Method and control circuit for monitoring an admissibility condition for an automated driving mode of a motor vehicle and motor vehicle with the control circuit

The method employs a control circuit in a motor vehicle to verify the road class for automated driving modes by combining map data with sensor information, addressing safety and reliability concerns in current systems by ensuring accurate and independent verification of permitted road classes.

DE102021120043B4Active Publication Date: 2025-05-22CARIAD SE
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
DE102021120043
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Filing Date
2021-08-02
Publication Date
2025-05-22
Estimated Expiration
2041-08-02

AI Technical Summary

Technical Problem

Current methods for determining the road class in automated driving systems, such as using digital HD maps or sensor-based state machines, face challenges like localization errors, outdated map data, and inability to specify uncertainty in state machines, leading to potential safety issues and high implementation costs.

Method used

A method using a control circuit in a motor vehicle to monitor the compliance condition for automated driving modes by reading map data from a digital road map and combining it with environmental data from sensors to determine the road class ahead, ensuring that the driving mode is only active on permitted road classes through independent verification by two signal paths.

Benefits of technology

This approach enhances safety and reliability by independently verifying the road class using both map data and sensor information, reducing the reliance on single error-prone sources and minimizing the risk of deactivating automated driving modes prematurely.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000013_0000
    Figure 00000013_0000
  • Figure 00000014_0000
    Figure 00000014_0000
Patent Text Reader

Abstract

Method for monitoring an admissibility condition (21) for an automated driving mode of a motor vehicle (10) by a control circuit (18), wherein the admissibility condition (21) includes that the driving mode may only be active when driving on roads (16) of at least one predetermined approved road class (29, 31), and for checking a current road class (29, 31) of a road segment (17) ahead in the direction of travel (15) of a road (16) currently being traveled in a first signal path (19): from a digital road map (27) to the road segment (17), map data indicating the road class (29, 31) of the road segment (17) are read out from a data memory and then checked to see whether they indicate the at least one permitted road class (29, 31), wherein, independently of the first signal path (19), in a second signal path (20) operated independently of the first signal path (19), environmental data (33) describing respective environmental features of the road segment (17) are received from an environmental sensor (32) or a plurality of environmental sensors (32), characterized in that independent of the first signal path (19) in the second signal path (20) operated independently of the first signal path (19), characteristics (38) of the road segment (17) are estimated by means of an estimation routine (35) on the basis of the surrounding data (33), and it is detected whether the estimated characteristics (38) in the road segment (17) lying ahead signal a probability (30, 41) of the presence of the at least one permitted road class (29, 31) that is greater than a predetermined threshold value, and checking the admissibility condition (21) comprises signaling the presence of the at least one permitted road class (29, 31) for the road segment (17) ahead only if both the first signal path (19) and the second signal path (20) confirm the existence of the road class (29, 31), and otherwise carrying out a predetermined blocking measure (22) for the driving mode.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The invention relates to a method for monitoring a permissive condition for an automated driving mode of a motor vehicle. The permissive condition includes that the driving mode may only be active when driving on roads that belong to at least one predetermined approved road class. The invention also relates to a control circuit that can implement the method in a motor vehicle, as well as to a motor vehicle with the control circuit.

[0002] In automated vehicles from SAE Level 3 (SAE - Society of Automotive Engineers) onwards, the driver temporarily transfers responsibility for the vehicle indirectly to the vehicle manufacturer. The permissible area of ​​operation for this is defined by the so-called Operational Design Domain (ODD), i.e. the vehicle manufacturer can specify where such an automated driving mode may be activated. The ODD is determined or restricted, among other things, by the road class. Road classes that are differentiated here include motorways, motorway-like roads, country roads or city streets. Some automated driving modes, such as the motorway pilot, are only designed for use on motorways or motorway-like roads and may therefore only be activated in these driving situations. In addition, they must automatically detect a change in road class and, if possible, proactively transfer vehicle control to the driver before deactivating.This means that high safety requirements must be placed on the system to reliably identify the corresponding road classes.

[0003] According to the current state of the art, road class recognition is usually based on so-called digital HD maps (road maps with entries for, for example, peripheral development and lane layouts). The road class is stored a priori in the HD map or, more generally, in a digital road map and then read from the map using vehicle location (e.g., via GNSS - Global Navigation Satellite System, e.g., GPS - Global Positioning System).

[0004] Recognizing road class based solely on digital road maps presents several challenges. Firstly, localization errors can occur, for example, due to GPS inaccuracies. Secondly, the map content must be continuously updated, which is difficult to implement. Highly automated systems with SAE levels greater than 2 generally require a safety integrity level of ASIL-D (ASIL - automotive safety level) for ODD recognition (according to ISO 26262). The resulting technical requirements would be associated with considerable effort and expense if implemented solely using road maps.

[0005] Alternatively, the road class can be determined based on sensor measurement data. According to DE 10 2012 218 362 A1, a state machine is used to determine a road class based on human experience and environmental data. If this fails to produce a result, the result of a decision tree induced by machine learning is used to estimate a road class. Determining the road class based on a state machine has the disadvantage that a state machine cannot indicate uncertainty regarding its own state depending on the reliability of its input data.

[0006] DE 102 54 806 A1 discloses combining environmental data from multiple sensors and map data from a digital road map by means of information fusion to obtain information about the road class currently being traveled. The disadvantage of such fusion is that a clear result from one information source can be relativized or weakened by uncertain results from another information source.

[0007] DE 10 2018 208 593 A1 discloses a method for checking whether a motor vehicle's driving mode can be changed safely. In two successive test stages, a check is first carried out to determine whether, based on the map data of a digital road map, a permissible road type for activating the driving mode exists anywhere nearby. If the first test stage is successfully completed, a second test stage uses environmental data from sensors to determine whether the motor vehicle is actually on a road of that road class. The test is only carried out for the road section currently being traveled on, which is why the method preferentially checks the activation of the driving mode, but not its deactivation upon leaving a permissible road class, i.e., upon exiting the ODD.

[0008] DE 10 2019 208 533 A1 describes a vehicle that only activates an autonomous driving mode if, based on the vehicle's current geoposition, it detects that the vehicle is on a road class approved for this driving mode. Since the geoposition can only be determined with a certain degree of variance, meaning it can also point to an area adjacent to the road, for example, an additional check is performed to determine whether there is either another road or only open space in the vicinity. If there are only open spaces adjacent to the road, it is assumed that the vehicle is on the only recorded road, and thus a positive validation of the road class can be achieved.

[0009] The invention is based on the object of checking in a motor vehicle for the operation of an automated driving mode whether a permissible road class exists for which the operation of the automated driving function is permitted.

[0010] The problem is solved by the subject matter of the independent patent claims. Advantageous developments result from technical features described in the dependent patent claims, in the following description, and in the figures.

[0011] As one solution, the invention comprises a method for monitoring a permissive condition for an automated driving mode of a motor vehicle by means of a control circuit. The permissive condition comprises that the driving mode may only be active when driving on roads of at least one predetermined permitted road class. To check the current road class of a road segment lying ahead in the direction of travel of a road currently being traveled, map data indicating the road class of the road segment are read from a data memory in a first signal path from a digital road map to the road segment and then checked to determine whether they indicate a road class that corresponds to the at least one road class permitted for the driving mode.The admissibility condition is checked in particular to determine, in the case of an already activated driving mode, whether the driving mode must be deactivated because the road segment ahead no longer corresponds to at least one permitted road class. As a driving mode, driver assistance for motorway driving, ACC (Automatic Cruise Control), adaptive cruise control, or an overtaking assistant can be provided in a conventional manner, to name just a few examples. In general, the automated driving mode can provide that, in the activated driving mode, longitudinal guidance (acceleration and / or braking) and / or lateral guidance (steering) is carried out automatically by at least one control unit of the motor vehicle without any driver intervention.

[0012] The road class can also be checked for the currently traveled road section or the currently traveled road segment, but according to the invention it is primarily a road segment ahead, for example a road segment that begins, for example, 10 centimeters to 10 meters in front of the motor vehicle and ends, for example, 30 meters to 1 kilometer in front of the motor vehicle. Using the road map, it can be checked whether the map data stored therein relating to this road segment ahead indicate that this road segment is a component or classified as an approved road class, i.e. as an element of the set of at least one approved road class. One or more than one road class can be defined as approved, which is why the formulation “at least one predetermined approved road class” is used here.Which road class is approved for which driving mode depends on the driving mode and can be determined by a specialist.

[0013] However, the use of a road map may result in outdated map data being available and / or the location of the motor vehicle in relation to the road map being subject to variance, tolerance or fluctuation, as is known per se for receivers of a position signal from a GNSS (Global Navigation Satellite System), such as GPS (Global Positioning System).

[0014] Therefore, independent of the first signal path, environmental data describing environmental features of the road segment (e.g. existing road signs or recognizable persons) are received from one or more environmental sensors in a second signal path independent of the first signal path, and on the basis of the environmental data, characteristics of the road segment (e.g. the characteristic “blue motorway sign” or “persons present”) are estimated using a predetermined estimation routine.

[0015] It is detected whether the estimated characteristics represent a probability of the presence of the at least one permitted road class in the road segment ahead, which is greater than a predetermined threshold. The environmental data can be the sensor data from the at least one environmental sensor and / or processed sensor data, for example, filtered sensor data and / or sensor data combined using sensor fusion. The respectively detected characteristic can include, for example, "pedestrians present," "central barrier present," "highway signs of a specific color present," to name just a few examples.A corresponding estimation routine for recognizing such characteristics from environmental data can be provided, for example, based on a machine learning algorithm, such as an artificial neural network, as is known from the state of the art for object recognition or machine vision. The existence probability that the characteristics indicate the presence of a specific road class or at least the presence of the set of permitted road classes can then be determined based on the estimation results for the individual characteristics, such as a confidence metric or estimation variance, when estimating the respective characteristic. This will be explained in more detail below.By combining estimated characteristics, such as the presence of a guardrail and / or the presence of pedestrians, the probability of existence for the presence of at least one permitted road class is then determined from the individual key figures of the characteristics, i.e. their estimation reliability. Thus, the second signal path also signals, independently of the first signal path, whether the road segment ahead belongs to the at least one permitted road class or not. The signal of membership is output if the probability of existence is greater than the threshold value. Otherwise, it is signaled that the road segment ahead (or a sub-interval thereof) does not belong to the at least one permitted road class.

[0016] The actual checking of the admissibility condition involves signaling the presence of the at least one permitted road class for the upcoming road segment only if both the first signal path and the second signal path confirm or signal the existence of the road class. Otherwise, a predetermined blocking measure is implemented for the driving mode. In other words, the two signal paths are evaluated independently of one another, and only if both signal paths independently signal or confirm the presence or existence of the at least one permitted road class in the area of ​​the upcoming road segment is it assumed that the admissibility condition is met for the driving mode. Otherwise, the blocking measure is implemented, which will be explained in more detail below.This means that you do not have to rely on signal paths being implemented absolutely error-free according to ASIL-D, but only the mutual adjustment of both signal paths must lead to the ASIL-D level.

[0017] For example, the two signal paths can be implemented as independently operated software modules, allowing them to perform the road class check independently of each other.

[0018] The invention also includes further developments which result in additional advantages.

[0019] It should be noted here: When estimating a characteristic, three aspects can be checked: firstly, whether the characteristic is present at all (“presence”) (a classifier, such as an artificial neural network, responds to the presence of the characteristic at all, e.g., to “people” in camera images), an “existence confidence” of the characteristic (the classifier signals the probability that the characteristic is present or how certain the estimate is), and a “state” of the characteristic (e.g., how many kilometers the characteristic has been present or what form it takes, such as “many pedestrians present” versus “few pedestrians present”). Corresponding estimation routines are available from the state of the art, for example, in connection with environmental monitoring in a motor vehicle for operating an autopilot.

[0020] A further development involves dividing the road segment ahead into several different distance intervals located one behind the other in the direction of travel, and performing a separate check of the admissibility condition for each distance interval using the two signal paths. For each distance interval, the map data and characteristics from the respective distance interval are used. In other words, the road segment ahead is not monitored as a uniform area across which the overall probability of existence of an approved road class is checked. Instead, subsections or subintervals with different distances from the vehicle (hence "distance intervals") are individually checked to determine whether they meet the admissibility condition.This has the advantage of preventing the "averaging out" of indicators or characteristics across the entire road segment. If, for example, the road no longer belongs to at least one permitted road class in the furthest distance interval, there may be corresponding characteristics in this furthest distance interval that clearly indicate this. For example, if the motor vehicle is driving on a motorway and changes into the turning lane, the admissibility condition "motorway present" may still be met for the immediately preceding distance interval, whereas at the end of the exit in a bend for leaving the motorway, the characteristic "bend with curve radius less than X meters" may be present, signaling that the road class "motorway" is no longer present here.However, this unique characteristic would potentially be averaged out when calculating the probability of existence for the "motorway" road class across the entire preceding road segment because the characteristics of the "motorway" road class predominate in the preceding distance intervals. However, if the probability of existence for the "motorway" road class is calculated separately for the distance intervals, the characteristic relating to the curve radius can be reliably identified for the last, furthest distance interval, and it can be signaled that no motorway or, in general, the permissible road class is present in this distance interval. Preferably, the road segment is divided into at least two distance intervals, preferably more than two, preferably more than three, and in particular more than four distance intervals.For each characteristic, a separate subdivision of the road segment into characteristic-related distance intervals of different lengths along the direction of travel can also be provided, for example distance intervals with a length of 50 meters, distance intervals with a length of 100 meters, to name just a few examples.

[0021] A further development includes the following: if a distance interval closest to the motor vehicle satisfies the admissibility condition (i.e., the driving mode may still be active) but at least one distance interval behind it violates the admissibility condition (i.e., the driving mode should be deactivated until this distance interval ahead is reached), the blocking measure comprises a handover routine for handing over longitudinal and / or lateral guidance of the motor vehicle to a driver of the motor vehicle. In other words, it is recognized for a future time or a future location along the direction of travel in the road segment that the driving mode must be deactivated there. Accordingly, the handover routine is triggered or executed as a blocking measure for blocking or deactivating the driving mode.This handover routine is known in the art and can, for example, include a notification or output to the driver that they should prepare to take over control of the vehicle. Additionally or alternatively, an emergency stop can be used to ensure that the driving mode does not have to be operated in an unapproved or impermissible road class if, for example, the handover procedure or routine fails for the driver.

[0022] Further training includes the fact that said blocking measure includes that in the event that the admissibility condition is violated in the present road segment, when the driving mode is deactivated, activation of the driving mode is blocked and / or when the driving mode is activated, a handover to a driver and / or an emergency stop is carried out and then the driving mode is deactivated. Generally, therefore, when the driving mode is deactivated, the blocking measure can prevent the driving mode from being activated. For example, a corresponding operating element in the motor vehicle, which is provided for activating the driving mode, can be switched off or faded out (in the case of a touch screen). When the driving mode is activated, the handover to the driver and / or the emergency stop is forced.

[0023] A further development comprises that at least in the second signal path (environmental data) the at least one permissible road class and / or at least one predetermined impermissible road class (driving mode must be deactivated here) is defined and the current existence probability for each of the road classes is calculated in parallel. Preferably, a memory function for at least one of the road classes or all of the road classes also takes into account previous road segments already traveled when calculating their existence probability, wherein this memory function comprises a degradation behavior of the characteristics estimated in these previous road segments (i.e. a forgetting factor) depending on a temporal and / or spatial distance from the respective previous road segment.In other words, for each location on the road, an estimate or calculation of the existence probability for each road class is available. An existence probability with a value greater than the specified threshold should then be signaled or determined for at least one road class, preferably for exactly one road class. However, by performing an estimation of the existence probability for all road classes in parallel or simultaneously, a previous or past road segment, or the characteristics estimated therein, can also be considered in the calculation of the existence probability and, for example, accumulated or taken into account using a regression.For example, if a central guardrail begins in a road segment and the motor vehicle passes this beginning of the central guardrail and the central guardrail continues into the road, then after the motor vehicle has traveled a predetermined minimum distance, for example 1 kilometer, the characteristic "central guardrail present" can be taken into account not only for the current road segment, but also, using the memory function, for previously passed or driven through road segments or locations on the road. However, a characteristic loses its meaning or significance when estimating the probability of existence of the road class of the current road segment, which is why characteristics from earlier in time and / or location are removed or hidden using the forgetting factor.As a time interval, it can be provided, for example, that a characteristic is hidden whose observation time or estimated time lies further back than a time value in a range of, for example, 10 seconds to 15 minutes. Additionally or alternatively, the described localization of the motor vehicle can be used to determine how large a spatial distance of the motor vehicle from the observation location for a characteristic has become. A distance in a range of more than 20 meters to more than 5 kilometers can be selected as the value for forgetting or hiding a characteristic, to name just a few examples.Therefore, if the clues or characteristics for a particular road class “multiply” or repeat or accumulate or aggregate during the journey, this can be detected by the memory function and taken into account when calculating the probability of existence for a road class that has not been detected so far.

[0024] A further development involves calculating, for each road class examined, an overall key figure for each characteristic based on measured classification uncertainties and / or existence uncertainties and / or state uncertainties when sensing the corresponding environmental features using the environmental data. The overall key figure indicates the degree of expression of this characteristic, with a higher absolute value of the overall key figure, for example, indicating a stronger expression. To determine the existence probability from the observed characteristics, each characteristic can be quantified using a single overall key figure.This allows the previously described classification uncertainty and / or existence uncertainty and / or state uncertainty for the estimation or observation of the characteristics (e.g., pedestrians or central barrier) and / or their existence (certainly present or present with X%) and the state (many or few pedestrians) to be summarized. This overall key figure describes how reliably the respective characteristic could be derived from the environmental data from at least one sensor. For example, an output from a machine learning algorithm, such as an output from an artificial neural network, can be used here.

[0025] A further development comprises performing, for each road class checked, a class-specific selection and / or a class-specific weighting from the total characteristics observed in the road segment according to relevance data stored in the control circuit, which indicate a relevance of the characteristics with respect to the respective road class. This prevents an estimation or calculation of an existence probability for a road class from being rendered uncertain by taking into account a characteristic that is irrelevant for the calculation of the existence probability, for example, for which there is a large observation uncertainty or estimation uncertainty and / or which is not a reliable indicator for a road class because it is typical for both an authorized and an unauthorized road class.This takes into account that there may be characteristics that are typical or characteristic of both a permitted road class and a prohibited road class. Therefore, for the distinction between permitted and prohibited road classes, this characteristic can be ignored or disregarded as irrelevant. The categories for selection can be, for example, absolutely necessary, conditionally necessary, or optional. An absolutely necessary characteristic can indicate that without an observation or estimation of this characteristic, the corresponding road class may not exist, or conversely, the characteristic must be missing for the road class to exist.A conditionally necessary characteristic means that it can, for example, currently be hidden, i.e. it does not have to be observed for the current road segment, but should be signaled, for example by means of the memory function, that this characteristic must have been present at least temporarily within a past period, for example within the last ten minutes, or within a predetermined distance, for example within the last kilometer.

[0026] A further development includes, in the event that the first signal path and the second signal path signal the same road class, the signaled road class is entered into an environment model (so-called dynamic digital environment map), based on which a semi-autonomous or fully autonomous driving function in the automatic driving mode plans a driving trajectory and selects a behavior rule to be followed and / or a traffic rule to be observed depending on the road class entered in the environment model. Thus, the verification of the admissibility condition is also used to control the road class signaled by both signal paths to determine the driving behavior of the motor vehicle. A behavior rule to be followed can, for example, indicate that overtaking is preferred or, conversely, that overtaking should be avoided.A traffic rule to be observed can be the traffic rule specified for the road class, which is implicitly specified by the road class (such as a maximum speed limit or speed limit for rural roads) and / or a driving style, e.g., the probability of an overtaking attempt. Thus, the signal paths are also used to provide information about the current road class in the environment model.

[0027] As a further solution, the invention comprises a control circuit for a motor vehicle, wherein the control circuit has a processor device that is configured to carry out an embodiment of the method according to the invention. The control circuit can have a data processing apparatus or a processor device that is configured to carry out an embodiment of the method according to the invention. For this purpose, the processor device can have at least one microprocessor and / or at least one microcontroller and / or at least one FPGA (Field Programmable Gate Array) and / or at least one DSP (Digital Signal Processor). Furthermore, the processor device can have program code that is configured to carry out the embodiment of the method according to the invention when executed by the processor device. The program code can be stored in a data memory of the processor device.The control circuit can be configured as a single control unit or as a network of multiple control units for the motor vehicle. The control circuit can additionally or alternatively comprise a central computer for a motor vehicle.

[0028] As a further solution, the invention comprises a motor vehicle with at least one sensor for generating sensor-based environmental data and with an embodiment of the control circuit according to the invention. The motor vehicle according to the invention is preferably designed as a motor vehicle, in particular as a passenger car or truck, or as a passenger bus or motorcycle.

[0029] The invention also encompasses combinations of the features of the described embodiments. The invention therefore also encompasses implementations that each have a combination of the features of several of the described embodiments, unless the embodiments are described as mutually exclusive.

[0030] Exemplary embodiments of the invention are described below. Shown are: Fig. 1 a schematic representation of an embodiment of the motor vehicle according to the invention with an embodiment of the control circuit according to the invention, by means of which an embodiment of the method according to the invention can be carried out; and Fig. 2 a sketch illustrating a calculation of the probability of existence of a permissible road class.

[0031] The exemplary embodiments explained below are preferred embodiments of the invention. In the exemplary embodiments, the described components of the embodiments each represent individual features of the invention that can be considered independently of one another, each of which also develops the invention independently of one another. Therefore, the disclosure is intended to encompass combinations of the features of the embodiments other than those shown. Furthermore, the described embodiments can also be supplemented by further features of the invention already described.

[0032] In the figures, the same reference symbols denote elements with the same function.

[0033] Fig. 1 shows a motor vehicle 10, which may be a motor vehicle, in particular a passenger car or a truck. An automated driving function 11, e.g., an autopilot, may be provided in the motor vehicle 10, which may, for example, control a drive motor 12 and / or brakes 13 for longitudinal guidance and a steering system 14 for lateral guidance. For example, only longitudinal guidance or lateral guidance may be provided. The autopilot 11 is active when a driving mode F is activated. It may be provided that the driving function 11 is only permitted for predetermined road classes or at least one predetermined road class, i.e., may only be active on at least one permitted road class, e.g., a motorway or country road. The motor vehicle 10 may currently be traveling along a direction of travel 15 on a road 16.For the road 16, it can be checked or determined in the motor vehicle 10 whether a road segment 17 lying ahead of the motor vehicle 10 in the direction of travel 15 corresponds to the at least one authorized or permissible road class. For this purpose, a control circuit 18 can be provided in the motor vehicle 10, which can be implemented, for example, by a control unit or a network of multiple control units and / or a central computer of the motor vehicle 10. Two mutually independent signal paths 19, 20 can be implemented in the control circuit 18, for example on the basis of a program code or software, in order to check an admissibility condition 21 indicating that the road segment 17 lying ahead corresponds to the at least one authorized road class. If this admissibility condition 21 is violated, a predetermined blocking measure 22 can be triggered by the control circuit 18.This can then deactivate driving mode F in driving function 11 or prevent its activation.

[0034] The signal path 19 can provide that a current geoposition 26 of the motor vehicle 10 is located or determined by means of a receiver 23 for a position signal 24 of a GNSS 25, for example, the GPS, and that these geoposition data of the geoposition 26 are received by the control circuit 18. Using a digital road map 27, for example from a navigation database 28, it can be determined whether the road 16 at the geoposition 26 corresponds to at least one permitted road class. A currently existing or traveled road class 29, or a road class ahead, can therefore be estimated from the map data of the digital road map 27. This road class 29, as signaled by the first signal path 19, can be used as input for the admissibility condition 21.The second signal path 20 can provide that, for one or more possible road classes, a probability of existence 30 of the road class and / or an estimated road class 31 determined by a threshold comparison is signaled through the second signal path 20. For this purpose, sensor data or environmental data 33 can be received from at least one environmental sensor 32. Examples of environmental sensors 32 are: a camera, a radar, a lidar, an ultrasonic sensor. The respective detection areas 34 of the environmental sensors 32 can be aligned with the road segment 17.Based on the environmental data, an estimation routine 35, for example an artificial neural network, can determine, based on environmental features 36 described by the environmental data 33, for example objects in the environment 37 of the motor vehicle 10 in the area of ​​the road segment 17 or in the adjacent area to the right and left of the road 16, which characteristics 38 are present in the road segment 17, for example whether a central barrier is present and / or whether pedestrians are present. Based on the respective expression or degree of the characteristic 38, a decision can be made or determined on the basis of relevance data 39 as to which of the characteristics 38 are to be used as the basis for which of the road classes being checked, in order to calculate, by means of a probability estimation 40, a respective existence probability 41 for the presence or existence of the respective road class in the road segment 17.Using a threshold value comparison 42, the respective existence probability 41 can then be compared with a threshold value 43 to determine which road class 31 is to be signaled as a further input by the second signal path 20 of the admissibility condition 21. Only if both signal paths 19, 20 signal the same road class 29, 31 does the blocking measure 22 remain deactivated or unused. If the blocking measure 22 is activated, a handover to the driver and / or an emergency stop is performed when driving mode F is active.

[0035] Fig. 2 illustrates how the estimated road class 31 is determined in the motor vehicle 10 for the road segment ahead, for example in the second signal path 20. It shows how different characteristics C1 to C6, for example "central barrier present", "pedestrian present", have been estimated from the environmental data 33 and how an overall key figure G has been determined for each characteristic 38. Furthermore, it is shown that the overall key figure G within the road segment 17 can be calculated independently or individually for each characteristic 38 at different distance intervals 50. For each characteristic 38, a separate length of the distance intervals 50 along the direction of travel 15 can be defined or provided. In the overlapping areas of the distance intervals 50, the overall key figures G of the different characteristics 38 can be combined with one another or used together.For example, the existence probability E 41 can be calculated for each of the different road classes K, represented here by four different road classes K1, K2, K3, K4, and different overall key figures G can be used or combined to determine the existence probability E of a respective road class K1, K2, K3, K4 using the relevance data 39 (represented by dashed lines). For each resulting distance interval, a separate estimate or signaling of an estimated road class 31 can then be made, for example the road class K1 to K4 with the greatest existence probability E or with an additional condition that the existence probability E must be greater than a predetermined minimum value. In . Fig.2 illustrates that a distance interval A1 closest to motor vehicle 10 corresponds to a permissible road class, while the subsequent distance interval A2 corresponds to an impermissible road class (each symbolized by different hatching). The admissibility condition 21 is thus violated within road segment 17 due to the distance interval A2. Accordingly, the blocking measure 22 is triggered, which, due to the still available distance interval 50 with a permissible road class, can, for example, execute or trigger a takeover procedure to transfer the driving task to the driver.

[0036] To meet the safety requirements for road class recognition in automated driving, a decomposition into two redundant ASIL B (D) road classifications is preferably performed according to ISO 26262. First, a road classification is performed using localization on an HD map (digital road map).

[0037] Second, classification is performed using environmental features detected by the vehicle's own environmental sensors, which represents the novelty. The goal of this new procedure is to estimate the road class by comparing the features detected in the environment and their conditions with the characteristics determined from laws, guidelines, and statistics.

[0038] In order to meet the safety requirements for road class recognition, a decomposition into two redundant ASIL B (D) road classifications is carried out according to ISO 26262. On the one hand, a road classification is carried out using localization on an HD map. On the other hand, a classification is carried out using environmental features recorded by the vehicle's own environmental sensors. The result of this classification is a distribution across the road classes defined in the function, whereby reliability is taken into account in the level of the probability value of each road class. An environmental feature is abstract information in the environment of the ego vehicle, which is recorded by various environmental sensors (e.g. radar, lidar and camera), processed and then combined by sensor data fusion. Important environmental features include lane markings, traffic signs, crash barriers, beacons, walls, vehicles and living beings.In addition to their geometric position or course in the vehicle's own coordinate system, they all have other states such as vehicle type, speeds and accelerations; color, dash type and width of lane markings; and sign type and content of traffic signs. To determine the road class, the states of all environmental features perceived in the surrounding area are analyzed. They are then compared with characteristics typical of the road class in order to determine a probability-based statement for the road class. Such characteristics are examined and defined a priori. Since these can vary considerably from country to country or change during the product's life cycle, they are stored in a database and kept up to date via updates.The starting point for defining the characteristics of different road classes is, on the one hand, the existing laws and guidelines of the respective country of operation. In Germany, for example, the StVO (Road Traffic Regulations), VwV-StVO (Value Regulations for Road Construction), or the various guidelines for the construction of roads are worth mentioning. These formulate guidelines for the alignment and cross-sections of lane markings, the type, size, and position of traffic signs, and the course and arrangement of guardrails. Similar regulations also exist for Chinese and American road construction. On the other hand, characteristics can be determined through data-driven, statistical analyses. Examples of this include the occurrence or frequency of certain traffic signs in the various road classes, the existence and behavior patterns of other road users, or even the road topology.This means that there are both characteristics that allow clear conclusions about the road class, and others that only provide an indication, in the form of an increased probability for one or more road classes. The goal of the method is to estimate the road class by comparing the characteristics recorded in the environment and their conditions with the characteristics determined from laws, guidelines, and statistics. The environmental characteristics used preferably correspond to safety integrity level ASIL-B (D). The environmental model is enriched with the estimated road class.

[0039] The advantage of this method is that the reliability of the input data has a direct influence on the probability value of a road class. If the sensor measurement data is poor, this can be taken into account when deciding on the availability of an automated driving function. Furthermore, uncertainty relationships can be established between two or more road classes. If all road classes with a significant probability value are approved for the automated driving function, a lower overall safety level for a specific road class may be sufficient to offer the driving function. The redundancy for recognizing the road class using HD map information enables greater safety integrity and reliability. Despite possible errors in localization or errors in the HD map (digital road map), the road class can thus be determined by the onboard sensors.Furthermore, the decomposition of road class recognition into two redundant processes reduces the technical complexity and safety requirements for its implementation. By incorporating information from various sensor principles (radar, lidar, camera) into the estimation, reliability can be increased, since, for example, purely camera-based road class recognition would be significantly more error-prone. In addition to activating and deactivating the system, the road class is also required for modeling traffic rules, which also increases reliability here.

[0040] The existing environmental features are recorded by the vehicle's own sensors and combined using sensor data fusion. The output of this fusion is various properties of the environmental features, such as classification, degree of existence, pose, kinematics and geometric progression. This data is used to generate an environmental model of the ego vehicle. From this environmental model, the characteristics to be checked are extracted using the states and relationships of the environmental features. The characteristics are higher-level indicators for properties that the environmental model exhibits. For example, the characteristic "pedestrian present" is determined from individual pedestrians in the environmental model. These extracted characteristics are compared with the characteristics stored as relevant or typical for the road class.For each characteristic, an overall key figure is calculated based on the measured classification uncertainties, existence uncertainties, and state uncertainties of the corresponding environmental features. The overall key figure provides information about the severity of this characteristic, with a higher absolute value indicating a stronger severity. The characteristics are updated in each calculation step, taking previous results into account. Over time, the various characteristics can exhibit a build-up or decay behavior in order to stabilize their behavior and to account for individual events that have only a limited temporal validity. For characteristics determined from location-specific environmental features, a location-dependent component can also be taken into account.For example, the distance to a certain environmental feature must exceed at least a certain distance in order for the temporal decay behavior of the characteristic to be applied.

[0041] The probability-based estimation of a road class is carried out by combining the characteristics relevant to that class. The way in which the characteristics are combined to estimate a road class is not fixed, but is queried via the database. Both which characteristics are used to estimate a road class and their influence and weighting on the estimation result can change. In addition, the characteristics relevant to a road class are divided into the categories absolutely necessary, conditionally necessary, and optional. Absolutely necessary characteristics must have a minimum level of significance for the road class to be declared as present. The same applies to conditionally necessary characteristics, with the difference that there is a condition that allows the road class to be declared as present nonetheless.Optional characteristics contribute to the plausibility of a road class, but have no direct influence on whether or not a road class can be declared as existing. All defined road classes are estimated in parallel and independently of one another, so that an estimated value is available for each defined road class at the end of a processing step. The existence probabilities of the characteristics are also evaluated depending on the distance in the longitudinal direction of the road from the ego vehicle. This results in distance intervals for each characteristic in which the expression of the characteristic and / or the existence probability can differ. The intervals of all characteristics are combined, and an estimate of the individual road classes is performed for each resulting interval. This also allows for the detection of any prior changes in the road class.By combining the individual estimates, a distribution of road classes with their determined probability values ​​is ultimately obtained.

[0042] The following characteristics are taken into account for estimating the road class: - Existence and properties of traffic lights (switch-on state, light signal colour, light signal shape, position, number) - Existence of cross traffic - Existence of oncoming traffic - Existence and nature of a separation from oncoming traffic (e.g. guardrail, - concrete barrier, wide green strip) - Existing curve radii (via lane markings or edge buildings) - Existence of intersections - Existence of roundabouts - Existence of entrances / exits - Existence and characteristics of traffic signs (type, size, color, position) - Existence and characteristics of pedestrians / cyclists (number, frequency, position, direction of movement) - Width of lanes - Properties of lane markings (width, dash type, gradient, reflectivity (can be measured by lidar sensor)) - Gradient (along and across the road) - Number of lanes - Existence of a hard shoulder - Existence and position of street lighting - Average speeds of vehicles per lane - Distances between vehicles (longitudinal and transverse) - Existence or frequency of vehicle types - Maximum speeds of vehicles - References - technical regulations for road construction in Germany - Guidelines for the construction of motorways

[0043] Overall, the examples show how an estimate of a road class can be provided based on sensor-based environmental data.

Claims

[1] Method for monitoring an admissibility condition (21) for an automated driving mode of a motor vehicle (10) by a control circuit (18), wherein the admissibility condition (21) includes that the driving mode may only be active when driving on roads (16) of at least one predetermined approved road class (29, 31), and for checking a current road class (29, 31) of a road segment (17) ahead in the direction of travel (15) of a road (16) currently being traveled in a first signal path (19): from a digital road map (27) to the road segment (17), map data indicating the road class (29, 31) of the road segment (17) are read out from a data memory and then checked to see whether they indicate the at least one permitted road class (29, 31), wherein, independently of the first signal path (19), in a second signal path (20) operated independently of the first signal path (19), environmental data (33) describing respective environmental features of the road segment (17) are received from an environmental sensor (32) or a plurality of environmental sensors (32), characterized by , that independent of the first signal path (19) in the second signal path (20) operated independently of the first signal path (19), characteristics (38) of the road segment (17) are estimated by means of an estimation routine (35) on the basis of the surrounding data (33), and it is detected whether the estimated characteristics (38) in the road segment (17) lying ahead signal a probability (30, 41) of the presence of the at least one permitted road class (29, 31) that is greater than a predetermined threshold value, and checking the admissibility condition (21) comprises signaling the presence of the at least one permitted road class (29, 31) for the road segment (17) ahead only if both the first signal path (19) and the second signal path (20) confirm the existence of the road class (29, 31), and otherwise carrying out a predetermined blocking measure (22) for the driving mode. [2] Method according to claim 1, wherein the road segment (17) lying ahead is divided into a plurality of different distance intervals (50) lying one behind the other in the direction of travel (15), and for each distance interval a separate check of the admissibility condition (21) is carried out by means of the two signal paths, wherein for each distance interval the map data and the characteristics (38) within the respective distance interval (50) are used. [3] Method according to claim 2, wherein, in the event that a distance interval closest to the motor vehicle (10) satisfies the admissibility condition (21) and at least one distance interval behind it violates the admissibility condition (21), the blocking measure (22) comprises a transfer routine for transferring a longitudinal guidance and / or transverse guidance of the motor vehicle (10) to a driver of the motor vehicle (10). [4] Method according to one of the preceding claims, wherein the blocking measure (22) comprises that, in the event that the admissibility condition (21) is violated in the road segment (17) ahead, activation of the driving mode is blocked when the driving mode is deactivated and / or a handover to a driver and / or an emergency stop is carried out when the driving mode is activated and then the driving mode is deactivated. [5] Method according to one of the preceding claims, wherein at least in the second signal path (20) for each permissible road class (29, 31) and / or for at least one predetermined impermissible road class (29, 31) the current existence probability (30, 41) is calculated in parallel and, for at least one of the road classes (29, 31), a memory function takes into account previous road segments (17) already travelled through in order to calculate their existence probability (30, 41), wherein the memory function comprises a degradation behavior of the characteristics (38) estimated in these previous road segments (17) as a function of a temporal and / or spatial distance from the respective previous road segment (17). [6] Method according to one of the preceding claims, wherein for each checked road class (29, 31) for each characteristic (38) an overall characteristic number is calculated on the basis of measured classification uncertainties and / or existence uncertainties and / or state uncertainties when sensing the corresponding environmental features using the environmental data (33), and the overall characteristic number signals an expression of this characteristic (38). [7] Method according to one of the preceding claims, wherein for each checked road class (29, 31) from the observed characteristics (38) a class-specific selection and / or a class-specific weighting is carried out according to relevance data (39) stored in the control circuit (18), which indicate a relevance of the characteristics (38) with respect to the respective road class (29, 31). [8] Method according to one of the preceding claims, wherein, in the event that the first signal path (19) and the second signal path (20) signal the same road class (29, 31), the signaled road class (29, 31) is entered into an environment model, on the basis of which a partially autonomous or fully autonomous driving function (11) plans a driving trajectory in the automatic driving mode and in doing so selects a behavior rule to be followed and / or a traffic rule to be observed depending on the road class (29, 31) entered in the environment model. [9] Control circuit (18) for a motor vehicle (10), wherein the control circuit (18) comprises a processor device which is configured to carry out a method according to one of the preceding claims. [10] Motor vehicle (10) with at least one sensor for generating sensor-based environmental data (33) and with a control circuit (18) according to claim 9.

Citation Information

Patent Citations

  • Road type estimation using sensor-based environmental data

    DE102012218362A1

  • Procedures for checking whether a change of driving mode can be carried out safely

    DE102018208593A1

  • Procedure for identifying a road class

    DE102019208533A1

  • Motor vehicle data processing method in which data or information derived from at least two sources are combined to provide a prediction relating to the type of road being traveled on or the course of the road

    DE10254806A1