DE102021120043A1 - Method and control circuit for monitoring a permissibility condition for an automated driving mode of a motor vehicle and motor vehicle with the control circuit - Google Patents

Abstract

The invention relates to a method for monitoring an admissibility condition (21) for an automated driving mode of a motor vehicle (10), the admissibility condition (21) comprising that the driving mode is only activated when driving on roads (16) of a predetermined authorized road class (29, 31 ) may be active. In a first signal path (19), map data from a digital road map (27) for a road segment (17) lying ahead are checked to determine whether they indicate the permitted road class (29, 31). The invention provides that, independently of the first signal path (19), environmental data (33) are received in a second signal path (20) from an environmental sensor (32) and the environmental data (33) are used to check whether there is a probability of existence (30, 41 ) a presence of the approved road class (29, 31) is greater than a predetermined threshold. The admissibility condition (21) is only met if both the first signal path (19) and the second signal path (20) confirm the existence of the road class (29, 31).

Description

The invention relates to a method for monitoring a permissibility condition for an automated driving mode of a motor vehicle. The admissibility condition includes that the driving mode may only be active when driving on roads that belong to at least one predetermined authorized road class. The invention also relates to a control circuit that can carry out the method in a motor vehicle, and a motor vehicle with the control circuit.

In the case of automated vehicles from SAE Level 3 (SAE - Society of Automotive Engineers), the driver temporarily hands over responsibility for the vehicle indirectly to the vehicle manufacturer. The permissible area of use 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 limited by the road class, among other things. Road classes that are differentiated here are, for example, motorways, roads similar to motorways, country roads or city streets. Some automated driving modes, such as the highway pilot, are only designed for use on freeways or roads similar to freeways and may therefore only be activated in these driving situations. In addition, they must automatically recognize a change in the road class and preferably proactively hand over control of the vehicle to the driver before deactivating. As a result, high security requirements are placed on the system to reliably recognize the relevant road classes.

According to the current state of the art, the road class is usually recognized on the basis of so-called digital HD maps (road maps with entries for e.g. peripheral buildings and lane courses). The road class is stored a-priori in the HD map or generally a digital road map and then read out on the map by localizing the vehicle (e.g. via GNSS - Global Navigation Satellite System, e.g. GPS - Global Positioning System).

Road class detection based solely on digital road maps brings with it some challenges. On the one hand, localization errors can occur, for example due to GPS inaccuracies. On the other hand, the map content must be continuously up-to-date, which is difficult to implement. Highly automated systems with SAE level greater than 2 usually require a safety integrity level ASIL-D (ASIL - automotive safety level) for the detection of the ODD (according to ISO 26262). The resulting technical requirements would be associated with great effort and costs if implemented solely using a road map.

Alternatively, the road class can be determined on the basis of sensor measurement data. According to the DE 10 2012 218 362 A1 uses a state machine that uses environmental data to determine a road class based on human experience. If this does not come to a conclusion, the result of a decision tree induced by machine learning is used and a road class is estimated. Determining the road class on the basis of a state machine has the disadvantage that a state machine cannot indicate any uncertainty with regard to its own state depending on the reliability of its input data.

From the DE 102 54 806 A1 It is known to combine environmental data from a number of sensors and map data from a digital road map by means of a fusion of information in order to obtain information about the road class currently being traveled on. The disadvantage of such a merger is that a clear result from one information source can be put into perspective or weakened by uncertain results from another information source.

From the DE 10 2018 208 593 A1 a method is known for checking whether a driving mode of the motor vehicle can be changed safely. In two consecutive test stages, it is first checked whether, according to the map data of a digital road map, there is even a permissible road type for activating the driving mode somewhere in the vicinity, and then, if the first test stage is passed successfully, a second test stage based on environmental data Sensors to detect whether the motor vehicle is actually on a road of this road class. The check only takes place for the section of road currently being traveled on, which is why the activation of the driving mode is preferably checked by means of the method, but not the deactivation when leaving a permissible road class, ie when exiting the ODD.

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 is present for which the operation of the automated driving function is permitted.

The object is solved by the subject matter of the independent patent claims. Advantageous developments result from technical Features described in the dependent claims, in the following description and in the figures.

As a solution, the invention includes a method for monitoring a permission condition for an automated driving mode of a motor vehicle by a control circuit. The admissibility condition includes that the driving mode may only be active when driving on roads of at least one predetermined authorized road class. To check a current road class of a road segment ahead in the direction of travel of a road currently being traveled on, in a first signal path from a digital road map to the road segment, map data which indicate the road class of the road segment are read from a data memory and then checked whether they indicate such a road class, which corresponds to the at least one road class permitted for the driving mode. The admissibility condition is checked in particular to check, in the case of an already activated driving mode, whether the driving mode has to be deactivated because the road segment ahead no longer corresponds to the at least one permitted road class. In a manner known per se, driver assistance for driving on the freeway, an ACC (automatic cruise control), an automatic distance control system, an overtaking assistant, for example, can be provided as a driving mode, to name just a few examples. In general, the automated driving mode can provide for longitudinal guidance (acceleration and/or braking) and/or lateral guidance (steering) to be carried out automatically in the activated driving mode by at least one control unit of the motor vehicle without any action on the part of the driver.

The road class can also be checked for the road section or road segment currently being traveled on, but according to the invention we primarily use 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, for example, in 30 meters ends up to 1 kilometer in front of the motor vehicle. The road map can be used to check whether the map data stored therein relating to this road segment ahead indicates that this road segment is part of or classified as an approved road class, ie as an element of the set of at least one approved road class. One or more than one road class can be defined as permissible, which is why the wording “at least one predetermined permissible 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.

However, the use of a road map can result in outdated map data being available and/or localization of the motor vehicle in relation to the road map being subject to a variance or tolerance or fluctuation, as is inherent to receivers of a position signal from a GNSS (Global Navigation Satellite System ), such as the GPS (Global Positioning System), is known.

Independent of the first signal path, in a second signal path that is independent of the first signal path, environmental data are received from an environmental sensor or multiple environmental sensors, which describe environmental features of the road segment (e.g. the presence of street signs or recognizable people), and based on the environmental data, characteristics of the road segment are determined using a predetermined estimation routine street segment (e.g. the characteristic "blue motorway signs" or "persons present").

It is detected whether the estimated characteristics of an existence probability of a presence of the at least one permitted road class in the road segment ahead is greater than a predetermined threshold value. The surroundings data can be the sensor data from the at least one surroundings sensor and/or processed sensor data, for example filtered sensor data and/or sensor data combined by means of a sensor fusion. The characteristic recognized in each case can include, for example: “Pedestrians present”, “Center barrier present”, “Motorway signs of a specific color present”, to name just examples. A corresponding estimation routine to recognize such characteristics from environmental data can be provided, for example, on the basis of an algorithm for machine learning, for example an artificial neural network, as is known per se from the prior art for object recognition or machine vision . The probability that the characteristics indicate the presence of a special road class or at least the presence of the set of permitted road classes can then be based on the estimation results for the individual characteristics, for example a key figure for the confidence or estimation variance, when estimating the respective characteristic . This is explained in more detail below. By combining estimated characteristics, for example the presence of a crash barrier and/or the presence of pedestrians, the individual key figures then result of the characteristics, i.e. their estimation reliability, the probability of the existence of the at least one approved road class. The second signal path thus 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 membership signal is issued if the existence probability is greater than the threshold value. Otherwise it is signaled that the road segment ahead (or a partial interval thereof) does not belong to the at least one permitted road class.

Said actual checking of the admissibility condition includes that the presence of the at least one approved road class for the road segment ahead is finally only signaled if both the first signal path and the second signal path confirm or signal the existence of the road class, and otherwise a predetermined blocking measure for the driving mode is carried out. In other words, the two signal paths are evaluated independently of one another and only if both signal paths independently of one another signal or confirm the presence or existence of the at least one permitted road class in the area of the road segment ahead is it assumed for the driving mode that the admissibility condition is met. Otherwise the blocking measure is carried out, which will be explained in more detail later.

This means that one is not dependent on the signal paths being implemented absolutely error-free according to ASIL-D, but only the mutual comparison of both signal paths must lead to the ASIL-D level.

The two signal paths can be implemented, for example, as software modules operated independently of one another, as a result of which they can check the road class independently of one another.

The invention also includes developments that result in additional advantages.

It should be noted here: When estimating a characteristic, three aspects can be checked, namely whether the characteristic is present ("presence") at all (a classifier, such as an artificial neural network, responds to the presence of the characteristic at all, e.g. on "persons" in camera images), a "confidence of existence" of the characteristic (the classifier signals the probability with which the characteristic is present or how reliable the estimate is) and a "state" of the characteristic (for example how many kilometers the characteristic has been present or which expression is present, such as "many pedestrians present" in contrast to "few pedestrians present"). Corresponding estimation routines are available per se from the prior art, for example in connection with monitoring the surroundings in a motor vehicle for operating an autopilot.

A development includes that the road segment ahead is divided in the direction of travel into several different distance intervals one behind the other in the direction of travel and a separate check of the admissibility condition is carried out using the two signal paths for each distance interval. The map data and the characteristics from the respective distance interval are therefore used for each distance interval. In other words, the road segment ahead is not monitored as a uniform area over which the probability of the existence of an approved road class is checked overall, but sections or partial intervals with different distances to the motor vehicle (hence "distance intervals") are individually checked to determine whether they meet the eligibility condition. This has the advantage that indicators or characteristics are prevented from being "averaged out" over the entire road segment. For example, if the road is no longer part of the at least one permitted road class in the furthest distance interval, corresponding characteristics that clearly signal this can be found in this furthest distance interval. For example, if the motor vehicle is driving on a freeway and changes to the turning lane, the admissibility condition "freeway available" can still be met for the distance interval immediately ahead, while at the end of the exit in a curve to leave the freeway the characteristic "curve with a curve radius smaller than X Meter" may be present, which signals that the "Autobahn" road class is no longer available here. Under certain circumstances, however, this unique characteristic would be averaged out in a calculation of the probability of existence for the “Autobahn” road class over the entire road segment ahead, because the characteristics for the “Autobahn” road class predominate in the preceding distance intervals. If, on the other hand, the probability of the existence of the road class "Autobahn" is calculated separately for the distance intervals, the characteristic relating to the curve radius can be reliably recognized for the last, most distant distance interval and signaled that in There is no longer a motorway or, in general, the permissible road class in this distance interval. The road segment is preferably divided into at least two distance intervals, preferably more than two, preferably more than three, in particular more than four distance intervals. 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 each characteristic, for example distance intervals with a length of 50 meters, distance intervals with a length of 100 meters, to name just a few examples.

A further development includes that in the event that a distance interval closest to the motor vehicle meets the admissibility condition (i.e. the driving mode may still be active) and at least one distance interval behind it violates the admissibility condition (i.e. the driving mode should be deactivated until this upcoming distance interval is reached ), The blocking measure includes a transfer routine for transferring a longitudinal guidance and / or lateral guidance of the motor vehicle to a driver of the motor vehicle. In other words, it is recognized for a future point in 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 carried out as a blocking measure for blocking or deactivating the driving mode. This handover routine is known per se from the prior art and can include, for example, an indication or an output to the driver that he should get ready to take control of the motor vehicle. In addition or as an alternative to this, an emergency stop can be used to ensure that the driving mode does not have to be operated in the area of a non-approved or impermissible road class if, for example, the takeover procedure or takeover routine for the driver fails.

A development includes that said blocking measure includes that in the event that the admissibility condition is violated in the road segment ahead, activation of the driving mode is blocked when the driving mode is deactivated and/or handover to a driver and/or an emergency stop when the driving mode is activated carried out and then the driving mode is deactivated. In general, 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 hidden (in the case of a touch screen). When driving mode is activated, driver handover and/or emergency stop are forced.

A further development includes that at least in the second signal path (environment 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 probability of existence for each of the road classes is calculated in parallel. A memory function for at least one of the road classes or all of the road classes preferably also takes into account past road segments that have already been traveled through when calculating their probability of existence, with this memory function taking into account a reduction behavior of the characteristics estimated in these past road segments (ie a forgetting factor) as a function of a temporal and/or or local distance to the respective past road segment. In other words, for each location on the road, there is an estimate or calculation of the probability of existence for each class of road. An existence probability with a value greater than said threshold value should then be signaled or determined for at least one road class, preferably for exactly one road class. However, since an estimation of the probability of existence is carried out in parallel or simultaneously for all road classes, a previous or past road segment or the characteristics estimated therein can also be taken into account in the calculation of the probability of existence and, for example, accumulated or taken into account by means of a regression. If, for example, a central barrier begins in a road segment and if the motor vehicle passes this start of the central barrier and the central barrier continues along the road, then, for example, after the motor vehicle has covered a distance of a predetermined minimum length, for example 1 kilometer, the characteristic “central barrier present ' can be taken into account not only for the current road segment, but also for past road segments or places on the road that have already been passed or driven through using the memory function. However, a characteristic loses its meaning or its informative value when estimating the probability of existence of the road class of the current road segment, which is why characteristics from the past in terms of 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 is further back than a time value in a range of, for example, 10 seconds to 15 minutes. Additionally or alternatively, it can be determined, for example by means of the described localization of the motor vehicle, how large a local distance of the motor vehicle has become from the observation point for a characteristic, with a distance in a range of more than 20 meters being the value for forgetting or hiding a characteristic up to more than 5 kilometers can be chosen, just to name examples. So if the indications or characteristics for a certain road class "multiply" or repeat or accumulate or aggregate during the journey, this can be recognized by means of the memory function and taken into account when calculating the probability of existence for a road class that has not yet been recognized.

A development includes that for each road class checked, for each characteristic, an overall index is calculated on the basis of measured classification uncertainties and/or existential uncertainties and/or status uncertainties when sensing the corresponding environmental features using the environment data, with the overall index signaling a degree of the extent of this characteristic, where, for example, a higher absolute value of the overall key figure means a stronger expression. In order to determine the probability of existence from the observed characteristics, each characteristic can be quantized by a single total number. With this, the already described classification uncertainty and/or uncertainty of existence and/or uncertainty of state 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) be summarized. This overall index describes how reliably the respective characteristic could be derived from the environmental data from the at least one sensor. Here, for example, an output of an algorithm for machine learning, for example an output of an artificial neural network, can be accessed.

A development includes that for each road class checked from the characteristics observed overall in the road segment, a class-specific selection and/or a class-specific weighting is carried out according to relevance data stored in the control circuit, which indicate a relevance of the characteristics with regard to the respective road class. It can thus be prevented that an estimation or calculation of an existence probability for a road class is made uncertain by considering a characteristic that is irrelevant for the calculation of the existence probability, for which there is, for example, a large observation uncertainty or estimation uncertainty and/or no reliable indication of a Road class is because it is typical of both a permitted and a prohibited road class. It is thus taken into account that there can also be characteristics that are typical or characteristic of both an approved road class and an impermissible road class, which is why this characteristic can be hidden or ignored as irrelevant for the differentiation between permissible and impermissible road classes. The categories for the selection can be specified, for example: absolutely necessary, conditionally necessary, optional. An absolutely necessary characteristic can indicate that without an observation or estimation of this characteristic, the corresponding road class can also not be present or the characteristic must be missing the other way around so that the road class can be present. A conditionally necessary characteristic states that this can, for example, currently be covered, i.e. does not have to be observed for the current road segment, but that the memory function should be used, for example, to signal that this characteristic was at least temporarily within a past period of time, for example within the last ten minutes. or must have been present within a predetermined distance, for example within the past kilometer.

A further development includes that 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), on the basis of which a partially autonomous or fully autonomous driving function plans a driving trajectory in the automatic driving mode and selects a behavioral rule to be followed and/or a traffic rule to be observed depending on the road class entered in the environment model. Thus, the checking of the admissibility condition is also used to control the road class signaled by both signal paths for determining a driving behavior of the motor vehicle. For example, a rule of conduct to be followed may state that overtaking is preferred or, conversely, that overtaking is to be avoided. A traffic rule to be observed may be the traffic rule intended for the road class implicitly given by the road class (such as a maximum speed or maximum speed for country roads) and/or a driving style, eg the likelihood of an overtaking attempt. Thus, the signal paths are also used for the information in the environment model mation of the currently available road class.

As a further solution, the invention comprises a control circuit for a motor vehicle, the control circuit having a processor device which is set up to carry out an embodiment of the method according to the invention. The control circuit can have a data processing device or a processor device that is set up 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 which is set up 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 designed as a control unit or as a combination of several control units for the motor vehicle. The control circuit can additionally or alternatively comprise a central computer for a motor vehicle.

As a further solution, the invention includes a motor vehicle with at least one sensor for generating sensor-based surroundings 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.

The invention also includes the combinations of features of the described embodiments. The invention also includes implementations that each have a combination of the features of several of the described embodiments, unless the embodiments were described as mutually exclusive.

• 1 eine schematische Darstellung einer Ausführungsform des erfindungsgemäßen Kraftfahrzeugs mit einer Ausführungsform der erfindungsgemäßen Steuerschaltung, durch welche eine Ausführungsform des erfindungsgemäßen Verfahrens durchgeführt werden kann; und
• 2 eine Skizze zur Veranschaulichung einer Berechnung einer Existenzwahrscheinlichkeit einer zulässigen Straßenklasse.

Exemplary embodiments of the invention are described below. For this shows:

• 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, through which an embodiment of the method according to the invention can be carried out; and
• 2 a sketch to illustrate a calculation of an existence probability of a permissible road class.

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 are to be considered independently of one another and that each also develop the invention independently of one another. Therefore, the disclosure is also intended to encompass combinations of the features of the embodiments other than those illustrated. Furthermore, the described embodiments can also be supplemented by further features of the invention that have already been described.

In the figures, the same reference symbols designate elements with the same function.

1 shows a motor vehicle 10, which can be a motor vehicle, in particular a passenger car or truck. An automated driving function 11, for example an autopilot, can be provided in motor vehicle 10, which can control a drive motor 12 and/or brakes 13 in motor vehicle 10, for example, for longitudinal guidance and a steering system 14 for lateral guidance. A longitudinal guide or a transverse guide can also be provided, for example. The autopilot 11 is active when a driving mode F is activated. Provision can be made here for the driving function 11 to be permitted only for predetermined road classes or at least one predetermined road class, ie may only be active on at least one permitted road class, for example a motorway or a country road. The motor vehicle 10 can currently be moving on a road 16 along a direction of travel 15 . For road 16, it can be checked or established in motor vehicle 10 whether a road segment 17 ahead of motor vehicle 10 in direction of travel 15 corresponds to the at least one approved or permitted road class. For this purpose, a control circuit 18 can be provided in motor vehicle 10, which can be implemented, for example, by a control unit or a combination of a plurality of control units and/or a central computer of motor vehicle 10. Two 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, which indicates that the road segment 17 ahead corresponds to the at least one authorized road class. If this admissibility condition 21 is violated, a predetermined degree of blocking 22 are triggered by the control circuit 18. This can then deactivate driving mode F in driving function 11 or prevent its activation.

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

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 crash barrier present”, “pedestrian presence”, have been estimated from the environmental data 33 and an overall index G has been determined for each characteristic 38 . It is also shown that the overall index G within the road segment 17 can be calculated independently or individually for each characteristic 38 in different distance intervals 50 . A separate length of the distance intervals 50 along the direction of travel 15 can be defined or provided for each characteristic 38 . 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. Thus, for different road classes K, represented here by four different road classes K1, K2, K3, K4, the probability of existence E, 41 can be calculated and in this case using the relevance data 39 (represented by dashed lines) for determining the probability of existence E of a respective road class K1, K2, K3, K4 different overall key figures G are taken as a basis or combined. An estimated road class 31 can then be estimated or signaled for each resulting distance interval, for example the road class K1 to K4 with the greatest probability of existence E or with an additional condition that the probability of existence E must be greater than a predetermined minimum value. In 2 it is illustrated 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 (in each case symbolized by different hatching). The admissibility condition 21 is thus violated within the road segment 17 due to the distance interval A2. Correspondingly, the blocking measure 22 is triggered, which, based on the still available distance interval 50 with a permissible road class, can carry out or trigger a takeover procedure for handing over the driving task to the driver, for example.

In order to meet the safety requirements for identifying a road class in automated driving, a decomposition according to ISO 26262 into two redundant ASIL B (D) road classifications is preferably carried out. On the one hand, roads are classified by localization on an HD map (digital road map).

On the other hand, there is a classification with the help of environmental features that are detected by the vehicle's own environmental sensors, which represents the novelty. The aim of this new method is therefore 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.

In order to meet the safety requirements for road class detection, a decomposition according to ISO 26262 into two redundant ASIL B (D) road classifications is carried out. On the one hand, roads are classified by localization on an HD map. On the other hand, a classification is carried out with the help of environmental features that are detected by the vehicle's own environmental sensors. The result of this classification is a distribution over the road classes defined in the function, where the reliability is taken into account in the level of the probability value of each road class. An environment feature is abstract information in the environment of the ego vehicle, which is recorded by various environment sensors (e.g. radar, lidar and camera), processed and then combined by a sensor data fusion. Important environmental features are e.g. lane markings, traffic signs, crash barriers, delineators, walls, vehicles and living beings. In addition to the geometric position or the course in the vehicle's own coordinate system, they all have other states such as vehicle type, speeds and accelerations for vehicles; Color, dash type and width for lane markings and plate type and content for traffic signs. To determine the road class, the states of all environmental features perceived in the area are analyzed. A comparison is then made 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 differ significantly in different countries or can change during the life cycle of the product, they are stored in a database and kept up to date via updates. On the one hand, the existing laws and guidelines of the respective country of use serve as a starting point for the definition of the characteristics of different road classes. In Germany, for example, the StVO, VwV-StVO or the various guidelines for the construction of roads should be mentioned. These formulate guidelines for the alignment and cross-sections of lane markings, the type, size and position of traffic signs or the course and arrangement of crash barriers. There are analogous sets of rules for Chinese and American road construction. On the other hand, characteristics can be determined by data-driven, statistical analyses. For example, the appearance or frequency of certain traffic signs in the various road classes, the existence and behavioral patterns of other road users or the road topology should be mentioned. As a result, there are both characteristics that clearly allow conclusions to be drawn about the road class and those that only provide an indication in the form of an increased probability for one or more road classes. The aim of the method is therefore 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 features used preferably correspond to safety integrity ASIL-B (D). The environment model is enriched with the estimated road class.

The advantage of the method is that the reliability of the input data has a direct impact on the probability value of a road class. If the sensor measurement data is poor, this can be taken into account when deciding whether an automated driving function is available. In addition, 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 for a specific road class may also be sufficient to offer the driving function. The redundancy of road class recognition using HD map information enables greater safety integrity and reliability. Despite possible localization errors or errors in the HD map (digital road map), the road class can be determined by the onboard sensors. Furthermore, the decomposition of the road class detection into two redundant methods reduces the technical complexity and the security requirements for its implementation. Because the information from different sensor principles (radar, lidar, camera) is taken into account in the estimation, increased reliability can also result, since, for example, a purely camera-based recognition of the road class would be significantly more error-prone. In addition to activating and deactivating the system, the road class is also used for the model compliance with traffic regulations, which increases reliability here as well.

The existing environmental features are recorded by the vehicle's own sensors and brought together by a sensor data fusion. At the end of this fusion there are various properties of the environmental features, such as classification, existence measure, pose, kinematics and geometric course. This data is used to generate an environment model of the ego vehicle. The characteristics to be checked are extracted from this environment model by the states and relationships of the environment features. The characteristics are high-level indicators for properties that the environment model has. For example, the characteristic "pedestrians present" is determined from individual pedestrians in the environment 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 status uncertainties of the corresponding environmental characteristics. The overall key figure provides information about the extent of this characteristic, with a higher absolute value of the number meaning a stronger expression. The characteristics are updated in each calculation step, taking into account the previous results. In the course of time, the various characteristics can show an increase or decrease in behavior in order to stabilize the behavior and to take into account individual events that only have a limited validity in time. A location-dependent component can also be taken into account for characteristics that are determined from location-specific environmental features. For example, the distance to a specific environmental feature must exceed at least a specific distance so that the degradation behavior over time of the characteristic is applied.

The probability-based estimation of a road class is made by combining the characteristics relevant to this class. The form in which the characteristics are combined to estimate a road class is not fixed, but is queried via the database. What is variable is which characteristics are used to estimate a road class, as well as their influence and weighting on the result of the estimation. In addition, the characteristics relevant for a road class are divided into the categories absolutely necessary, conditionally necessary and optional. Characteristics that are absolutely necessary must have a minimum specification so that the road class can be output as existing. The same applies to conditionally necessary characteristics, with the difference that there is a condition that allows the road class to still be output as present. Optional characteristics contribute to the plausibility check of a road class, but have no direct influence on whether a road class can be output as existing or not. All defined road classes are estimated in parallel and independently of each other, so that at the end of a processing step there is an estimated value for each defined road class. The probabilities of existence of the characteristics are also evaluated as a function of the distance in the longitudinal direction of the roadway from the ego vehicle. This results in distance intervals for each characteristic, in which the expression of the characteristic and/or the probability of existence can differ. The intervals of all characteristics are combined and an estimate of the individual road classes is made for each resulting interval. In this way, pending changes in the road class can also be detected. At the end of the day, the combination of the individual estimates results in a distribution of the road classes with their determined probability values.

• - Existenz und Eigenschaften von Ampeln (Einschaltzustand, Lichtzeichenfarbe, Lichtzeichenform, Position, Anzahl)
• - Existenz von Querverkehr
• - Existenz von Gegenverkehr
• - Existenz und Beschaffenheit einer Trennung zum Gegenverkehr (z.B. Leitplanke,
• - Betonbarriere, breiter Grünstreifen)
• - Vorhandene Kurvenradien (über Fahrstreifenmarkierung oder Randbebauungen)
• - Existenz von Kreuzungen
• - Existenz von Kreisverkehren
• - Existenz von Auf-/Abfahrten
• - Existenz und Eigenschaften von Verkehrsschildern (Typ, Größe, Farbe, Position)
• - Existenz und Eigenschaften von Fußgängern / Radfahrern (Anzahl, Häufigkeit, Position, Bewegungsrichtung)
• - Breite der Fahrstreifen
• - Eigenschaften der Fahrstreifenmarkierungen (Breite, Strichelungstyp, Verlauf, Reflektivität (kann durch Lidarsensor gemessen werden))
• - Steigung/Gefälle (längs und quer der Fahrbahn)
• - Anzahl der Fahrstreifen
• - Existenz eines Seitenstreifens
• - Existenz und Position von Straßenbeleuchtung
• - Durchschnittsgeschwindigkeiten der Fahrzeuge pro Fahrstreifen
• - Abstände zwischen den Fahrzeugen (längs und quer)
• - Existenz bzw. Häufigkeit von Fahrzeugtypen
• - Maximalgeschwindigkeiten der Fahrzeuge
• - Referenzen
• - technische Regelwerke für das Straßenwesen in Deutschland
• - Richtlinien für die Anlage von Autobahnen

The following characteristics are taken into account when estimating the road class:

• - Existence and properties of traffic lights (switch-on status, light signal color, light signal shape, position, number)
• - Existence of cross traffic
• - Existence of oncoming traffic
• - Existence and nature of a separation from oncoming traffic (e.g. crash barrier,
• - concrete barrier, wide green strip)
• - Existing curve radii (via lane markings or edge buildings)
• - Existence of intersections
• - Existence of roundabouts
• - Existence of on/off ramps
• - Existence and properties of road signs (type, size, color, position)
• - Existence and characteristics of pedestrians / cyclists (number, frequency, position, direction of movement)
• - Width of lanes
• - Lane marking properties (width, dash type, gradient, reflectivity (can be measured by lidar sensor))
• - Uphill/downhill (along and across the road)
• - Number of lanes
• - Existence of a hard shoulder
• - Existence and position of street lighting
• - Average vehicle speeds per lane
• - Distances between vehicles (longitudinal and transverse)
• - Existence or frequency of vehicle types
• - Maximum vehicle speeds
• - Credentials
• - technical regulations for the road system in Germany
• - Guidelines for the construction of motorways

Overall, the examples show how an estimate of a road class can be provided on the basis of sensor-based environment data.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• DE 102012218362 A1 [0005]
• DE 10254806 A1 [0006]
• DE 102018208593 A1 [0007]

Claims (10)

Method for monitoring a permission condition (21) for an automated driving mode of a motor vehicle (10) by a control circuit (18), wherein the permission condition (21) comprises that the driving mode only when driving on roads (16) of at least a predetermined approved road class ( 29, 31) may be active, 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 on in a first signal path (19): from a digital road map ( 27) map data for the road segment (17), which indicate the road class (29, 31) of the road segment (17), are read from a data memory and checked to see whether they indicate the at least one permitted road class (29, 31), characterized in that that independently of the first signal path (19) in a second signal path (20) operated independently of the first signal path (19) from an environment sensor (3 2) environmental data (33) describing environmental characteristics of the road segment (17) are received by one or more environmental sensors (32) and characteristics (38) of the road segment (17) are estimated and detected using the environmental data (33) by means of an estimation routine (35), whether the estimated characteristics (38) in the road segment (17) ahead signal an existence 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 that the presence of the at least one permitted road class (29, 31) for the road segment (17) ahead is only signaled if both the first signal path (19) and the second signal path (20) indicate the existence of the road class (29, 31) confirm, and otherwise a predetermined blocking action (22) for the driving mode is performed. procedure after claim 1 , wherein the road segment (17) ahead is subdivided in the direction of travel (15) 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 using the two signal paths, wherein the map data and the characteristics (38) within the respective distance interval (50) are used for each distance interval. procedure after 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 thereafter violates the admissibility condition (21), the blocking measure (22) includes a transfer routine for transferring longitudinal guidance and/or lateral guidance of the Motor vehicle (10) to a driver of the motor vehicle (10). Method according to one of the preceding claims, wherein the blocking measure (22) comprises that in the event that in the road segment (17) ahead the admissibility condition (21) is violated, activation of the driving mode is blocked when the driving mode is deactivated and/or when it is activated Driving mode carried out a handover to a driver and / or an emergency stop and then the driving mode is deactivated. Method according to one of the preceding claims, the current probability of existence (30, 41) being calculated in parallel 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). and for at least one of the road classes (29, 31) to calculate its probability of existence (30, 41), a memory function takes into account past road segments (17) that have already been traveled through, the memory function taking into account a degradation behavior of the characteristics estimated in these past road segments (17). (38) as a function of a temporal and/or spatial distance from the respective previous road segment (17). Method according to one of the preceding claims, wherein for each checked road class (29, 31) for each characteristic (38) an overall index based on measured classification uncertainties and/or existence uncertainties and/or status uncertainties when sensing the corresponding environmental features using the environmental data (33) is calculated and the overall index indicates an expression of this characteristic (38). 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 according to relevance data (39) stored in the control circuit (18) which indicate a relevance of the Specify characteristics (38) with respect to the respective road class (29, 31), is carried out. 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 in an environment model based on which a partially autonomous or fully autonomous driving function (11) plans a driving trajectory in the automatic driving mode and 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. Control circuit (18) for a motor vehicle (10), the control circuit (18) having a processor device which is set up to carry out a method according to one of the preceding claims. Motor vehicle (10) with at least one sensor for generating sensor-based environment data (33) and with a control circuit (18). claim 9 .

Images