System and method for learning a following distance for automatic longitudinal guidance of a vehicle

Abstract

System (1) for learning a following distance for an automatic longitudinal guidance of a vehicle is designed to adapt a normal operation following distance (N) to a target object driving ahead of the vehicle, which is set for a normal operation of the automatic longitudinal guidance, as a function of a detected override of the vehicle driver of the automatic longitudinal guidance, so that a changed normal operation following distance (N) is set for a future following drive in the normal operation of the automatic longitudinal guidance. The system is also designed to set a learning operation following distance (L) to the target object driving ahead, which is increased compared to the normal operation following distance (N), as a function of a situation, so that the driver has more opportunities to set a following distance, which is reduced compared to the learning operation following distance (L), desired by the driver, by overriding the automatic longitudinal guidance.

Description

Technical Field

[0001] This invention relates to a system for learning following distances for automatic longitudinal guidance of a vehicle, and to a vehicle equipped with such a system. Furthermore, this invention relates to a computer-implemented method for learning following distances for automatic longitudinal guidance of a vehicle, a processing apparatus, a computer program for performing this method, and a computer-readable storage medium on which the computer program is stored. Background Technology

[0002] Many modern vehicles are equipped with autonomous driving functions, particularly those enabling automatic longitudinal guidance. A widely used example of an autonomous driving function is automatic distance-speed adjustment, often referred to as distance-adjustable speed control or ACC (adaptive cruise control). Here, if the vehicle ahead is traveling slower than a target speed set for the vehicle, the following distance is adjusted to the target distance using automatic engine and braking intervention. The target following distance can be, for example, a specific second interval; that is, it can be preset to a specific time interval in which a vehicle equipped with ACC should follow the vehicle ahead.

[0003] Because different drivers have different following distance preferences, drivers can usually set a target following distance within certain limits. For example, it could be stipulated that drivers can globally set the following distance across a certain number of distance levels, with corresponding distance-speed characteristic curves stored for each level. However, these predefined distance levels only partially reflect individual needs and habits. While they reflect the average of ordinary drivers, they do not offer the possibility of adapting to specific driver preferences.

[0004] In conventional systems, the preset distance-speed characteristic curve follows a fixed pattern. At low speeds, the distance per second is typically high. The higher the speed, the smaller the corresponding distance. However, this doesn't necessarily correspond to the driver's desired behavior. In congested city traffic, drivers might, for example, prefer tighter driving and choose larger distances on highways and rural roads to easily reach their destination. These preferences are not reflected in the conventional distance levels because they don't correspond to the average driver. By limiting the settling of only a few distance levels, individual driver preferences can only be considered to a limited extent.

[0005] In the prior art, several solutions are already known to learn desired distances from the driver’s driving behavior and to take these desired distances into account in automated longitudinal guidance.

[0006] For example, one proposed approach is to learn individual driving behavior related to following distance using neural networks (see, for example, HUANG, Xiuling; SUN, Jie; SUN, Jian, A car-following model considering asymmetric driving behavior based on long short-term memory neural networks. Traffic Research Part C: Emerging Technologies, 2018, Vol. 95, pp. 346-362).

[0007] DE 10 2015 016993 A1 describes a method for acquiring a driving profile for a vehicle's driver assistance system, comprising the following steps: in an inactive driver assistance system, recording distance measurement signals from the vehicle's distance measurement system and driving dynamic measurement signals of the vehicle; storing the recorded distance measurement signals and driving dynamic measurement signals in a storage device; obtaining driving behavior from the stored distance measurement signals and driving dynamic measurement signals, wherein the distance and / or time interval to the vehicle ahead are obtained from the distance measurement signals; and storing the acquired driving behavior. Furthermore, a driver assistance adjustment method is proposed, particularly an ACC adjustment method using this method, and a driver assistance system for executing the driver assistance adjustment method.

[0008] Conventional solutions, such as those mentioned in the paragraph above, only allow for training drivers in manual driving when distance adjustment is disabled. However, in Germany, for example, in vehicles equipped with ACC, more than 30% of the mileage has already been traversed using activated distance adjustment, making these conventional solutions unsuitable for drivers' distance preferences on these significant proportions of the routes.

[0009] However, among users who frequently use the ACC system and are therefore most likely to evaluate and question the automatic learning of distance behavior, there is a need to effectively adapt to the expected distance behavior. Therefore, it is desirable to adapt the driver's presets even during the phases when distance adjustment is activated.

[0010] A practical example is a driver who prefers to use ACC (Adaptive Cruise Control) on rural roads and highways, but prefers manual control in urban scenarios due to the high dynamics of driving conditions. If the ACC system uses different distance-speed characteristic curves for urban and highway scenarios, then the urban characteristic curve is fully trained due to the high proportion of manual driving. However, if the driver immediately activates the distance-speed cruise control system after entering the highway, then the standard characteristic curve for highways, or a characteristic curve that is only minimally adapted to the standard characteristic curve due to the low proportion of manual driving on highways, is always used.

[0011] Therefore, a technological solution is needed that can adapt to the driver's habits during activated driving assistance, especially on highways or highway-like roads. Summary of the Invention

[0012] The objective of this invention is to describe a computer-implemented method for effectively learning following distance based on driver behavior detected during activated distance adjustment, and a corresponding system capable of performing this method.

[0013] This task is addressed by the subject matter of the independent claims. Advantageous embodiments are described in the dependent claims.

[0014] The first aspect of the invention relates to a system for learning following distances for automatic longitudinal guidance of a vehicle.

[0015] Vehicles can be motor vehicles, in particular. Here, the term "motor vehicle" should be understood specifically as a land vehicle that moves by mechanical force and is not connected to rails. In this sense, a motor vehicle can be constructed, for example, as a passenger car, a motorcycle, or a tractor-trailer.

[0016] Automatic longitudinal guidance of a vehicle can be implemented within the scope of autonomous driving functions, such as ACC (Adaptive Cruise Control) or more comprehensive functions, such as supporting a combination of longitudinal and lateral guidance.

[0017] Within the scope of this document, the term "autonomous driving function" generally refers to a vehicle function capable of achieving autonomous driving. Here, the term "autonomous driving function" is understood as driving with automatic longitudinal and / or lateral guidance. This may, for example, involve longer periods of driving on highways or limited time driving within a parking area. The term "autonomous driving" includes autonomous driving with any degree of automation. Exemplary levels of automation are Driver Assistance, Semi-Autonomous Driving, Conditional Automated Driving, High Automated Driving, and Full Automated Driving (each with an increasing degree of automation). The five levels of automation described above correspond to SAE Levels 1 through 5 according to the SAE J3016 (SAE - Society of Automotive Engineers) standard as of April 30, 2021. In Driver Assistance (SAE Level 1), the system performs longitudinal or lateral guidance in specific driving situations, expecting the driver to assume all remaining aspects of the dynamic driving task. In Semi-Autonomous Driving (SAE Level 2), the system assumes longitudinal and lateral guidance in specific driving situations, where, as in Driver Assistance, the driver must continuously monitor the system. In Conditional Automated Driving (SAE Level 3), the system provides longitudinal and lateral guidance under specific driving conditions, without requiring continuous monitoring by the driver; however, the driver must be able to take over driving for a certain period as required by the system. In Highly Automated Driving (SAE Level 4), the system takes over vehicle guidance under specific driving conditions, eliminating the need for a driver as a backup, even when the driver does not respond to intervention requests. In Fully Automated Driving (SAE Level 5), the system performs all aspects of dynamic driving tasks that are still under human driver control, regardless of road and environmental conditions.

[0018] The system design according to the invention is used to adapt the normal following distance for normal operation of automatic longitudinal guidance to a target object (especially a vehicle in front of the vehicle) traveling in front of the vehicle, based on overtaking detected by the vehicle driver during the learning phase of automatic longitudinal guidance, such that the normal following distance is adjusted for future following during normal operation of automatic longitudinal guidance.

[0019] Therefore, the system can learn the normal operating following distance in such a way that it stores, for example, a new normal operating following distance for use in subsequent following maneuvers during normal operation, wherein the new normal operating following distance is based on a change in the driver's detected overtaking behavior relative to the previous normal operating following distance. Here, the previous normal operating following distance itself may also be an (intermediate) result of this learning process. Alternatively, the previous normal operating following distance may, for example, be a preset standard following distance.

[0020] For example, the normal following distance can be speed-specific. The normal following distance can be obtained, in particular, from a preset or learned distance-speed characteristic curve (hereinafter referred to as the "characteristic curve"), and, for example, is assigned to the vehicle's current speed based on this characteristic curve as the target following distance. Here, the learning of the driver's distance behavior is not limited to a single normal following distance, but can be performed generally on this characteristic curve. Details regarding the training of the distance-speed characteristic curve are further explained below in conjunction with possible embodiments of the system and method according to the invention.

[0021] Overtaking in automatic longitudinal guidance can particularly include shortening the following distance through driver accelerator pedal manipulation. Even when distance adjustment is activated, the driver can shorten the distance to the vehicle ahead as desired by using the accelerator pedal. This is known as overtaking (übertreten), where the driver assumes the longitudinal dynamic requirements of distance adjustment. The changed (new) normal operating following distance can in particular be a reduced distance compared to the previous normal operating following distance. Therefore, the changed normal operating following distance can take into account the driver's preference for a smaller following distance, expressed by the driver's active shortening of the following distance, in future following maneuvers (and possibly already in further directions of the current following maneuver). The driver-desired distance behavior of automatic longitudinal guidance can be learned progressively and better in this way.

[0022] The system is also designed to set a learning operating following distance for a target object ahead, increased compared to the normal following distance, based on the context. This learning operating following distance is named as such because it should be able to effectively adapt the normal following distance to the driver's preferences based on more detected overdrive interventions.

[0023] This invention is based on the idea that a driver actively adapting their following distance by using the accelerator pedal only results in a decrease in the stored normal following distance. Therefore, without the solution according to the invention, drivers with a greater distance expectation (compared to the previous normal following distance) typically have to intervene by braking to deactivate the ACC system in order to increase the following distance. Thus, only drivers with a smaller distance expectation can actively provide the motivation to train their distance behavior.

[0024] By proposing here to increase the initial following distance to the learning operating following distance, a larger number of drivers can set their corresponding desired distances, and thus, the automatic longitudinal guidance can be trained to their preferred distance behavior. In particular, the system can intentionally set a learning operating following distance larger than the normal operating following distance set for normal operation, at least initially—that is, directly after the distance adjuster is activated or a following situation occurs—in order to subsequently detect whether and, if necessary, to what extent the driver reduces the following distance by manipulating the accelerator pedal.

[0025] Therefore, by increasing the learning running following distance, distance behavior of automatic longitudinal guidance can be effectively trained even when the distance adjuster is activated, because the driver has more opportunities to set the desired following distance, compared to the learning running following distance, through overtaking of the automatic longitudinal guidance. This means that, compared to the normal operating following distance, (at least in the statistical average across all driver preferences) there is a more frequent expectation of shortening the following distance when the learning running following distance is increased. Therefore, more overtaking intervention can serve as a data basis for training the normal operating following distance.

[0026] According to the implementation method, the description of "setting the learning operating following distance according to the situation" can be understood in particular as follows: the system must first determine, as a prerequisite for setting the learning operating following distance, rather than the smaller normal operating following distance, that at least one switching condition is met. The switching condition is so named because it can be understood as the condition used to switch between the normal operating following distance and the learning operating following distance, and is understood as the actual following distance to be set. As further described below with respect to the implementation variations of the method presented herein, such a switching condition may, for example, relate to the type of road currently being traveled. For example, the switching condition might be that the vehicle is currently traveling on a highway or a highway-like road.

[0027] According to the implementation method, the system is designed to temporarily set a learning following distance and gradually shorten the following distance to the target vehicle ahead from the learning following distance towards the normal following distance, as long as the driver does not exceed the automatic longitudinal guidance. This avoids the situation where drivers who do not want to actively exceed the automatic longitudinal guidance are forced by the system to drive at a following distance that may be too large for their perceived distance for a longer period of time.

[0028] A second aspect of the invention is a vehicle, particularly a motor vehicle, having a system according to the first aspect of the invention.

[0029] A third aspect of the invention is a computer-implemented method for learning following distances for automatic longitudinal guidance of a vehicle.

[0030] This method can be performed, for example, by means of a system according to the first aspect of the invention, and in particular by means of one or more (data) processing devices of such a system. Therefore, the above and subsequent description of the system according to the invention and its possible design schemes similarly applies to possible implementations of the method according to the invention, and vice versa.

[0031] Its automatic longitudinal guidance should be adapted to vehicles by means of a processing device designed to set the normal following distance to the target object traveling ahead during normal operation of automatic longitudinal guidance, i.e., generating corresponding control signals for automatic longitudinal guidance.

[0032] The method includes the steps described below, which can be performed using the processing apparatus described above and / or using one or more additional processing apparatuses.

[0033] In one step, a learning running following distance is set that is increased compared to the normal following distance to the target object traveling ahead.

[0034] In a further step, the driver's overtaking of the automatic longitudinal guidance is detected, and the following distance is shortened by this overtaking compared to the learned following distance. This is particularly relevant in scenarios where the driver actively takes over automatic longitudinal guidance via the accelerator pedal to reduce the following distance. If the driver demands higher acceleration than the automatic longitudinal guidance system, then the driver controls the vehicle's longitudinal guidance in a manner similar to inactive distance adjustment, so that the set following distance can be considered effective for their preferred distance behavior, as in purely manual driving (and where, if necessary, in other conditions, such as recognizing the actual following driving scenario).

[0035] Subsequently, in another step, based on the detected overshoot during automatic longitudinal guidance, a normal following distance is adapted for future following maneuvers during normal operation of automatic longitudinal guidance. For example, the normal following distance can be stored as part of a distance-speed characteristic curve that is gradually adapted to the driver's desired behavior based on the detected overshoot process.

[0036] According to the improved method, as an additional step before the above steps, the method may also include checking whether the switching conditions are met. In this case, it can be specified that the learning running following distance (rather than the normal running following distance) is set only when the check determines that the switching conditions are met. The switching conditions can also be a combination of multiple individual conditions. For example, it can be specified that the switching conditions are considered met only when multiple individual conditions are cumulatively met.

[0037] Switching conditions (e.g., as one of several single conditions) may include, for example, that the vehicle is traveling on a specific type of road, particularly on a highway or highway-like road. This condition may be determined, for example, in a known manner based on map information and / or sign recognition from a navigation system.

[0038] As already mentioned, the present invention is particularly advantageous for use on highways or highway-like roads. There, due to the high proportion of stages with active distance adjustment, the number of scenarios in which the driver manually sets their desired distance without active distance adjustment may be too small to effectively adapt the normal following distance to the driver's distance preference based solely on this. The method according to the invention can be remedied in that it also incorporates training in active driving assistance, and hereby provides the driver with more possibilities for setting their desired following distance by setting an increased learned operating following distance through over-control of distance adjustment using the accelerator pedal.

[0039] The switching conditions may involve specific measurement parameters relevant to following another vehicle, relative to alternative or additional locations associated with the currently traversed road type. For example, switching conditions may include: the actual following distance being within a predefined range and / or the actual speed of the vehicle, the actual speed of the target object, and / or the relative speed between the target object and the vehicle ahead being within the corresponding predefined range.

[0040] According to the implementation method, the normal following distance adaptation based on the detected overdrive includes: during the learning phase, performing multiple steps for at least one support point speed of the distance-speed characteristic curve, which are further described in detail below. According to a variant, the steps can be performed separately for multiple support point speeds of the distance-speed characteristic curve. Alternatively or additionally, it can be specified that the steps are performed at multiple time points during the learning phase, for one support point speed or for multiple support point speeds.

[0041] The basic approach used here to adapt distance adjustment to individual driving behavior is to train a specific distance characteristic curve. The goal is to store driver preferences and adapt the adjusted distance according to the driver's expectations. Therefore, distance adjustment should be adapted to the driver's individual driving profile in order to design distance adjustment for the driver's comfort as much as possible.

[0042] Specifically, this should be achieved by automatically adapting a stored distance-speed characteristic curve to the driver's observed driving behavior, which, within the range of automatic longitudinal guidance, is used to adjust the distance to the vehicle ahead in a speed-related manner. The distance-speed characteristic curve assigns a target following distance to a certain number of speed values, referred to as support point speeds within the range of this specification. The target following distance can be, for example, a time interval (hereinafter sometimes referred to as a second interval), which indicates at what time offset the vehicle should follow the vehicle ahead. If the second interval is, for example, 1.5 seconds, then the vehicle's longitudinal guidance is adjusted so that the vehicle arrives at the observed location (i.e., longitudinally, for example, along the shared lane of the vehicle and the vehicle ahead) 1.5 seconds after the vehicle ahead. Alternatively, however, the target following distance can also be expressed as a real distance (i.e., with units of length).

[0043] The adaptation of the characteristic curve can be carried out during the defined learning phase, while following other vehicles under the influence of the overdrive distance adjuster.

[0044] To determine the learning phase, it is necessary to identify as accurately as possible when the driver adjusts their desired following distance. Road traffic is dynamic. New vehicles may merge, accelerate, new speed limits may appear, or sudden braking may be required. In all these situations, the driver intentionally does not maintain a constant distance behind the target object. Instead, the distance can vary dynamically and may temporarily be greater than or less than the actual desired distance. In possible implementation variations of the proposed method or corresponding learning algorithm, these situations should therefore preferably be kept out of consideration; that is, the learning phase is specifically set outside these driving situations. Advantageously, the learning phase is limited to stable following situations with a relatively constant following distance, in which it can be assumed that the driver intentionally adjusts their desired following distance.

[0045] Regarding the methods and means of identifying such advantageous following situations, reference is made to the applicant's patent application filed with the German Patent and Trademark Office on January 15, 2024, entitled "Method and System for Identifying Following Situations," the contents of which are incorporated herein. The technical solutions described herein can be advantageously used within the scope of the methods currently described to define one or more learning stages.

[0046] In a step of the method, a previous (i.e., previously valid) target following distance is provided for the support point speed of the distance-speed characteristic curve. Providing this may include, for example, reading the previous target following distance from a data memory, where different speed values ​​and associated target following distances are stored.

[0047] In a further step, the detected actual following distance is provided. The actual following distance can be detected during the learning phase, for example, using a suitable sensor system (e.g., radar). Here, as mentioned above, the learning phase can occur, for example, when the driver intentionally sets a desired following distance and consistently travels behind the vehicle in front at that desired following distance.

[0048] In a separate step, a new (i.e., updated) target following distance is determined based on the actual following distance to support the point speed. Alternatively, the new target following distance can be determined, for example, based on a previous target following distance.

[0049] Other possible influencing parameters in determining the new target following distance are described in detail in the applicant's patent application filed with the German Patent and Trademark Office on January 15, 2024, entitled "Method for Learning Following Distance and System for Automatic Longitudinal Guidance of Vehicles". The contents of that patent application are incorporated in this patent application.

[0050] According to an embodiment of the method presented herein, the method described in the aforementioned patent application "Method for learning following distance and system for automatic longitudinal guidance of a vehicle" is used to adapt the normal operating following distance based on the detected overdrive.

[0051] For example, it can be stipulated that the new target following distance is determined based on the overtaking duration of the automatic longitudinal guidance. Here, the overtaking duration should not be understood, for example, merely as the duration of a specific accelerator pedal operation by which the driver exercises overtaking control of the automatic longitudinal guidance. Rather, the overtaking duration refers to the entire duration during which the following distance is changed (especially reduced) due to overtaking (which can also be achieved, for example, through multiple accelerator pedal operations spaced apart from each other in time) relative to the following distance preset by the automatic longitudinal guidance.

[0052] In particular, determining the new target following distance based on the actual following distance can be performed in such a way that the longer the overtaking duration, the greater the change in the new target following distance compared to the previous target following distance. For example, the new target following distance can be determined computationally based on a factor that increases with the duration of the overtaking automatic longitudinal guidance. Specifically, it can be specified that this factor increases more than linearly with the overtaking duration, for example, with the square of the overtaking duration. This is further illustrated below by way of an exemplary highway factor used to calculate the new target following distance based on the actual following distance.

[0053] The longer the driver's overtaking of the vehicle's automatic longitudinal guidance is maintained, particularly on highways or highway-like roads, the larger the highway factor becomes. However, in the current context, the limitation on overtaking on highways or highway-like roads should be understood as optional. In principle, this factor (which is used to calculate the new target following distance and increases with overtaking duration) can also be used during training on roads far from highways or highway-like roads.

[0054] If a new target following distance is determined, then in a subsequent step, the previous target following distance in the distance-vehicle characteristic curve is replaced by the new target following distance. During this process, the corresponding updated distance-vehicle characteristic curve, for example, can be stored to assign the new target following distance to the support point velocity.

[0055] The updated characteristic curves, in particular, reveal the adapted normal following distance, which should be adjusted at the appropriate speed during future normal operation with automatic longitudinal guidance.

[0056] The fourth aspect of the invention is a (data) processing apparatus having at least one processor and designed to execute the method according to the third aspect of the invention by means of the at least one processor. Here, according to some embodiments, spatially distributed processing apparatuses may also be involved. The processing apparatus may, for example, include a plurality of spaced-apart partial processing apparatuses. Here, the processing apparatus for setting the normal operating following distance or learning operating following distance to a target object traveling ahead in the method according to the third aspect of the invention may be one of a plurality of partial processing apparatuses, and one or more additional processing apparatuses used in the method may be one or more additional partial processing apparatuses of the processing apparatus according to the fourth aspect of the invention.

[0057] The processing unit may include, for example, a controller (or a portion thereof) for controlling automatic longitudinal guidance, such as a controller located on the vehicle. However, implementations in which the processing unit is wholly or partially located outside the vehicle are also possible. For example, it is conceivable that the processing unit or a portion thereof is part of a back-end server, with the vehicle connected to the back-end server via a wireless communication link.

[0058] According to an embodiment of the system according to the first aspect of the invention, the system includes a processing device according to the fourth aspect of the invention or is connected in data technology to such a processing device (e.g., located on or at the rear of a vehicle).

[0059] A fifth aspect of the invention is a computer program comprising commands that, when executed by a processing device (e.g., a processing device according to the fourth aspect of the invention), cause the processing device to perform a method according to the third aspect of the invention. Here, the computer program can be divided into multiple individual subroutines, each of which can be executed by different processing devices (e.g., multiple individual processors) that are spatially distant from each other if necessary. The processing device according to the fourth aspect of the invention can be designed, and in particular programmed, to execute the computer program according to the fifth aspect of the invention.

[0060] A sixth aspect of the invention is a computer-readable storage medium comprising instructions that, when executed by a processing device, such as a processing device according to a fourth aspect of the invention, cause the processing device to perform a method according to a third aspect of the invention. In other words, a computer program according to a fifth aspect of the invention can be stored on a computer-readable storage medium. Attached Figure Description

[0061] The present invention will now be described in detail with reference to the embodiments and the accompanying drawings.

[0062] Figure 1 An exemplary and schematic diagram illustrates a system for learning following distances for automatic longitudinal guidance of a vehicle.

[0063] Figure 2A The steps of a method for learning the following distance for automatic longitudinal guidance of a vehicle are illustrated exemplaryly and schematically.

[0064] Figure 2B Exemplary and schematic illustrations show in Figure 2A Within the scope of the method, adapt and set possible steps for future normal following distance when following other vehicles.

[0065] Figure 3 An example is shown of a distance-speed graph, which has a distance-speed characteristic curve for normal operating following distance and a distance-speed characteristic curve for learning operating following distance.

[0066] Figure 4 Examples are shown of normal operating following distance and learning operating following distance compared to a normal distribution of the following distance expected by the driver. Detailed Implementation

[0067] Subsequently, referring to method 4 used for learning following distance... Figure 2A Steps 40 to 44, illustrated in the block diagram, are explained in Figure 1 System 1 is shown schematically in the diagram.

[0068] System 1 is also designed to adapt the normal following distance N between the vehicle and the target object traveling in front of the vehicle for normal operation of automatic longitudinal guidance based on the detected overshoot by the vehicle driver, so that the normal following distance is adjusted for future following distance settings during normal operation of automatic longitudinal guidance.

[0069] System 1 is also designed to set a learning running following distance L that is increased compared to the normal running following distance N, based on the situation, so that the driver has more opportunities to set the following distance that the driver desires to be reduced compared to the learning running following distance L by overtaking the automatic longitudinal guidance.

[0070] For this purpose, system 1 includes a distance sensor system 12 for detecting the actual following distance between the vehicle and a vehicle traveling in front (the vehicle in front). For example, the distance sensor system 12 may have one or more radar sensors. In addition to following distance, the relative speed between the vehicle and the vehicle in front may also be determined, for example, by means of the radar sensors.

[0071] In addition, system 1 also includes a speed sensor system 13 for detecting the actual speed of the vehicle. The speed sensor system 13 may include, for example, one or more wheel speed sensors, one or more acceleration sensors (from which the actual speed can be calculated) and / or a receiver for a global navigation satellite system (wherein the actual speed can be determined based on changes in the vehicle's position detected by the global navigation satellite system).

[0072] The (data) processing device 10 of System 1 is connected to the distance sensor system 12 and the speed sensor system 13 in a signaling technology, and is designed to receive a distance signal A, including the actual following distance, from the distance sensor system 12, and a speed signal G, including the actual speed, from the speed sensor system 13. Furthermore, the processing device 10 can, for example, receive a relative speed signal V from the distance sensor system 12, and the speed of the vehicle ahead can be determined by combining the relative speed signal with the speed signal G.

[0073] According to Figure 1 In this embodiment, the processing device 10 is also connected to the output device 16 in a signaling technology manner. The processing device 10 can generate control signals H for the output device 16, thereby outputting these prompts or information to the driver. Figure 1 In this embodiment, output device 16 is exemplary and schematically shown as a speaker, since cues can be output in an acoustic form (e.g., as voice output). Additionally or alternatively, output device 16 may include, for example, a display for displaying visual information.

[0074] If the learning phase of the learning algorithm used to learn the following distance is active, the driver can be informed of this via output device 16. Here, the driver may be notified, for example, that the current overtaking of the automatic longitudinal guidance has resulted in a reduction of the previously stored normal following distance N.

[0075] Furthermore, the processing device 10 is connected to the data memory 11 of the system 1 in a signaling technology and is designed to receive characteristic curve information K, such as speed values ​​and target following distances respectively associated with the speed values, from the data memory based on the distance-speed characteristic curve (hereinafter referred to as characteristic curve) stored in the data memory.

[0076] The processing device 10 is also designed to further refer to the following based on the detected driving behavior (especially based on the detected actual distance and actual speed). Figure 2B The updated characteristic curve information K' is determined in a detailed manner, and the updated characteristic curve information is output to the data memory 11 for storage, so that the updated characteristic curve is stored in the data memory 11.

[0077] The data storage 11 is connected in signal technology to another processing device in the form of a controller 14, which is designed to control the automatic longitudinal guidance of the vehicle.

[0078] The controller 14 is designed to set a normal following distance N to the target vehicle ahead during normal operation of automatic longitudinal guidance. Here, the controller 14 may, for example, read characteristic curve information K, K' (which may include, in particular, the normal following distance for the current speed of the vehicle) from the data memory 11, and generate a control signal S for controlling automatic longitudinal guidance based on the characteristic curve information K, K', in particular to adjust the normal following distance N to the vehicle ahead (as the target following distance).

[0079] The controller 14 can output a control signal S to the vehicle's longitudinal guidance actuator 15, particularly the braking system and / or drive module. Subsequently, the longitudinal guidance actuator 15 can control automatic longitudinal guidance according to the control signal S, and in particular adjust the target following distance to the vehicle ahead. Figure 1 In the diagram, the longitudinal guiding actuator 15 is drawn with a dashed line because it can in principle be a general longitudinal guiding actuator known in the prior art, and it need not be understood as part of the system 1 according to the invention.

[0080] System 1 is designed to execute method 4 for learning following distance, using controller 14 and processing device 10. Figure 2A Steps 40 to 44 are illustrated in the schematic block diagram.

[0081] In particular, the controller 14 is designed to set a learning operating following distance L with respect to the vehicle ahead that is increased compared to the normal operating following distance N, based on the situation (step 41).

[0082] In addition, the processing device 10 is designed to detect (step 42) the vehicle driver's overtaking of the automatic longitudinal guidance (by overtaking to shorten the following distance compared to the learning running following distance L) and then, based on the detected overtaking of the automatic longitudinal guidance, adapt the normal following distance N for future following driving during normal operation of the automatic longitudinal guidance (step 43).

[0083] exist Figure 1 In the illustrated embodiment, the processing device performs the steps described above for detecting 42 overtaking by the vehicle driver in the automatic longitudinal guidance and for adapting 43 the normal operating following distance based on the detected overtaking. The processing device 10 is separate from the controller 14, which controls the vehicle's automatic longitudinal guidance and, in particular, also performs the method steps of setting 41 the learning operating following distance L. Here, the controller 14 is connected to the processing device 10 in signal technology and receives, for example, distance and speed information A, G, V from the processing device as the basis for the vehicle's longitudinal adjustment.

[0084] However, the following implementations are also possible, in which one and the same processing device performs steps 41-43 of method 4. In other words, refer to Figure 1 The processing device 10 may be the same as or a part of the controller 14 (or vice versa).

[0085] Method 4 may also include checking whether the switching conditions are met as an additional step 40 prior to steps 41-43. Figure 2A In this context, this step is shown with a dashed box because it is an optional additional step. In this case, it can be specified that the learning operating following distance L (instead of the normal operating following distance N) is set only if the check determines that the switching conditions are met. Switching conditions may include, for example, the vehicle being driven on a specific type of road, particularly on highways or highway-like roads.

[0086] In an additional optional step 44 of method 4, controller 14 may generate a control signal S for controlling automatic longitudinal guidance based on an adapted, updated normal operating following distance (i.e., based on corresponding updated characteristic curve information K') during subsequent normal operation of automatic longitudinal guidance, in particular to adjust the adapted normal operating following distance or the target following distance with the vehicle ahead according to the corresponding adapted characteristic curve. The control signal S may be output to the longitudinal guidance actuator 15.

[0087] Starting from the adapted normal following distance, Method 4 can be repeated in subsequent situations, where the modified normal following distance has the effect of the previous normal following distance. In other words, the learned normal following distance (changed if necessary relative to the previous implementation) which is greater than the previously modified normal following distance can be adjusted to provide the driver with an opportunity to shorten the following distance by using overtaking with the accelerator pedal. Based on this overtaking, the previously modified normal following distance can be further adapted, thereby gradually and better learning the distance behavior desired by the driver.

[0088] To adapt to the normal following distance of vehicle 43, the processing unit 10 of system 1 can be designed to perform speed calculations based on one or more support points of the distance-speed characteristic curve at one or more time points during the learning phase. Figure 2B Method steps 431 to 434.

[0089] In detail according to Figure 2B Before steps 431 to 434, refer to... Figure 3 A brief description of the distance-velocity characteristic curves shown therefor.

[0090] Figure 3 An example of a characteristic curve in the form of a graph is shown, in which the second interval d is plotted against the velocity v. Here, the initially valid characteristic curve is shown in the graph as small circles connected by solid lines (graph note: "Base"). The data points shown as small circles are assigned a second interval to the support point velocities (hereinafter referred to as "support points"). Between the support points, the characteristic curve is derived by (linearly) interpolation.

[0091] exist Figure 3 The exemplary characteristic curve shown uses five support points. The more support points, the more personalized the characteristic curve. Therefore, if available storage space allows in a specific application, using more than five support points is theoretically meaningful. However, due to the interpolation that is always performed between the data points, it is not necessary to choose a step size of approximately 1 km / h or smaller between the support points to obtain favorable results. The most common speed limits, such as 30 km / h, 50 km / h, 80 km / h, 100 km / h, and 120 km / h, and some higher speeds, such as 140 km / h, 160 km / h, or 180 km / h, are sufficient to provide a perceptible impact for the driver. If system speed limits (i.e., especially for autonomous driving functions) are added for this purpose, then, for example, 10 support points and 10 associated distance values ​​need to be stored.

[0092] The characteristic curve shown, for example, assigns a normal following distance N of slightly more than 1.4 seconds to a speed of approximately 56 m / s. Figure 3 As shown by the (thin) dashed auxiliary line.

[0093] The characteristic curve (solid line) shown, specifically for setting the normal following distance N at a speed of approximately 56 m / s, can be changed over time by a learning algorithm and thus adapt to the distance behavior desired by the driver.

[0094] However, according to the advantageous implementation, the individual second intervals cannot be trained completely freely, but certain constraints can be set, and they must move within those constraints. Figure 3 This is illustrated in the diagram by way of the lower and upper boundary characteristic curves (see note: "Boundary"). The characteristic curve being learned must lie between the upper and lower boundary characteristic curves.

[0095] The upper limit can be provided, for example, by the performance of the sensor system. Classic radar sensors have an effective range of approximately 150 meters, while more advanced components have a maximum range of 300 meters. For objects at greater distances, target loss may occur, potentially leading to uncomfortable longitudinal braking behavior for the driver. If distant targets are repeatedly lost, it can result in incomprehensible acceleration and braking by the driver. In following a vehicle traveling at 190 km / h, the following distance corresponding to the radar's maximum field of view of 150 meters corresponds, for example, to a following distance of 2.84 seconds. However, to avoid erroneous braking or other undesirable effects, it is preferable to also include an additional buffer. Therefore, according to... Figure 3 In this embodiment, the 2.5-second value that forms the upper boundary characteristic curve is assumed to be the maximum interval. This can be further increased or decreased depending on the sensor performance.

[0096] Regarding the lower boundary, legal requirements on the one hand and standard conditions on the other can be particularly considered as boundary conditions. For example, the distance derived from the trained characteristic curve should not be lower than the distance derived from the characteristic curve corresponding to the minimum distance level set by the driver. If the ACC function has, for example, three distance levels, then the minimum distance level should form the lower boundary. Figure 3 The lower boundary characteristic curve in the curve corresponds to the boundary that has been defined in this way.

[0097] In a system with three selectable distance levels, the characteristic curve stored for the intermediate distance level can, for example, be used as a starting point for training the characteristic curve. This characteristic curve can serve as a known standard value for the driver, who may have already driven previous models of the same vehicle brand. Based on this, learning can be quickly achieved in both positive and negative directions (regarding following distance), allowing the driver to immediately perceive the positive impact.

[0098] exist Figure 3 In the graph, to illustrate method 4 according to the invention, a (thick) dashed characteristic curve is also plotted, which sets the corresponding second interval for speeds above 10 m / s as the improved learning running following distance compared to the solid characteristic curve. A (thin) dashed auxiliary line, for example, illustrates that the learning running following distance L is set for a speed of approximately 56 m / s, which is approximately 1.6 seconds and therefore slightly less than the normal operating following distance N by 0.2 seconds.

[0099] Step 41 of the method for setting a learning operating following distance L that is increased compared to the normal operating following distance N can be performed based on this modified characteristic curve. If the switching conditions of the type described above are met, then, for example, it is possible to switch between a solid characteristic curve with the normal operating following distance N and a dashed characteristic curve with the learning operating following distance L, thereby setting the learning operating following distance L instead of the normal operating following distance N at a speed of, for example, 56 m / s.

[0100] In the example shown, the dashed characteristic curve deviates from the solid characteristic curve only at relatively high speeds exceeding 10 m / s, because the increased learning running following distance according to the invention's setting 41, as already mentioned, may be particularly important at highway driving, where speeds of 10 m / s and slower are generally not important.

[0101] Specific driving conditions or time periods during driving can be identified as suitable for training characteristic curves. This can particularly relate to following driving with a relatively constant following distance, implemented by the driver (or at least influenced by the driver). In this case, a learning phase can be initiated, in which the following description is implemented separately at multiple time points, e.g., in consecutive calculation cycles. Figure 2B Steps 431-434.

[0102] In step 431, using the characteristic curve information K, the support point velocity v is assigned to the previously valid distance-velocity characteristic curve. n Provide the previous (i.e., previously valid) target following distance d n Providing, for example, may include reading the previous target following distance d from the data storage 11. nDifferent speed values ​​and the corresponding target following distances are stored in this data memory according to the characteristic curves.

[0103] In another step 432, the actual following distance d to be detected is provided. ist Here, the actual following distance d ist For example, the actual following distance d can be detected using the distance sensor system 12 during the driver overtaking automatic longitudinal guidance phase. Subsequently, the actual following distance d is detected. ist It can be received along with the distance signal A from the distance sensor system 12 on the processing device 10 side and provided for further data processing.

[0104] In another step 433, based on the actual following distance d ist and the previous following distance d n For support point speed v n Determine the new (i.e., updated) target following distance d. n’ .

[0105] Subsequently, in another step 434, the support point v in the characteristic curve is... n The previous target following distance d n The distance d from the new target vehicle n’ Alternatively, in this process, the updated characteristic curve information K' can be stored, for example, in the data memory 11, thus the updated characteristic curve is stored in the data memory 11, and the support point v is given according to the updated characteristic curve information. n Assigning a new target vehicle with a following distance d n’ .

[0106] In step 433, determine the new target following distance d. n’ At that time, according to the embodiments described in detail below, at multiple time points during the learning phase (e.g., in each calculation cycle), the corresponding new target following distance d n’ Based on the corresponding previously valid (e.g., determined in a previous calculation period) target following distance d n and the corresponding current actual following distance d ist It is determined according to the following equation:

[0107] The learning rate factor appears particularly on the right side of the equation. α Highway factors f AB Learning duration factor f t Speed ​​difference factor f δv Credibility factor fk And depends on the actual following distance d ist Distance d from the previous target vehicle n The factor of deviation between them.

[0108] It should be noted that the following embodiments are also possible, in which the following embodiment is used to determine the new target following distance d. n’ In the expression, for example, besides the previous target following distance d n In addition, only one or some of the other influence factors mentioned above may be used.

[0109] In a possible implementation, the influence factor described above, which depends on the actual following distance d, is used. ist Distance d from the previous target vehicle n At least one factor of the deviation between them, such as the last factor in the second term on the right-hand side of the equation above. Additionally, the velocity difference factor is preferably considered. f δv and / or learning duration factor f t and / or credibility factor f k and / or highway factor f AB .

[0110] The aforementioned potential influencing factors are described in detail in the applicant's patent application, "Method for learning following distance and system for automatic longitudinal guidance of vehicle".

[0111] The highway factors were then described more precisely. f AB As a concrete example of implementation, the following distance of the new target vehicle (433) is determined based on the duration of the overdrive automatic longitudinal guidance.

[0112] If a driver activates the distance adjustment system and overshoots it several times without determining a real change in distance to the vehicle ahead, the driver can deactivate the system early or switch back to the standard distance level. Therefore, it makes sense to set up logic for this specific situation that can be applied to the learning objectives significantly more quickly.

[0113] This can be achieved by, within the scope of the learning algorithm, determining the new target following distance d. n’ When considering highway factors f AB The longer a vehicle driver maintains automatic longitudinal guidance overtaking on a highway or similar road, the greater the highway factor becomes.

[0114] Here, for highway factors fAB The specific design takes several criteria into consideration. Firstly, this additional function should ideally be activated only when the vehicle is on a highway or similar road. In urban traffic, overtaking may occur frequently, making it less suitable for distance targets. Furthermore, usage rates in urban traffic are not as high as on other road types.

[0115] In this embodiment, the highway factor f AB It is named this way because it is only effective in highway environments. However, in general, this factor does not necessarily have to be designed so that its effect is limited only when vehicles are traveling on highways or highway-like roads. On other road types, overtaking can also be considered in the manner described here when determining the new target following distance, i.e., in particular, the longer the detected overtaking duration, the greater the change in the new target following distance compared to the previous target following distance.

[0116] In other words, highway factor f AB The optimal influence should be based on the duration of the over-control automatic longitudinal guidance. t ueb And increase. Because a large influence is needed during a long learning phase, a quadratic relation is chosen, for example. Highway factor. f AB For example, it can be determined using the following equation:

[0117] in α AB It is a fixed value used for standardizing time, and t ueb It refers to the duration of the overrun, which can be considered in the highway factor for algorithmic processing purposes. f AB The middle is limited to the maximum value t max .

[0118] Exceeding the duration t ueb For example, it could be similar to the duration of learning. t l The formula is calculated, as described in the aforementioned patent application "Method for Learning Following Distance and System for Automatic Longitudinal Guidance of Vehicles". In other words, at the start of the learning phase, this can also be set here. d t0 and v t0However, it is preferable to set an additional time interval at the beginning. This helps to ensure that the initial values ​​are actually set to the correct initial values ​​and avoids possible erroneous training operations. Otherwise, even a slight deviation could subsequently lead to the interruption of adaptive acceleration through the additional factor. During this additional time interval, the driver can adjust their desired distance by overtaking. This takes a certain amount of time, and the vehicle must still accelerate accordingly to establish the driver's constant desired distance. Here, once all learning conditions are met, the driver actively overtakes the accelerator pedal, and the road type is determined to be either a highway or a highway-like road, then the relevant timer is started. Only when the timer expires... t ueb It was only set to 0, and d t0 and v t0 It was then set to the current value. Subsequently, a process similar to learning the duration began. t l Formula for calculating highway factors f AB .

[0119] Figure 4 The charts in the example illustrate the normal operating following distance N and the learned operating following distance L compared to a statistically normal distribution of the following distance expected by the driver.

[0120] As illustrated in the diagram, the increase in following distance according to the invention, from the normal following distance to the learning following distance, provides users of the automatic longitudinal guidance function with more opportunities, statistically speaking, to set the driver's desired following distance through over-control of the automatic longitudinal guidance. Thus, training of the desired distance behavior can be conducted more effectively. Here, increasing the following distance to the learning following distance L is significant, allowing the driver to easily correct, i.e., especially shorten, the following distance via the accelerator pedal. In other words, due to the initial increase in following distance, the probability of the driver wanting to reduce the following distance increases.

[0121] If we assume a Gaussian-normal distribution for the desired following distance, then shifts in the distance characteristic curve (especially increases in following distance from the learned following distance L) can have a significant impact. Figure 4 As shown in the chart, while more drivers will therefore have to adapt their distances, on the other hand, drivers are now also given the option to use a self-learning system, which they would otherwise have to adjust their desired distances by disabling the system or the self-learning function.

[0122] The drawn vertical dashed lines mark the normal following distance N (left) and the learning operating following distance L (right). The curve itself represents the driver's desired distance. All users located to the left of the corresponding vertical dashed line in a normal distribution can adjust their desired distance by overtaking it using the accelerator pedal. The desired distance is less than the desired distance of the corresponding initial characteristic curve, and therefore can be achieved in this way. All other users located to the right of the corresponding vertical dashed line must deactivate the system or manually change the distance level. By shifting the distance characteristic curve, the number of users is significantly increased, and users can set their desired distance by overtaking it. The learning algorithm is therefore also able to provide drivers with rapid adaptation to their individual expectations, who primarily use the distance adjustment system on highways. Due to overtaking, this possibility can also be realized using active driving assistance and expand the system's adaptability.

Claims

1. A system (1) for learning following distances for automatic longitudinal guidance of a vehicle, wherein, The system (1) is designed to adapt and set a normal following distance (N) for normal operation of the automatic longitudinal guidance and a target object traveling in front of the vehicle based on the detected overtaking of the automatic longitudinal guidance by the driver of the vehicle, such that a normal following distance (N) is set for future following during normal operation of the automatic longitudinal guidance, and wherein the system is further designed to set a learning operating following distance (L) for the target object traveling in front, which is increased compared to the normal following distance (N), according to the situation, so that the driver has more opportunities to set the actual following distance that the driver desires to be reduced compared to the learning operating following distance (L) by overtaking the automatic longitudinal guidance.

2. The system (1) according to claim 1, wherein, The system (1) is designed to temporarily set the learning running following distance (L) and then gradually bring the actual following distance to the target object traveling ahead closer to the normal running following distance (N), as long as the driver does not exceed the control of the automatic longitudinal guidance.

3. A vehicle having a system (1) according to any one of the preceding claims.

4. A computer-implemented method (4) for learning following distances for automatic longitudinal guidance of a vehicle, wherein, The vehicle includes a processing device (14) designed to set a normal following distance (N) to a target object traveling ahead during normal operation of the automatic longitudinal guidance, and wherein the method (4) includes the following steps performed by means of the processing device (14) and / or by means of one or more additional processing devices (10): - Set (41) an increased learning running following distance (L) with the target object traveling ahead compared to the normal running following distance (N); - Detection (42) of the driver's overtaking of the automatic longitudinal guidance, thereby shortening the actual following distance compared to the learned running following distance (L); and - In the normal operation of the automatic longitudinal guidance, the normal following distance (N) is adapted for future following driving (43), wherein the normal following distance (N) is adapted according to the detected overtaking of the automatic longitudinal guidance.

5. The method (4) according to claim 4, wherein, As an additional step prior to steps (41)-(43) above, the method (4) includes checking (40) whether the switching condition is met, and wherein the learning running following distance (L) is set only if the check (40) finds that the switching condition is met.

6. The method (4) according to claim 5, wherein, The switching conditions include: the vehicle is traveling on a specific road type, especially on highways or highway-like roads.

7. The method (4) according to any one of claims 4 to 6, wherein, The adaptation of the normal following distance (43) includes: performing the following steps during the learning phase for at least one support point speed of the distance-speed characteristic curve: - Provide (431) the previous target following distance for the speed of the support point; - Provides (432) the actual following distance detected; - Based on the actual following distance, determine the new target following distance (433) for the speed of the support point; and - Replace the previous target following distance in the distance-speed characteristic curve (434) with the new target following distance.

8. The method (4) according to claim 7, wherein, The following distance to the new target is determined based on the duration of the automatic longitudinal guidance provided by the overdrive control (433).

9. The method (4) according to claim 8, wherein, The new target following distance is determined based on the actual following distance in the following manner (433), that is, the longer the overtaking duration, the greater the change in the new target following distance compared with the previous target following distance.

10. A processing apparatus (10, 14) having one or more processors, wherein, The processing device is designed to perform the method (4) according to any one of claims 4 to 9 by means of the one processor or by means of the plurality of processors.

11. A computer program comprising instructions that, when executed by at least one processing device (10, 14), cause the processing device to perform the method (4) according to any one of claims 4 to 9.

12. A computer-readable storage medium comprising instructions that, when executed by at least one processing device (10, 14), cause the processing device to perform the method (4) according to any one of claims 4 to 9.

Images