dry)*P(μ_max|dry)+p(wet)*P(μ_max|wet)+p(snow)*P(μ_max|snow)+p(ice)*P(μ_max|ice)… dh P(μ_max)=∑over i=1 to n for P(μ_max|road condition_i)×P(road condition_i) where i is to be understood as an index for all road conditions considered.

[0058] Thus, for example, the conditional probability distribution P(µ_max|dry) does not enter into the resulting friction coefficient distribution if the probability p(dry) is equal to zero.

[0059] Thus, all conditional probabilities according to the probability of the corresponding road surface condition are included in the resulting friction coefficient distribution.

[0060] Thus, all road weather information relating to the current and / or future waypoint of the road can be linked as a probability vector with the conditional probabilities, resulting in a more accurate friction coefficient distribution.

[0061] The resulting friction coefficient distribution can also be discrete or continuous.

[0062] A percentage risk value is then determined, indicating the risk of underestimating the actual coefficient of friction. This represents the safety-critical case. The risk value is preferably set at 10% or within the range of 10%, which is within the acceptable range for vehicle safety.

[0063] Then, based on the percentage risk value, in particular 10% in this case, a scalar quantile of the resulting friction coefficient distribution is determined, which is set as the risk-adjusted friction coefficient at a current and / or future waypoint.

[0064] Furthermore, this determined coefficient of friction can be displayed to the driver as information. This allows the driver to adjust their driving style, especially their speed, to this coefficient of friction. The coefficient of friction can also be transmitted to a driver assistance system.

[0065] Fig. Figure 2 shows the method according to the invention in a second embodiment.

[0066] First, the conditional probability distributions which specify a coefficient of friction under the condition of a certain road surface state are provided.

[0067] These conditional probability distributions are determined or adjusted based on the current operating parameters. These parameters can include at least the vehicle's speed and / or the condition of the tires, such as pressure, temperature, tread depth, and contact pressure, which can be easily detected by an internal sensor system. Such an adjustment can be easily performed, for example, using regression functions.

[0068] This allows the accuracy of the conditional probability distributions to be increased. All further steps are analogous to the initial implementation in Fig. 1.

[0069] The risk-adjusted coefficient of friction can also be transmitted to a driver assistance system 1.

[0070] Fig.Figure 3 shows such a driver assistance system 1 according to the invention. This can, for example, be a braking system which has the method according to the invention or to which the coefficient of friction thus determined at a current and / or future waypoint on a road with a given road surface of a vehicle is transmitted.

[0071] To determine the probability vector for the road condition, for example a sensor system comprising lidar / radar sensors 2 and cameras 3 may be available.

[0072] The driver assistance system 1 according to the invention, in particular when it is designed as a braking system, thus enables improved driving safety through the precise knowledge of the coefficient of friction between the tires and the road. Reference symbol list 1 Driver assistance system 2 Lidar / Radar sensors 3 cameras

Claims

[1] Method for generating a resulting friction coefficient distribution of the maximum friction coefficient of a vehicle at a current and / or future waypoint on a road with a given road surface, the method comprising the steps: - Providing all known conditional probability distributions that estimate a coefficient of friction under the condition of a specific road surface condition, - Providing all possible current road weather information in relation to the current and / or future waypoint of the road as a probability vector for the road surface condition, wherein the possible current road weather information in relation to the current and / or future waypoint of the road is generated by a dedicated vehicle sensor system, and wherein a road weather model is used to generate the recorded current road weather information as a probability vector for the road surface condition, - Generating the resulting friction coefficient distribution of the friction coefficient at a current and / or future waypoint by determining the total probability distribution based on the conditional probability distributions in conjunction with the probability vector. [2] Method according to claim 1, characterized by, that the respective conditional probability distribution takes the given road surface into account or that the respective conditional probability distribution is adapted with respect to the given road surface. [3] Method according to claim 1 or 2, characterized by , that the total probability distribution is determined by the conditional probabilities weighted by the probability vector and subsequently added together. [4] Method according to any one of the preceding claims, characterized by , that the probability vector for the road condition can be estimated using a dedicated sensor system and stored information. [5] Method according to claim 4, characterized by that the stored information includes the specific data of the tires used. [6] Method according to any one of the preceding claims, characterized by, that the conditional probability distributions, which specify the maximum coefficient of friction with respect to all existing road surface conditions, can be loaded from an external server. [7] Method according to any one of the preceding claims, characterized by , that the conditional probability distributions, which specify the maximum coefficient of friction with respect to all existing road surface conditions, are stored in an internal database. [8] Method according to any one of the preceding claims 4 to 7 characterized by , that the conditional probability distributions are determined or adjusted taking into account the current operating parameters, especially the tires. [9] Method according to any one of the preceding claims, characterized by , that a percentage risk value is set, which indicates the risk of overestimating the current maximum friction value at a current and / or future waypoint. [10] Method according to claim 9, characterized by , that a scalar quantile of the resulting friction coefficient distribution is determined based on the percentage risk value, which is set as the risk-adjusted friction coefficient at a current and / or future waypoint. [11] Method according to claim 10, characterized by that the risk-adjusted coefficient of friction is at least displayed to the driver on a screen. [12] Method according to claim 10 or 11 characterized by , that the risk-adjusted coefficient of friction is transmitted as an input value to existing driver assistance systems (1). [13] Driver assistance system (1) comprising a method according to claim 12, characterized by , that the driver assistance system is trained to use the friction coefficient transmitted as an input value to adjust various operating parameters.

Images