DE102020200620A1 - Detection of a turning process of a second vehicle - Google Patents

Abstract

The present invention relates to a device (24) for detecting a turning process of a second vehicle (12) in the vicinity of a first vehicle (14), having: an input interface (26) for receiving a wirelessly transmitted message from a vehicle-to-vehicle Communication unit (16) of the second vehicle, the message comprising foreign vehicle data with information on a current state of the second vehicle; a classification unit (28) for determining whether the second vehicle will perform a turning process based on an evaluation of the external vehicle data in a pre-trained classification method; and an output unit (30) for outputting an output signal with information on the turning process when the second vehicle is going to perform a turning process. The present invention further relates to a method and a system (10) for detecting a turning process of a second vehicle (12) in the vicinity of a first vehicle (14).

Description

The present invention relates to a device for detecting a turning process of a second vehicle in the vicinity of a first vehicle. The present invention further relates to a system and a method for detecting a turning process of a second vehicle in the vicinity of a first vehicle.

Modern vehicles (cars, vans, trucks, motorcycles, etc.) include a large number of systems that provide the driver with information and control individual functions of the vehicle in a partially or fully automated manner. The surroundings of the vehicle and other road users are recorded via sensors. Based on the recorded data, a model of the vehicle environment can be generated and changes in this vehicle environment can be reacted to. As a result of the advancing development in the field of autonomous and semi-autonomous vehicles, the influence and scope of such driver assistance systems (Advanced Driver Assistance Systems, ADAS) are increasing. The development of ever more precise sensors makes it possible to record the environment and the traffic and to control individual functions of the vehicle completely or partially without the intervention of the driver. Driver assistance systems can contribute in particular to increasing traffic safety and improving driving comfort.

In addition to using sensors on your own vehicle, data can also be exchanged with other vehicles (vehicle-to-vehicle communication) and / or with infrastructure objects (e.g. traffic lights, traffic control systems, cameras, etc., vehicle-to-infrastructure Communication). Such so-called V2X systems and the formats and contents of the exchanged messages have meanwhile been specified more precisely in the IEEE 1609 standard family. For the future, all vehicles are to be equipped with appropriate communication units. The exchanged data can be used, for example, to adapt the behavior of an autonomous or partially autonomous vehicle to a current situation.

One challenge is to react adequately to the behavior of other road users. For this it is first necessary to detect an intention or a planned driving maneuver of another vehicle. A reaction of the own vehicle can then be determined. In this Correlation is given in Tang et al., "Turn Prediction at Generalized Intersections", IEEE 2015 , discloses an approach to predicting a turn at an intersection. In particular, an intention of a vehicle driver is predicted. The disclosed model uses maps of lane guidance in an intersection and makes a prediction based on previous positions and the movement of the vehicle. A data set with several hundred routes at different intersections is generated. The prediction of the turning process is based on a hidden Markov model, a support vector machine and a dynamic Bayesian network.

A disadvantage of previous approaches for recognizing the behavior of another vehicle is that they often only provide reliable results in previously known environments. In addition, sensor data are often required as input signals that may or may not be present, for example because a camera cannot detect the other vehicle due to a lack of direct line of sight.

Based on this, the present invention therefore has the task of creating an approach for the reliable detection of a turning process of a second vehicle. It should be recognized whether or not a second vehicle in the vicinity of one's own vehicle will make a turning maneuver in the immediate future. In particular, an approach is to be created that makes it possible to provide reliable results in any environment.

• einer Eingangsschnittstelle zum Empfangen einer drahtlos übermittelten Nachricht von einer Fahrzeug-zu-Fahrzeug-Kommunikationseinheit des zweiten Fahrzeugs, wobei die Nachricht Fremdfahrzeugdaten mit Informationen zu einem aktuellen Zustand des zweiten Fahrzeugs umfasst;
• einer Klassifikationseinheit zum Ermitteln, ob das zweite Fahrzeug einen Abbiegevorgang durchführen wird, basierend auf einer Auswertung der Fremdfahrzeugdaten in einem vortrainierten Klassifikationsverfahren; und
• einer Ausgabeeinheit zum Ausgeben eines Ausgabesignals mit Informationen zu dem Abbiegevorgang, wenn das zweite Fahrzeug einen Abbiegevorgang durchführen wird.

To achieve this object, the present invention relates in a first aspect to a device for detecting a turning process of a second vehicle in the vicinity of a first vehicle, with:

• an input interface for receiving a wirelessly transmitted message from a vehicle-to-vehicle communication unit of the second vehicle, the message comprising external vehicle data with information on a current state of the second vehicle;
• a classification unit for determining whether the second vehicle will perform a turning process based on an evaluation of the external vehicle data in a pre-trained classification method; and
• an output unit for outputting an output signal with information on the turning process when the second vehicle is going to perform a turning process.

In a further aspect, the present invention relates to a vehicle-to-vehicle communication unit with a device as described above.

• einer Fahrzeug-zu-Fahrzeug-Kommunikationseinheit in einem zweiten Fahrzeug; und
• einer Vorrichtung oder einer Fahrzeug-zu-Fahrzeug-Kommunikationseinheit wie zuvor beschrieben.

Furthermore, one aspect of the present invention relates to a system for detecting a turning process of a second vehicle in the vicinity of a first vehicle, with:

• a vehicle-to-vehicle communication unit in a second vehicle; and
• a device or a vehicle-to-vehicle communication unit as described above.

Further aspects of the invention relate to a corresponding method and a computer program product with program code for performing the steps of the method when the program code is executed on a computer, as well as a storage medium on which a computer program is stored which, when executed on a computer, causes the method described herein to be carried out.

Preferred embodiments of the invention are described in the dependent claims. It goes without saying that the features mentioned above and those yet to be explained below can be used not only in the respectively specified combination, but also in other combinations or on their own, without departing from the scope of the present invention. In particular, the method and the computer program product can be implemented in accordance with the configurations described for the device in the dependent claims.

According to the invention, it is provided that data (external vehicle data) are received from a second vehicle in the immediate vicinity of a first vehicle via vehicle-to-vehicle communication. Based on this received third-party vehicle data, a pre-trained classification method is used to determine whether the second vehicle will perform a turning maneuver. In this respect, the invention makes use of the fact that data that are already transmitted via vehicle-to-vehicle communication can be used as an input for a machine learning approach. It is predicted whether the second vehicle will perform a turning maneuver. This information can then be output to a driver of the first vehicle (own vehicle). It is also possible for an actuator of the first vehicle to be controlled directly on the basis of the detection. For example, a braking or acceleration process can be initiated or influenced when a second vehicle is turning.

The processing according to the invention of transmitted vehicle-to-vehicle communication data of the second vehicle in a classification method enables a reliable forecast of the turning behavior of the second vehicle to be made. It is not necessary to use sensors from the first vehicle. In addition, there is no need to receive any data from sensors in the infrastructure. In this respect, for example, line of sight and recording of an image signal by means of a camera are not necessary. By using a machine learning method, a high level of reliability can be achieved even in previously unknown environments and situations. Compared to the prior art, there is greater flexibility and forecast accuracy. By providing the output signal with information on the turning process, the safety of road users, in particular of the first and second vehicles, can be improved. The risk of accidents is reduced. In addition, the energy efficiency can be increased by implementing a predictive driving style and avoiding unnecessary braking and / or acceleration processes. In particular in the field of autonomous and semi-autonomous vehicles, energy can be saved and safety increased.

In a preferred embodiment, the pre-trained classification method is a random forest classifier. A random forest classifier is a machine learning technique for classification and regression and creates multiple decision trees to provide a prediction. Such a classifier works well with a mixture of numerical and categorical features on the input side. The classifier can, for example, be trained based on collected data from various vehicles. This results in a high level of forecasting accuracy when predicting the turning process of the second vehicle.

In a preferred embodiment, the classification unit is designed to determine a turning direction of the second vehicle based on the evaluation of the external vehicle data in the pre-trained classification method. The output unit is designed to output the determined turning direction. In addition to the (binary) decision as to whether a turning process is taking place, it can be determined whether a turning to the left or to the right or in a different direction is to take place. As a result, additional information can be provided in the further processing of the output signal. The reaction of the first vehicle can be better adjusted. Safety and energy efficiency can be further improved.

In a preferred embodiment, the device comprises a filter unit for determining a position and direction of travel of the second vehicle based on the received message. The classification unit is designed to determine the turning process only if the second vehicle is a vehicle driving ahead of the first vehicle. In principle, a turning process can be detected for all vehicles in the vicinity of the first vehicle from which messages are received. It is advantageous if a turning process is detected, in particular for vehicles driving ahead. In the case of a vehicle in front, your own driving style can be adjusted in order to avoid unnecessary braking or acceleration and to prevent an accident.

In a preferred embodiment, the device comprises a control unit for controlling a motor and / or a brake of the first vehicle based on the output signal. This makes it possible to adapt an acceleration and / or braking process of the first vehicle when the second vehicle is going to perform a turning process. It is possible for a control unit of the first vehicle to be activated directly in order to adapt a driving style of the first vehicle. In this way, a predictive driving style can be implemented in a partially autonomous or autonomous vehicle. The result is improved safety for all road users and higher energy efficiency.

The input interface is preferably designed to receive a standardized Dedicated Short Range Message, DSRC. In particular, a Cooperative Awareness Message, CAM, a Decentralized Environmental Notification Message, DENM, a Signal Phase and Timing Message, SPAT and / or a Map Data Message, MAP can be received. In particular, standardized messages from other vehicles can be received and processed. The use of CAM messages is particularly advantageous. On the one hand, the data contained as standard in CAM messages can be used as input variables for the classification. On the other hand, it is possible to use optional fields. Through the use of standardized messages, interaction with any other vehicles can take place. A prior adaptation or negotiation of a communication protocol is not necessary. This results in improved safety and increased efficiency for all road users involved.

In a preferred embodiment, the external vehicle data include a steering wheel angle, a steering wheel angle confidence, a steering wheel angle change rate, a vehicle wheel angle, an acceleration, an acceleration confidence, a yaw angle, a yaw angle confidence, a speed, a direction and / or a position. A confidence is understood here to mean a value that expresses a certainty or an uncertainty of a measured value. The various data mentioned allow a machine learning model to be set up with a high level of informative value with regard to an imminent turning maneuver of a vehicle. It is possible to train the classifier based on using multiple numerical and categorical features. During the training phase, the various data can be collected while a vehicle is in operation.

In a preferred embodiment, the input interface is designed to receive a RAW message. The external vehicle data include a friction confidence and / or a rate of rotation.

In a preferred embodiment, the classification method is pre-trained based on recorded third-party vehicle data from several vehicles. In particular, a training data set can be obtained based on collected data from several other vehicles. It is possible for the training data to be updated or supplemented while the first vehicle is in operation, possibly based on data received from other vehicles. This results in a high level of informative value of the classifier with regard to an upcoming turning maneuver of the second vehicle.

In one embodiment, the device is integrated into a vehicle-to-vehicle communication unit. In other words, a corresponding classification method is carried out in a processor of a vehicle-to-vehicle communication unit. The result is an expanded and improved data processing of the received data.

The detection of a turning process is understood here to mean, in particular, a binary decision as to whether a turning process is imminent. Immediately imminent means in the near future, with a longer advance warning time having a positive effect. A precise time horizon does not have to be defined. In other words, the prognosis as to whether a turning maneuver is imminent should be made as early as possible. At the latest when the turning process has already been initiated, the classifier should provide reliable detection of the turning process. A turning direction can in particular be right or left. It is also possible for a further or more precise specification of a direction to take place, for example by means of a number of degrees. A pre-trained classification method is a machine learning algorithm that has been trained using training data.

• 1 eine schematische Darstellung eines erfindungsgemäßen Systems zum Detektieren eines Abbiegevorgangs;
• 2 eine schematische Darstellung einer erfindungsgemäßen Vorrichtung;
• 3 eine schematische Darstellung eines Fahrzeugs mit einer erfindungsgemäßen Fahrzeug-zu-Fahrzeug-Kommunikationseinheit; und
• 4 eine schematische Darstellung eines erfindungsgemäßen Fahrzeugs.

The invention is described and explained in more detail below with the aid of a few selected exemplary embodiments in connection with the accompanying drawings. Show it:

• 1 a schematic representation of a system according to the invention for detecting a turning process;
• 2 a schematic representation of a device according to the invention;
• 3 a schematic representation of a vehicle with a vehicle-to-vehicle communication unit according to the invention; and
• 4th a schematic representation of a vehicle according to the invention.

In the 1 is a system according to the invention 10 for detecting a turning process of a second vehicle 12th in an environment of a first vehicle 14th shown. The system 10 comprises a vehicle-to-vehicle communication unit 16 in the second vehicle and a vehicle-to-vehicle communication unit 18th in the first vehicle 14th . The representation is a top view from a bird's eye view of a street 20th in the area of a confluence 22nd to understand. The first vehicle 14th and the second vehicle 12th move in the direction of the indicated arrows. The vehicle-to-vehicle communication unit 18th of the first vehicle 14th is in connection with a device according to the invention 24 for detecting a turning process of the second vehicle 12th .

According to the invention, the vehicle-to-vehicle communication units of the two vehicles are in communication, as indicated by the arrow. The surroundings of the first vehicle 14th is defined in particular by the fact that it corresponds to a communication range of a corresponding communication unit. The second vehicle uses the communication link 12th Foreign vehicle data to the first vehicle 14th transmitted. The transmission is preferably carried out periodically by means of standardized messages. These foreign vehicle data include information on a current state of the second vehicle.

In the device according to the invention 24 in the example shown inside the first vehicle 14th is arranged, the received third-party vehicle data is evaluated based on a machine learning approach. A classification method is carried out that automatically recognizes a turning maneuver of the second vehicle 12th enables. Based on the transmitted third-party vehicle data, the device 24 determined whether the second vehicle 12th at the confluence 22nd wants to make a turn. Such a turning process usually leads to the travel of the second vehicle being slowed down 12th . This can lead to a rear-end collision if the first vehicle 14th drives onto the second vehicle. It is also possible that in the first vehicle 14th a braking process becomes necessary that would not be absolutely necessary if the intention of the second vehicle were to occur 12th would be known in advance. The first vehicle 14th can implement a more predictive driving style with the knowledge determined, since it is immediately from an imminent turning maneuver of the second vehicle 12th will be notified.

In the 2 is the device according to the invention 24 shown schematically. The device comprises an input interface 26th , a classification unit 28 as well as an output unit 30th . Optionally, the device includes 24 also a filter unit 32 as well as a control unit 34 . The units and interfaces can be implemented partially or completely in software and / or in hardware. In particular, the units can be designed as a processor, processor modules or also as software for a processor. The device 24 can in particular be integrated into a vehicle-to-vehicle communication unit. For example, the device 24 be integrated into a standard communication unit and run as software on a processor of such a communication unit. It is also possible that the device according to the invention 24 is implemented as a mobile device or as software for a mobile device (smart phone app or similar).

Via the input interface 26th a wirelessly transmitted message is received from a vehicle-to-vehicle communication unit of the second vehicle. The input interface 26th for example, be connected to a corresponding low-energy short-range communication unit. Such a communication unit can, for example, implement the corresponding ETSI standard. The input interface 26th can also be designed for wireless communication itself.

In particular, the input interface 26th be designed to periodically receive a Cooperative Awareness Message (CAM). Such a standardized message contains different data which reflect the current state of the second vehicle. The data contained can in particular a steering wheel angle, a steering wheel angle confidence, a steering wheel angle change rate, a vehicle wheel angle, an acceleration, an acceleration confidence, a yaw angle, a yaw angle confidence, a speed, a direction and / or a position include. It goes without saying, however, that it is also possible to receive other vehicle-to-vehicle communication messages. In particular, application-specific messages can also be received in RAW format, which contain additional data.

In the classification unit 28 a pre-trained classification process is then carried out with the external vehicle data as input data. In particular, it has been found to be advantageous if a random forest classifier is used. This gives good results with mixed numerical and categorical features. It is therefore possible that on the one hand sensor data and on the other hand property data are processed. In the classification unit 28 a determination is made as to whether the second vehicle is about to make a turn. A prediction is made as to whether a turning process will be carried out. The external vehicle data allow recognition of the turning process at the latest when the turning process has already been initiated. The turning process is advantageously detected before it is initiated. A prediction can be made, so to speak, as to whether or not the second vehicle will make a turning maneuver.

The pre-trained classification method is preferably designed to recognize whether a turning process is being carried out and, in addition, also to make a statement about the direction in which the turning process is being carried out. It is particularly advantageous if at least one statement is derived as to whether a maneuver to turn left or right is imminent. But it is also possible to make an even more precise statement.

The classification process is pre-trained in order to provide the appropriate data evaluation and categorization. For this purpose, training data can be used in a training phase that was recorded in advance for several vehicles and annotated manually and / or automatically. This means that data is recorded and the time at which a turning process is carried out is determined. These data are used to train the classification method and, in particular, allow multiple decision trees to be generated in a random forest classifier. By using the correct training data, the reliability of the detection of the turning process can be increased.

Via the output unit 30th an output signal is provided which includes information on the turning process. In particular, the output signal includes binary information as to whether a turning process is being carried out or is imminent. In addition, it is possible that the output signal also outputs the determined turning direction. The output signal can be processed in a vehicle control unit, for example, in order to inform the driver of the vehicle whether a vehicle traveling ahead will perform a turning maneuver. It is also possible that an acceleration and / or braking process is acted on directly in an autonomous or partially autonomous vehicle in order to adapt it.

On the one hand, this can prevent accidents. On the other hand, it is also possible that the energy efficiency is improved when driving, since unnecessary braking and / or acceleration processes are avoided.

In the optionally available filter unit 32 a position and direction of travel of the second vehicle is determined. Based on this, it is possible to determine whether the second vehicle is a vehicle traveling ahead of the first vehicle. Information relating to a turning maneuver by the vehicle in front is particularly relevant because there is a risk of rear-end collisions and the driving style of the driver's own vehicle must be adapted when the vehicle in front is turning. The position in the direction of travel of the second vehicle can be determined, for example, from an evaluation of a direction of reception of the message by means of a directional antenna. It is also possible for a position and direction within the external vehicle data to be compared with a position and direction of the own vehicle in order to determine whether or not the second vehicle is a vehicle in front.

In the optionally available control unit 34 the output signal is processed directly in order to control an engine and / or a brake of the first vehicle when the second vehicle is going to perform a turning maneuver. In an autonomous or semi-autonomous vehicle in particular, this creates opportunities for improving safety and improving energy efficiency.

In the 3 is schematically a first vehicle 14th shown that is a vehicle-to-vehicle communication unit 18th and a device according to the invention 24 includes. The device 24 is to a vehicle control unit 36 connected, via which a motor and / or a brake of the first vehicle 14th can be controlled.

In the 4th a method according to the invention is shown schematically. The method comprises the steps of receiving S10 a wirelessly transmitted message, of determining S12 whether the second vehicle is making a turning maneuver and outputting S14 an output signal. The method can be designed, for example, as software for a vehicle-to-vehicle communication unit, a vehicle control device or a mobile device. For example, the method can also be implemented as a smartphone app that communicates with corresponding control units of the first vehicle.

The invention has been comprehensively described and explained with reference to the drawings and the description. The description and explanation are to be understood as an example and not restrictive. The invention is not limited to the disclosed embodiments. Other embodiments or variations will become apparent to those skilled in the art using the present invention and a careful analysis of the drawings, the disclosure, and the following claims.

In the claims, the words “comprising” and “having” do not exclude the presence of further elements or steps. The undefined article “a” or “an” does not exclude the presence of a plural. A single element or a single unit can perform the functions of several of the units mentioned in the patent claims. An element, a unit, an interface, a device and a system can be implemented partially or completely in hardware and / or in software. The mere mention of some measures in several different dependent patent claims should not be understood to mean that a combination of these measures cannot also be used advantageously. A computer program can be stored / distributed on a non-volatile data carrier, for example on an optical memory or on a semiconductor drive (SSD). A computer program can be distributed together with hardware and / or as part of hardware, for example by means of the Internet or by means of wired or wireless communication systems. Reference signs in the patent claims are not to be understood as restrictive.

List of reference symbols

10

system

12th

second vehicle

14th

first vehicle

16

Vehicle-to-vehicle communication unit of the second vehicle

18th

Vehicle-to-vehicle communication unit of the first vehicle

20th

road

22nd

Confluence

24

contraption

26th

Input interface

28

Classification unit

30th

Output unit

32

Filter unit

34

Control unit

36

Vehicle control unit

QUOTES INCLUDED IN THE DESCRIPTION

This list of the documents listed 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.

Non-patent literature cited

• The context is described in Tang et al., "Turn Prediction at Generalized Intersections", IEEE 2015 [0004]

Claims (13)

Device (24) for detecting a turning process of a second vehicle (12) in the vicinity of a first vehicle (14), comprising: an input interface (26) for receiving a wirelessly transmitted message from a vehicle-to-vehicle communication unit (16) of the second vehicle, the message comprising external vehicle data with information on a current state of the second vehicle; a classification unit (28) for determining whether the second vehicle will perform a turning process based on an evaluation of the external vehicle data in a pre-trained classification method; and an output unit (30) for outputting an output signal with information on the turning process when the second vehicle is going to perform a turning process. Device (24) after Claim 1 , wherein the pre-trained classification method is a random forest classifier. Device (24) according to one of the preceding claims, wherein the classification unit (28) is designed to determine a turning direction of the second vehicle (12) based on the evaluation of the external vehicle data in the pre-trained classification method; and the output unit (30) is designed to output the determined turning direction. Device (24) according to one of the preceding claims, with a filter unit (32) for determining a position and direction of travel of the second vehicle (12) based on the received message, the classification unit (28) being designed to only determine the turning process when the second vehicle is a vehicle traveling ahead of the first vehicle. Device (24) according to one of the preceding claims, with a control unit (34) for controlling a motor and / or a brake of the first vehicle (14) based on the output signal in order to adapt an acceleration and / or braking process of the first vehicle, if the second vehicle (12) will perform a turning maneuver. Device (24) according to one of the preceding claims, wherein the input interface (26) is designed to receive a standardized Dedicated Short Range Message, DSRC, in particular a Cooperative Awareness Message, CAM, a Decentralized Environmental Notification Message, DENM, a Signal Phase and Timing Message, SPAT and / or a Map Data Message, MAP. Device (24) according to one of the preceding claims, wherein the foreign vehicle data a steering wheel angle, a steering wheel angle confidence, a steering wheel angle change rate, a vehicle wheel angle, an acceleration, an acceleration confidence, a yaw angle, a yaw angle confidence, a speed, a direction and / or a Position. Device (24) according to one of the preceding claims, wherein the input interface (26) is designed to receive a RAW message and the external vehicle data include a friction confidence and / or a rate of rotation. Device (24) according to one of the preceding claims, wherein the classification method is pre-trained based on recorded third-party vehicle data of several vehicles. Vehicle-to-vehicle communication unit (18) with a device (24) according to one of the preceding claims. System (10) for detecting a turning process of a second vehicle (12) in the vicinity of a first vehicle (14), comprising: a vehicle-to-vehicle communication unit (16) in a second vehicle; and a device (24) or a vehicle-to-vehicle communication unit (18) according to one of the preceding claims. A method for detecting a turning process of a second vehicle (12) in the vicinity of a first vehicle (14), comprising the steps: Receiving (S10) a wirelessly transmitted message from a vehicle-to-vehicle communication unit (16) of the second vehicle, the message comprising external vehicle data with information on a current state of the second vehicle; Determining (S12) whether the second vehicle will perform a turning process based on an evaluation of the external vehicle data in a pre-trained classification method; and Outputting (S14) an output signal with information on the turning process when the second vehicle is going to perform a turning process. Computer program product with program code for performing the steps of the method according to Claim 12 when the program code is executed on a computer.

Images