DE102019211098B4 - Method, device and computer program for determining a traffic light phase of a traffic light of a traffic light system - Google Patents

Abstract

Method for determining a traffic light phase of a traffic light of a traffic light system, the traffic light being on a route of a vehicle, with: - (S301) calling up an event time (Tevent) at which the vehicle is at the traffic light; - (S302) calling up a first detection time (Ti) at which the vehicle (1) detects the traffic light for the first time; - (S303) retrieving a last detection time (Te) at which the vehicle (1) detects the traffic light for the last time; - (S304) retrieving a phase curve the traffic light, which extends from the first detection time (Ti) to the last detection time (Te); Has a predetermined cycle time and phase durations (Xrot) of the reference traffic light cycle are modeled as random variables.

Description

The present invention relates to a method, a device and a computer program for estimating a traffic light phase of a traffic light of a traffic light system.

There are automatic driving functions of a motor vehicle that process information from traffic lights and surrounding vehicles in order to react accordingly in traffic.

So describes the DE 10 2011 004 425 A1 a method for estimating the status of a traffic light system. It is determined that a vehicle is approaching a traffic light system, ie that a traffic light system is imminent on the route. According to the method, a check is made as to whether the driver intends to delay or is willing to delay. A status of the traffic light system is estimated as a function of this. For example, if the driver's intention or willingness to decelerate was recognized when approaching a traffic light, it can be assumed that the traffic light is emitting a red light signal. If there is a camera-based system for recognizing the surroundings that can recognize a traffic light system, the information about the signal color of the traffic light system, which is often not very reliable in itself, could also be taken into account when estimating the traffic light status.

From the DE 10 2012 207 620 A1 a system and a method for light signal detection are known. The method and the system can determine a location of a vehicle, acquire an image using a camera assigned to the vehicle, analyze the image together with the location of the vehicle and / or previously acquired information about the location of light signals or other objects (eg traffic signs) and using this analysis to determine an image of a light signal within the acquired image. The position of the signal can be determined and saved for later use. The identification of the signal can be used to provide an output such as the status of the signal (e.g. green light).

The DE 10 2016 013 972 A1 describes a driver assistance system with first receiving means for receiving traffic light signal phase information from a traffic light that is ahead on the path of an ego vehicle carrying the driver assistance system, second receiving means for collecting information about vehicles that are on the path between the ego vehicle and the traffic light , and computing means for forecasting the end of the traffic jam and the time of resolution of a traffic jam formed by these vehicles in front of the traffic light, for calculating and outputting a recommended speed at which the ego vehicle can travel in order to reach the end of the traffic jam after the traffic jam has been resolved.

The DE 20 2015 004 892 U1 describes a driver assistance system with an environment sensor for detecting the current phase of a traffic light, its distance from the driver assistance system or from a vehicle carrying the driver assistance system and the remaining time until the next phase change of the traffic light, a computing unit for predicting the phase when it arrives at the Traffic light based on the information from the environmental sensor and a user interface to display the forecast phase.

The DE 10 2013 223 022 A1 describes a method for estimating the probability of a predetermined light signal being displayed by a light signal system at discrete times.

The DE 10 2015 207 807 A1 describes a device for a vehicle for determining a probability that a signal display of a traffic control system has a predetermined state when the device approaches the signal display, a location system being provided, the location system being used to determine a time for the device to approach the signal display , a data memory being provided, the learned probabilities of the state of the signal display being stored in the data memory for predetermined times, a computing unit being provided, the computing unit being designed to determine a probability for the state of the signal display depending on the time of the approach read out from the data memory and further process.

The challenge is therefore that it is not recognized by the driver but by means of a method, a device and / or a computer program which information is relevant for an automatic vehicle function. In particular in the case of a vehicle that is driving on a section of the roadway controlled by a traffic light system and is moving towards the traffic light system, the relevant traffic light information such as the current traffic light phase of a traffic light must be reliably recorded or determined. A Traffic light phase is a signal that a traffic light can indicate. As a rule, a traffic light can display the traffic light phases “green”, “yellow”, “red” and “yellow and red”.

However, due to the limited opening angle of the camera optics, the vehicle-side camera systems cannot recognize a traffic light phase of a traffic light system if the vehicle is too close to the traffic light system. But the present traffic light phase of the traffic light system at the time of driving over / stopping at the traffic light system is particularly important, for example for various vehicle functions.

The object of the present invention is to provide a method, a device and a computer program which at least partially overcome the above-mentioned challenge.

This object is achieved by a method according to claim 1, a device according to claim 14 and a computer program according to claim 15.

Further advantageous embodiments of the invention emerge from the subclaims and the following description of preferred exemplary embodiments of the present invention.

• - Abrufen eines Event-Zeitpunkts, zu dem sich das Fahrzeug an der Ampel befindet;
• - Abrufen eines ersten Erfassungszeitpunkts, zu dem das Fahrzeug die Ampel erstmalig erfasst;
• - Abrufen eines letzten Erfassungszeitpunkts, zu dem das Fahrzeug die Ampel letztmalig erfasst;
• - Abrufen eines Phasenverlaufs der Ampel, der sich von dem ersten Erfassungszeitpunkt bis zum letzten Erfassungszeitpunkt erstreckt;
• - Bestimmen einer Ampelphase der Ampel zum Event-Zeitpunkt durch Vergleichen des abgerufenen Phasenverlaufs mit einem Referenzampelzyklus, wobei der Referenzampelzyklus eine vorbestimmte Umlaufzeit aufweist und Phasendauern des Referenzampelzyklus als Zufallsvariablen modelliert sind.

A first aspect of the disclosure relates to a method for determining a traffic light phase of a traffic light of a traffic light system. The traffic light is on a route of a vehicle. The procedure consists of the following steps:

• - Retrieving an event time at which the vehicle is at the traffic light;
• - Retrieving a first detection time at which the vehicle detects the traffic light for the first time;
• - Retrieving a last detection time at which the vehicle detected the traffic light for the last time;
• - Retrieving a phase curve of the traffic light, which extends from the first time of detection to the last time of detection;
• Determination of a traffic light phase of the traffic light at the time of the event by comparing the retrieved phase profile with a reference traffic light cycle, the reference traffic light cycle having a predetermined cycle time and phase durations of the reference traffic light cycle being modeled as random variables.

For the vehicle, the status (the current traffic light phase) of the traffic light that controls the vehicle's route is usually decisive. In the case of a multi-lane road, the lane in which the vehicle is moving is the route.

A traffic light can signal several traffic light phases, in particular a green phase, a yellow phase, a red phase and a red-yellow phase. As a rule, the traffic light also switches in the aforementioned phase sequence within a predetermined time. Such a complete phase sequence of the traffic light is also called a switching cycle. This predetermined time is called the traffic light cycle time. In other words, the cycle time indicates the time that a complete switching cycle of the traffic light requires.

The event time is a time when the vehicle is at the traffic light. The vehicle can stop at the traffic lights or drive over them. “At the traffic light” means in particular that the vehicle is on its route in the direction of travel up to a maximum of 25 m, preferably up to 15 m, in front of the traffic light. The traffic light position is detected by a detection device on the vehicle. The acquisition device comprises a camera device for acquiring image information and a GPS system (such as, for example, a navigation system) for acquiring position data of the vehicle and the traffic light. The traffic light position is then sent, for example, to a database and stored there.

In addition, the first and the last recording time are called up. At these times, the traffic light is detected by the vehicle (in other words, by the detection device on the vehicle side) for the first time or for the last time. So when the vehicle moves towards the traffic light, the first time of detection (in terms of time) corresponds to the first time the vehicle viewed the traffic light. Accordingly, the last time of detection is the last time the vehicle viewed the traffic light. In terms of time, the last time of detection is usually before the time of the event, since due to a limited opening angle of a camera of the detection device, the traffic light can no longer be detected / visible at the time of the event. However, there are also situations in which the traffic light is still in the detection area of the detection device and the event time therefore corresponds to the last detection time.

In one step, the traffic light phase of the traffic light at the time of the event is determined. The phase progression called up is compared with the reference traffic light cycle. The reference traffic light cycle is a modeled complete phase curve of a switching cycle with a predetermined cycle time. A reference traffic light cycle is used, since the vehicle is generally unable to acquire complete information about the traffic light, such as the cycle time or the phase duration of the traffic light.

In order to model a reference traffic light cycle as generally as possible, the phase durations of the reference traffic light cycle are modeled as random variables. These random variables are distributed in corresponding value ranges. For example, the red phase duration of the reference traffic light cycle can have a value from the range [10s; 84s]. Other value ranges are also possible. In this way, possible value ranges for phase durations of the reference traffic light cycle can be derived from the guidelines for traffic light systems (RiLSA).

The unknown phase durations of the traffic lights detected by the vehicle can therefore be taken into account by means of the reference traffic light cycle. By comparing the retrieved phase profile with a reference traffic light cycle, a probability of the presence of a corresponding traffic light phase at the time of the event can be determined. In other words, it can be estimated whether a certain traffic light phase is present when driving over the traffic light or when stopping at the traffic light, even if the detection device cannot detect the traffic light.

In a variant, the predetermined cycle time can also be modeled as a random variable. This allows the reference traffic light cycle to be modeled in an even more general way.

In particular, the random variable representing the predetermined cycle time can have a value in the range of [30s; 120], whereby this range of values for the cycle time is taken from the RiLSA. Other, in particular country-specific, value ranges for the random variable are also possible.
The random variables for the cycle time and the phase durations of the reference traffic light cycle can be evenly distributed in their corresponding value ranges. In other words, each cycle time or phase duration from the corresponding value ranges has the same corresponding probability. Alternatively, the distributions can be different. In addition, it is possible to make the probability that a cycle time or a phase duration assumes a specific value from the corresponding value range dependent on the day, time and / or location.

Furthermore, the phase durations of the reference traffic light cycle can be dependent on the predetermined cycle time. In other words, the phase durations of the reference traffic light cycle can each be specified as a fraction of the predetermined cycle time. In total, the fractions result in a value of 1.

In an alternative, the reference traffic light cycle can consist of a red phase and a green phase. The comparatively short yellow phase can thus be assigned to the red phase and the red-yellow phase to the green phase. The reference traffic light cycle can thus only consist of a green phase and a red phase. The phase durations of the reduced reference traffic light cycle, i.e. in this case the green phase duration and the red phase duration, can each be modeled as random variables (depending on one another). The reference traffic light cycle is thus made less complex. It then follows from this that the circulation time of the reference traffic light cycle can be determined from the green phase duration and the red phase duration of the reference traffic light cycle.

• - Ermitteln eines Phasenwechsel-Zeitpunkts, zu dem ein Phasenwechsel der Ampel vor dem letzten Erfassungszeitpunkt erfolgt; und
• - Ermitteln einer Wahrscheinlichkeit, dass die bestimmte Ampelphase zum Event-Zeitpunkt vorliegt, durch Vergleichen eines Zeitintervalls, das sich von dem Phasenwechsel-Zeitpunkt und zu dem Event-Zeitpunkt erstreckt, mit der Dauer einer entsprechenden Ampelphase des Referenzampelzyklus (unter der Berücksichtigung, dass die Dauer der entsprechenden Ampelphase als Zufallsvariable modelliert ist), wobei die entsprechende Ampelphase einer aufgrund des Phasenwechsels vorliegenden Ampelphase entspricht.

In a variant, the determination of the traffic light phase can further include:

• - Determination of a phase change point in time at which a phase change of the traffic light takes place before the last detection point in time; and
• - Determining a probability that the specific traffic light phase is present at the event time by comparing a time interval that extends from the phase change time and the event time with the duration of a corresponding traffic light phase of the reference traffic light cycle (taking into account that the Duration of the corresponding traffic light phase is modeled as a random variable), the corresponding traffic light phase corresponding to a traffic light phase present due to the phase change.

A phase change occurs when the traffic light phase of the traffic light changes (plausibly). Possible phase changes can be from a green phase to a yellow phase, from a yellow phase to a red phase, from a red phase to a red-yellow phase or from a red-yellow phase to a green phase. Here it is a phase change immediately before the first acquisition time or the last acquisition time.

In one step, the probability that the traffic light phase determined above is present at the time of the event can be determined. The time interval (duration) that extends from the phase change point in time and the event point in time is compared with the duration of the corresponding traffic light phase of the reference traffic light cycle. The corresponding traffic light phase corresponds to a traffic light phase that is present due to the phase change that occurs. If, for example, there is a red traffic light phase at the recorded traffic light after a phase change, the duration of the red traffic light phase of the reference traffic light cycle is used for comparison.

This traffic light phase, which is present after the phase change that has occurred, can also be determined from the phase profile that has been called up. If, for example, the phase change takes place between the first detection time and the last detection time, the traffic light phase present after the phase change can, so to speak, be read from the retrieved phase curve. If the phase change occurs before the first detection time (and therefore there is no phase change in the retrieved phase curve), it can be deduced that the traffic light phase present in the retrieved phase curve (and also the only one) is due to the phase change.

It should be noted here that the duration of the corresponding traffic light phase of the reference traffic light cycle is a random variable. The result, i.e. that a certain traffic light phase is present at the time of the event, occurs with a probability, which in turn is dependent on the probabilities, that the duration of the corresponding traffic light phase of the reference traffic light cycle assumes corresponding values from the corresponding value range for the duration.

In one variant, the phase change point in time can lie within the retrieved phase profile. The phase change and the phase change time can therefore be detected by the vehicle-side detection device and are therefore known. Therefore, the time interval between the phase change point in time and the event point in time can easily be determined.

• - Durchführen eines der folgenden Schritte in Abhängigkeit der Ampelphase des abgerufenen Phasenverlaufs:
• - Verwenden des letzten Erfassungszeitpunkts als Phasenwechsel-Zeitpunkt für das Ermitteln der Wahrscheinlichkeit, dass die bestimmte Ampelphase zum Event-Zeitpunkt vorliegt; oder
• - Modellieren eines Phasenwechsel-Zeitpunkts als Zufallsvariable, wobei der modellierte Phasenwechsel-Zeitpunkt vor dem ersten Erfassungszeitpunkt liegt.

In a further variant, the phase change time can be before the first detection time and the determination of the traffic light phase can also include:

• - Carry out one of the following steps depending on the traffic light phase of the phase sequence called up:
• - Using the last detection time as a phase change time for determining the probability that the specific traffic light phase is present at the event time; or
• - Modeling a phase change point in time as a random variable, the modeled phase change point in time lying before the first detection point in time.

The phase change can also take place before the first acquisition time. Consequently, the retrieved phase curve cannot have a phase change, since the retrieved phase curve can only be retrieved for a period between the first and the last acquisition time. In the absence of a phase change within the retrieved phase curve, the retrieved phase curve has only one traffic light phase.

Then, depending on the present traffic light phase of the retrieved phase profile, the last acquisition time can be used as the phase change time or a phase change time can be modeled as a random variable.

In particular, if the retrieved phase profile indicates a traffic light phase of yellow or red-yellow, it can be assumed that the phase change time corresponds to the last time of detection. This assumption takes into account that a yellow phase and a red-yellow phase have comparatively short durations. Because after a traffic light switches to one of these phases, the next traffic light phase usually follows within a few seconds, i.e. a red phase after the yellow phase or a green phase after the red-yellow phase.

If the phase profile called up is a green or a red phase, the phase change point in time can be modeled as a random variable. In particular, it is assumed here that the modeled phase change point in time lies in a predetermined time interval, which is in particular from a point in time that is one cycle time before the last detection point in time to the first Acquisition time extends. It should also be noted here that the result, ie that a certain traffic light phase is present at the time of the event, occurs with a probability that is now additionally dependent on probabilities, namely that the phase change time assumes corresponding values from the predetermined time interval (value range) . Additionally dependent to the extent that the above-mentioned probabilities that the corresponding phase durations of the reference traffic light cycle assume corresponding values from the corresponding value ranges for the durations must also be taken into account.

• - Durchführen eines der folgenden Schritte in Abhängigkeit der Ampelphase des abgerufenen Phasenverlaufs:
• - Verwenden des letzten Erfassungszeitpunkts als Phasenwechsel-Zeitpunkt für das Ermitteln der Wahrscheinlichkeit, dass die bestimmte Ampelphase zum Event-Zeitpunkt vorliegt; oder
• - Modellieren eines Phasenwechsel-Zeitpunkts als Zufallsvariable, wobei der modellierte Phasenwechsel-Zeitpunkt vor dem ersten Erfassungszeitpunkt liegt.

In an alternative, the determination of the traffic light phase (described above) can further include the following if the phase change time is before the first detection time:

• - Carry out one of the following steps depending on the traffic light phase of the phase sequence called up:
• - Using the last detection time as a phase change time for determining the probability that the specific traffic light phase is present at the event time; or
• - Modeling a phase change point in time as a random variable, the modeled phase change point in time lying before the first detection point in time.

Otherwise, the time of the phase change lies within the retrieved phase curve and can be determined as described above.

In an alternative, if there are at least three phase change times in the retrieved phase curve, at least one can be derived from the cycle time of the (detected) traffic light and phase duration of the (detected) traffic light, taking into account the at least three phase change times. The determination of the traffic light phase at the time of the event can thus be made more precise by adapting the cycle time or phase duration of the reference traffic light cycle to the actual cycle time or phase duration of the detected traffic light. In particular, it is again assumed that a yellow phase and a red-yellow phase have comparatively short durations.

This applies above all to a switching cycle that has the four traffic light phases mentioned so far. In general, the number of existing phase change times must be at least one less (i.e. one less or equal to or higher) than the number of traffic light phases of any switching cycle in order to determine at least one of the cycle time of the (detected) traffic lights and phase durations of the ( recorded) of any switching cycle.

Furthermore, the traffic light phase at the time of the event can correspond to a green phase of the traffic light or a red phase of the traffic light. This reduction to the results relevant to the vehicle simplifies the complexity of determining (the probability) of the traffic light phase at the time of the event.

In an alternative, the reference _{traffic light cycle can take into account a transition time (t trans} ) which follows a phase change to the red phase and in which the red phase is viewed as a green phase. This transition period takes into account that some drivers want to cross the traffic light quickly after a phase change to the red phase of a traffic light. This corresponds to the behavior of some drivers who want to get across the intersection quickly, even though the red phase of the traffic light is already present.

In one variant, the time of the event, the first time of recording, the last time of recording, the phase profile called up and the reference traffic light cycle can be stored in a database.

Furthermore, the traffic light information and / or the time of the event can be detected by a detection device on the vehicle. This enables the vehicle to record the traffic light information that is relevant to itself.

A second aspect of the disclosure relates to a device for determining a traffic light phase of a traffic light of a traffic light system, the traffic light being on a route of a vehicle, the device being designed and set up to carry out one of the methods described above.

A third aspect of the disclosure relates to a computer program for determining a traffic light phase of a traffic light of a traffic light system, the traffic light being on a route of a vehicle, the computer program being designed to cause a data processing device to execute one of the methods described above when it is executed.

• 1 schematisch einen durch eine Ampelanlage gesteuerten Fahrbahnabschnitt;
• 2 schematisch ein Verfahren zum Aufbereiten von Daten über die Ampelanlage;
• 3 schematisch ein Ampelbild der in 1 gezeigten Ampelanlage;
• 4 schematisch ein beispielhaftes Ampelbild der in 1 gezeigten Ampelanlage;
• 5 schematisch ein Verfahren zur Gruppierung von Ampeln der in 1 gezeigten Ampelanlage;
• 6 schematisch ein Verfahren zum Bestimmen einer Ampelphase einer Ampel; und
• 7 schematisch und beispielhaft zwei Phasenverläufe.

Embodiments of the invention will now be described by way of example and with reference to the accompanying drawings. It shows:

• 1 schematically a section of the roadway controlled by a traffic light system;
• 2 schematically a method for processing data about the traffic light system;
• 3 schematically a traffic light image of the in 1 shown traffic lights;
• 4th schematically an exemplary traffic light image of the in 1 shown traffic lights;
• 5 schematically a method for grouping traffic lights of the in 1 shown traffic lights;
• 6th schematically a method for determining a traffic light phase of a traffic light; and
• 7th schematically and by way of example two phases.

In 1 is schematically a road section Fb shown by a traffic light A. is controlled. The road section Fb has three lanes F1, F2, F3, which in turn are represented by four corresponding traffic lights tp ₁ , tp ₂ , tp ₃ , tp ₄ , tp _{5 of} the traffic light system A. being controlled. The traffic lights A. can through a the road section Fb driving vehicle 1 are recorded. In particular, the individual traffic lights tp ₁ -tp ₅ the traffic lights A. detected. The vehicle points to the detection 1 a detection device 3 on, which is designed in such a way, a phase curve and / or position of each traffic light tp ₁ -tp ₅ capture. Furthermore, the detection device 3 at least one of the position, position progression and orientation of the vehicle 1 capture. The detection device 3 can, for example, comprise a camera device and a GPS device. The vehicle 1 has a radio interface 5 to communicate with a server / database (not shown) and to exchange data.

In the context of the disclosure, “detection by the vehicle 1” also means “detection by the vehicle-side detection device 3”.

A phase curve of a traffic light is also referred to as a signal curve. A phase progression indicates phase durations as well as corresponding traffic light phases. A phase progression for a traffic light can be represented as a large number of time-traffic light phase pairs. A traffic light phase is a signal that a traffic light can display. As a rule, a traffic light has the (traffic light) phases “green”, “yellow”, “red” and “yellow and red”, whereby the traffic light phases are usually also displayed in this order by the traffic light. Other orders are also possible.

It goes without saying that the vehicle 1 can also exemplify a large number of vehicles, so that the traffic light system A. can be detected by a variety of vehicles. This is the case for every crossing over the traffic lights A. by a vehicle 1 A vehicle-specific, or otherwise called crossing-specific, data record can be created from the multitude of vehicles. This crossing-specific data record is described later.

With the present traffic light system A. the traffic lights tp ₁ , tp _{2 control} the lane F1 and the traffic lights tp ₃ , tp _{4 control} the lanes F2, F3. The traffic lights tp ₁ , tp ₂ thus form a first traffic light group and the traffic lights tp ₃ , tp _{4 form} a second traffic light group. The traffic lights tp ₁ , tp _{2 of} the first traffic light group therefore have the same first phase profile and the traffic lights tp ₃ , tp _{4 of} the second traffic light groups have the same second phase profile. In other words, the respective traffic lights of the traffic light groups have the same phase progression. Finally, the lane F3 can also be controlled by a separate right-turn traffic light tp _5.

In the 2 is a method for creating a traffic light image 7th shown. With traffic light image 7th is meant a data record that can be stored in a database, for example, and positions of traffic lights in a traffic light system, e.g. the one in 1 shown traffic lights A. with the traffic lights tp ₁ -tp ₅ include. In other words, the position of a traffic light system, and in particular its traffic lights, can be displayed / represented by means of a traffic light image.

In a step S101, traffic light information, so-called traffic light objects, n _c retrieved, for example from an existing database. Every traffic light object n _c is representative of a corresponding traffic light of the traffic lights tp ₁ -tp ₅ and therefore includes at least position data of the corresponding traffic light. Furthermore, the traffic light objects include n _c also information about the traffic light phases of the corresponding traffic lights.

Every traffic light object n _c is created specifically for the crossing. Here "crossing" means that the vehicle 1 the lane section from the multitude of vehicles Fb drives on to the traffic lights A. closes and the traffic lights A. then crossed. During such a passage c, the vehicle detects 1 by means of the detection device 3 , especially continuously, the positions and the phases of the individual traffic lights tp ₁ -tp ₅ .

Each of the traffic light objects n _c , in particular the traffic light position represented by the corresponding traffic lights tp ₁ -tp ₅ , can for example be represented as a point (traffic light object point) p _n in a global coordinate system. The entirety of all traffic light objects n _c can therefore be represented as a point cloud {p _n } with n = 1 .... N in a global coordinate system K.

For the in 2 The methods shown, the terms “traffic light objects” and “traffic light object points” can be used interchangeably, since the traffic light object points correspond to a representation of the traffic light objects in the global coordinate system K.

It is possible to display the traffic light object points in a “semi-global” coordinate system that is valid for a predetermined area around a given reference point (origin of the “semi-global” coordinate system). The predetermined area can also extend several hundred meters from the reference point in all possible directions, in particular in the east, north and above (away from the earth's surface).

The traffic light object points p _n can have an undesirably high scatter in a longitudinal direction along a course of the lane section Fb exhibit. The traffic light object points p _n can therefore be pre-filtered in order to create the traffic light image. This pre-filtering is described below.

In a step S102, a course d _{traj of} the lane section is determined Fb retrieved. This is the course of the road that leads to the traffic lights A. feeds. _{Movement history data} r traj of the vehicle can be used for this purpose 1 can be retrieved, the at least one of the position, the position history and the orientation of the vehicle 1 represent. In other words, the course d _{traj of} the road section Fb can be determined from the movement _history data r traj. The course d _{traj of} the roadway section can be used here Fb also representative of an orientation of the vehicle 1 be.

The course r _{traj of} the road section Fb becomes like the traffic light objects n _c also, shown in the global coordinate system K.

It is possible for step S101 to take place before, after or at the same time as step S102.

In a step S103, the traffic light object points p _{n are set} to a longitudinal direction d _{traj_intersec of} the course d _{traj of} the lane section Fb projected, the longitudinal direction d _{traj_intersec} the course d _{traj of} the road section Fb represents on the traffic light line. The traffic light line is the position along the course of the road section Fb , at which the traffic lights A. is located. This can be, for example, that position along the course d _{traj of} the roadway section at which (in the direction of travel of the vehicle 1 ) the vehicle for the first time at a traffic light object n _c or traffic light point p _n passes. In particular, the traffic light line can thus be determined specific to the crossing (and thus also specific to the vehicle). For the method described here, a cross-crossing traffic light line can also be formed, in that the mean of the traffic light lines is formed from each crossing. In step S103, the traffic light line is also determined / retrieved.

Since the traffic light object points p _n and the course d _{traj of} the lane section Fb can be represented in the coordinate system K (as vectors), the projection for each traffic light object p _{n takes place} according to the following formula: $$x{\text{'}}_{p_n}={\overrightarrow{p}}_n{}^T{\overrightarrow{d}}_{traj\_intersec}$$

von dem Medianwert über allen projizierten Ampelobjektpunkte {x'_{p n} } liegen Es werden also die projizierten Ampelobjektpunkte {x'_{p n} } verworfen, die folgende Bedingungen erfüllen: $$x{\text{'}}_{p_n}<\underset{n}{\text{median}}\left\{x{\text{'}}_{p_n}\right\}+2{\sigma }_{x{\text{'}}_{p_n}}$$

In a step S104, those lights are discarded object points p _n, the more than two standard deviations in the longitudinal direction d _traj _ _intersec ${\sigma }_{x{\text{'}}_{p_n}}$

of the median value over all projected traffic light object points {x ' _{p n} } lie The projected traffic light object points {x ' _{p n} } discarded that meet the following conditions: $$x{\text{'}}_{p_n}<\underset{n}{\text{median}}\left\{x{\text{'}}_{p_n}\right\}+2{\sigma }_{x{\text{'}}_{p_n}}$$

abhängig ist. Dieser vorbestimmte Grenzwert kann bspw. ein fester (absoluter) Wert sein.In other words, traffic light object points p _n are sorted out, taking into account a scatter (in the longitudinal direction) of all traffic light object points {p _n }. Alternatively, it is also possible to provide a predetermined limit value which, in particular, does not depend on the standard deviation ${\sigma }_{x{\text{'}}_{p_n}}$

is dependent. This predetermined limit value can, for example, be a fixed (absolute) value.

The discarded traffic light object points p _n represent traffic lights that are not on the for the lane section with a high degree of probability Fb valid traffic light line of the traffic light system A. such as pedestrian lights, traffic lights for lane sections from other directions)

Steps S102, S103 and S104 represent the above-mentioned pre-filtering of the traffic light object points p _n .

Furthermore, the traffic light objects p _n remaining after the pre-filtering are subdivided into groups by means of a clustering method. A clustering method here means a method for discovering similarity structures in (large) databases, such as the traffic light objects p _n present here. The groups of "similar" objects found by the clustering process are called clusters.

In a step S105, the traffic light object points p _{n are} projected onto an image plane E which is perpendicular to the course d _{traj of} the lane section Fb is at the level of the traffic light line. In other words, the image plane E is that plane which is perpendicular to the direction of travel of the vehicle 1 and along the course d _{traj of} the road section Fb is at the level of the traffic light line.

des Verlaufs d_traj des Fahrbahnabschnitts Fb projiziert, wobei die Querrichtung $d_{traj\_intersec}^{\perp }$

des Verlaufs d_traj des Fahrbahnabschnitts Fb auf der Ampellinie darstellt und in einer Ebene des Fahrbahnabschnitts Fb liegt. Höhen der Ampeln tp₁-tp₅ werden ebenfalls durch die Ampelobjektpunkte p_n repräsentiert und können (ohne in eine Richtung projiziert zu werden) übernommen werden. Demnach lässt sich jedes auf die Bildebene E projiziertes Ampelobjekt p'_n wie folgt durch einen 2D-Vektor des Koordinatensystems K darstellen: $$\overrightarrow{p}{\text{'}}_n=\left[\begin{array}{c}{\overrightarrow{p}}_n{}^T{\overrightarrow{d}}_{traj\_intersec}\\ {\overrightarrow{p}}_n{}^T\overrightarrow{k}\end{array}\right]$$

Each traffic light object point is projected onto the image plane E in a transverse direction $d_{traj\_intersec}^{\perp }$

of the course d _{traj of} the road section Fb projected with the transverse direction $d_{traj\_intersec}^{\perp }$

of the course d _{traj of} the road section Fb represents on the traffic light line and in a plane of the road section Fb lies. Heights of the traffic lights tp ₁ -tp ₅ are also represented by the traffic light object points p _n and can be adopted (without being projected in one direction). Accordingly, each can be projected onto the image plane E traffic light object p _'n by one as follows 2D vector of the co-ordinate system K represent: $$\overrightarrow{p}{\text{'}}_n=\left[\begin{array}{c}{\overrightarrow{p}}_n{}^T{\overrightarrow{d}}_{traj\_intersec}\\ {\overrightarrow{p}}_n{}^T\overrightarrow{k}\end{array}\right]$$

der kanonische Basisvektor des Koordinatensystems K.Here is $\overrightarrow{k}$

is the canonical basis vector of the coordinate system K.

In a step S106, a (first) clustering method is carried out _{on the traffic light object points p 'n projected in the direction of travel. The traffic light object points p 'n} projected in the direction of travel can thus be grouped in order to recognize / determine traffic light groups. The traffic light object points p ' _n are clustered with regard to their position (data).

Various clustering methods can be used for this. As a rule, the well-known “Density-Based Spatial Clustering of Applications with Noise” (DBSCAN) algorithm is used. This algorithm works on a density basis and can recognize several clusters. Noise points are ignored and returned separately.

The algorithm forms clusters of so-called "densely connected" points, i.e. of points that are not further than a specified distance (neighborhood length) from a "core point" in the same cluster t. A core point is a point which is closer than the neighborhood length ε to at least a predetermined minimum number minPts of further core points in the same cluster t.

For the present method, a neighborhood length ε _{1 is} two meters and the predetermined minimum number minPts = 3. In the present case, a comparatively high neighborhood length ε ₁ = 2m is selected in order to avoid the comparatively high scatter in the traffic light objects p ' _n , in particular with regard to their positions (in which Coordinate system) to counteract.

Other values are also possible in order to adapt the clustering method to different circumstances.

The result of the clustering method from step S106 are clusters t with t = 1... T from the traffic light object points p ' _n projected in the direction of travel. Each of these clusters t represents a traffic light object group that is representative of a “physical” traffic light tp ₁ -tp ₅ is.

In a step S107, the traffic light objects belonging to a traffic light cluster t, that is to say a traffic light t, are referenced with the corresponding index t, so that each projected traffic light object point has p ' _{n c, t} lets specify. The projected traffic light object point p ' _{n c, t} can therefore be assigned to corresponding traffic light clusters t and corresponding crossings c.

It should be noted that a traffic light group t can include more than one traffic light object (and thus a traffic light position) which can originate or originate in the same crossing c. The reason for this is that double detection by the detection device 3 is possible, these traffic light objects being at least partially time-overlapping or not time-overlapping.

In a step S108, the results of the first clustering method are post-filtered.

For this purpose, traffic light object points p ' _n that cannot be assigned to any traffic light group t and thus represent noise are discarded.

Alternatively or additionally, in step S108 traffic light clusters t with a number of traffic light object points p ' _n that is smaller than a predetermined minimum number, for example one fifth of the number of the second largest traffic light cluster, are discarded. The traffic light object points p ' _{n of} the relevant traffic light clusters t are also discarded. This ensures that no traffic light clusters t that are formed from scattered traffic light object points p ' _n that do not correspond to any physical traffic light t are taken into account for the further method.

As an alternative or in addition, a main component analysis is carried out for each traffic light cluster t in step S108.

Due to the comparatively high first neighborhood length ε ₁ = 2m, it can happen that two closely spaced traffic lights t are combined into a cluster t by the clustering process. As a result, the two closely adjacent traffic lights t are not recognized as such, but are only represented by a traffic light cluster t and thus only shown as one traffic light t. Such a traffic light cluster t can be determined by means of the principal component analysis.

Principal component analysis is a method of multivariate statistics. It is used to structure, simplify and visualize extensive data sets by approximating a large number of statistical variables using a smaller number of linear combinations (main components) that are as meaningful as possible.

For the main component analysis of a traffic light cluster t, all traffic light object points p ' _{n associated with the corresponding traffic light cluster t are determined t} determined. The following formula can be used for this: $$\left\{p{\text{'}}_{n_t}\right\}={\displaystyle \underset{c=1}{\overset{\mathrm{C.}}{\cup }}\left\{p{\text{'}}_{n_{c,t}}\right\}}$$

In the above formula, c stands for the corresponding crossing from c = 1 ... C.

From this traffic light object point cloud {p ' _{n t} } Using the principal component analysis, the principal components (principal directions) and the corresponding (maximum) eigenvalues λ _{max t} determined. An eigenvalue λ _maxt corresponds to the length of a corresponding main component. The principal component analysis is used to determine how far a traffic light cluster t is “pulled apart”.

This means that those traffic light clusters can be identified as “double clusters” whose main component is longer than a predetermined value. In other words, double clusters exist when eigenvalues λ _{maxt of} the double cluster exceed a predetermined eigenvalue limit value.

In the present case, the predetermined eigenvalue limit value can be twice the median value of all eigenvalues λ _maxt over all traffic light clusters t. To identify the double clusters, the following condition can apply: $${\lambda }_{ma{x}_t}>2\underset{t}{{\displaystyle median}}\left\{{\lambda }_{ma{x}_t}\right\}$$

It is also possible to use other predetermined eigenvalue limit values. For example, the predetermined eigenvalue limit value can be a different factor of the median value of the eigenvalues λ _maxt over all traffic light clusters than the above twofold factor.

The clustering process (second clustering process) is applied again to the identified double clusters. In particular, the DBSCAN algorithm is used here again, but with a neighborhood length ε ₂ which corresponds to half of the first neighborhood length ε _{1 of} the first clustering method. In the present case, the neighborhood length is ε ₂ = 1m. The neighborhood length ε _{2 of} the second clustering method can also have a different value as long as it is lower than the first neighborhood length ε ₁ . It is important that the second clustering process again detects “sub-clusters” within the double clusters.

The resulting “sub-clusters” are treated as traffic light clusters t. From the DBSCAN algorithm with the neighborhood _{length ε 2 of} the second clustering method, traffic light object points {p ' _n } that cannot be assigned to a “sub-cluster” are discarded.

In a step S109, all crossing-specific data records that have less than a predetermined minimum number of traffic light objects after the above steps are discarded. It is therefore checked whether, after the previous steps, it can happen that there is only a predetermined minimum number of traffic light objects from one pass c.

Here this predetermined minimum number is two. In other versions, a different predetermined minimum number of traffic light objects is also possible. What is decisive is that this minimum number is selected in such a way that it is ensured that a reliable assignment of the traffic light object points p ' _{n t} to the traffic light clusters t can take place.

If a traffic light cluster t no longer has any traffic light objects as a result of step S109, these “empty” traffic light clusters are likewise discarded.

In a step S110, the traffic light image becomes 7th created. For this purpose, a spatial mean value is formed from each of the remaining traffic light clusters t, which then represents a position of the physical traffic light t, which is represented by the corresponding traffic light cluster t. _{The traffic light object points p 'n} remaining from the previous steps S101 to 108 _t are the physical traffic lights tp ₁ -tp ₅ assignable.

A result of the previously described method in FIG 3 shown. Here is the traffic light picture 7th shown that the traffic lights A. out 1 represents. The traffic light objects n _{c, t} from all crossings c are as (cross-shaped) point cloud {p ' _{n t} } shown. It can be seen that in the traffic light image 7th the traffic light objects n _{c, t} are grouped into four clusters t = 1 to t = 4 (clusters t ₁ -t _4). Each of the clusters corresponds to one of the traffic lights tp ₁ to tp ₄ . The hatched diamonds represent centers of the clusters t (spatial mean values), which each represent a spatial traffic light position of the corresponding physical traffic light.

The traffic light objects that were recorded for traffic light tp ₅ are assigned to the cluster for traffic light tp ₄ . This is because the traffic light tp ₅ (as a right-turning traffic light) is located very close to or even adjacent to the traffic light tp ₄ . In the 3 The horizontal axis (x-axis) shown represents a direction transverse to the road section and the vertical axis (y-axis) a direction perpendicular to the earth's surface towards the sky. 3 corresponds to a view from the vehicle, so to speak 1 on the traffic lights A. . The axis values are given in meters. For example, the traffic light cluster t _{2 is} approx. 4 m to the left and 7 m above the vehicle 1 .

In the 4th is an example and schematic of another traffic light image 7 ' the traffic lights A. shown for further explanation. In the traffic light picture 7 ' is a variety of traffic light objects n _{c, t} shown representative of the traffic lights tp ₁ -tp ₅ the traffic lights A. are.

As described above, the traffic light objects n _{c, t} recorded specific to the crossing. As a rule, "crossing" means that the vehicle 1 the lane section from the multitude of vehicles Fb drives on to the traffic lights A. approaches and then crosses it. During a passage c, the vehicle is recorded, in particular continuously 1 the positions and the phases of the individual traffic lights tp ₁ -tp ₅ . Using the in 2 The procedure shown is from the totality of the traffic light image 7 ' included traffic light objects n _{c, t} Traffic light clusters t ₁ , t ₂ , t ₃ , t _{4 are} formed, the traffic light clusters t ₁ -t _{4 being} the physical traffic lights tp ₁ -tp ₅ the traffic lights A. correspond.

So can be seen from the traffic light image 7 ' determine that in this example the traffic light objects n _{c, t} can be created from four passes c = 1 ... 4. It should be mentioned that these four crossings c = 1 ... 4 not only by the same vehicle 1 , but also may have been carried out by different vehicles from the multitude of vehicles. In the 4th are the traffic light objects n _{c, t} shown in different shapes, each shape standing for a corresponding passage c. It can thus be seen that a traffic light object n _1,1 or n _{1,3 is} recorded for _{each traffic light tp 1 and traffic light tp 3} from a crossing c = 1 and assigned to the traffic light _{clusters t 1} , t ₃ . For the same crossing c = 1 tp for the traffic light and the traffic light ₂ tp ₄ each two traffic objects n n _1.2 and _1.4 and the traffic light detected cluster t _2, t _4, respectively.

The fact that more than one traffic light object is generated for a traffic light during a passage may be due, among other things, to the fact that the detection device 3 of the corresponding vehicle, the corresponding traffic light is incorrectly recorded and several traffic light objects are created for this traffic light and assigned to it. It is also possible that during the acquisition by the acquisition device 3 a line of sight between the sensing device 3 and the corresponding traffic light is interrupted by an obstacle, so that after the obstacle has disappeared, the corresponding traffic light is detected by the detection device 3 is recorded again and therefore a new traffic light object is generated for it and assigned to this traffic light.

It should be noted that the in 4th shown different shapes of the traffic light objects n _{c, t} serve only for clarification and the disclosure is not limited to the fact that the traffic light objects n _{c, t} in a special form in the traffic light image 7 ' being represented.

In 5 is a method of grouping traffic lights tp ₁ -tp ₅ shown. The groups are formed in such a way that each traffic light within a (traffic light) group has the same phase behavior or a similar phase progression.

In a step S201, the traffic light image 7th or the traffic light objects n _{c, t} retrieved that or the traffic light system A. represents. As described above, the traffic light image 7th the traffic light objects recorded over several crossings c n _{c, t} for every traffic light tp ₁ -tp ₅ the traffic lights A. on. That in the 5 The method shown does not specifically require the traffic light image 7th the traffic lights A. (as an input data record). Traffic light images representing other traffic light systems can also be used.

In a step S202, inconsistent traffic lights are identified. It should be noted that the physical traffic lights tp ₁ -tp ₅ by corresponding traffic light clusters t ₁ -t ₄ in the traffic light image 7th be represented.

A traffic light is inconsistent if the phases of individual traffic light objects that are assigned to the same traffic light appear to be inconsistent with one another. This usually indicates that there is a (right or left) turning light at the traffic light concerned, which only lights up occasionally. These turning lights are usually so close to the traffic light concerned that the in 2 The method shown wrongly assigns the traffic light object representing the turning traffic light to the traffic light concerned. This situation is, for example, in the 1 shown traffic lights A. in front. Here the traffic light tp _{5 is arranged} close to the traffic light tp ₄ , so that the in 2 The method shown does not recognize two traffic light clusters for the traffic lights tp ₄ and tp ₅ , but only the traffic light cluster t ₄ . In other words, the two traffic lights tp ₄ and tp ₅ are marked by the in 2 The procedures shown are not kept apart.

In the exemplary traffic light picture 7 ' the traffic light object n _2.4 and traffic light object n _{2.5 of} the traffic lights tp ₄ or tp _{5 are} so close to one another that they are assigned to the _{same traffic light cluster t 4.} Hence this is in 4th The traffic light object n _{2.5 shown} (the traffic light object created for a crossing c = 2 for the traffic light t ₅ ) in the _{traffic light cluster t 4} .

bestimmt. Die Korrelationsfaktoren ${\rho }_{n_{c,t},n{\text{'}}_{c,t}}\left({\rho }_{obj}\right)$

werden wie folgt gebildet: $${\rho }_{n_{c,t},n{\text{'}}_{c,t}}=\frac{T{m}_{n_{c,t},n{\text{'}}_{c,t}}}{T{t}_{n_{c,t},n{\text{'}}_{c,t}}}$$

In order to counteract this circumstance of inconsistent traffic lights or traffic light clusters, _{all possible traffic light object pair combinations are determined for each traffic light cluster t 1} to t ₄ for each crossing c and a correlation factor for each of the determined traffic light object pair combinations ${\rho }_{n_{c,t},n{\text{'}}_{c,t}}$

certainly. The correlation factors ${\rho }_{n_{c,t},n{\text{'}}_{c,t}}\left({\rho }_{Obj}\right)$

are formed as follows: $${\rho }_{n_{c,t},n{\text{'}}_{c,t}}=\frac{T{m}_{n_{c,t},n{\text{'}}_{c,t}}}{T{t}_{n_{c,t},n{\text{'}}_{c,t}}}$$

und $T{t}_{n_{c,t},n{\text{'}}_{c,t}}$

stehen für die Überlappungsdauer bzw. Übereinstimmungsdauer zwischen dem ersten Ampelobjekt n_c,t und dem zweiten Ampelobjekt n'_c,t des Ampelobjektpaars.Here is n _{c, t} a first traffic light object of the traffic light object pair and n'c _{, t} a second traffic light object of the traffic light object pair. $T{m}_{n_{c,t},n{\text{'}}_{c,t}}$

and $T{t}_{n_{c,t},n{\text{'}}_{c,t}}$

stand for the duration of the overlap or the period of correspondence between the first traffic light object n _{c, t} and the second traffic light object n ' _{c, t of} the pair of traffic light objects.

geclustert. Vorliegend erfolgt das in der Regel mittels eines hierarchischen Clusteralgorithmus. Die resultierenden Cluster werden gezählt. Anschließend wird für jedes Ampelcluster t₁ bis t₄ der Mittelwert der Anzahl der resultierenden Cluster über alle Überfahrten c ermittelt. Wenn dieser ermittelte Mittelwert einen vorbestimmten Grenzwert überschreitet, wird das Ampelcluster als inkonsistent bewertet. Der vorbestimmte Grenzwert kann bspw. 1,2 sein.All traffic light objects assigned to a crossing c are determined taking into account the corresponding correlation factors ${\rho }_{n_{c,t},n{\text{'}}_{c,t}}$

clustered. In the present case, this is usually done using a hierarchical cluster algorithm. The resulting clusters are counted. The mean value of the number of resulting clusters over all crossings c is then determined for each traffic light cluster t ₁ to t _4. If this determined mean value exceeds a predetermined limit value, the traffic light cluster is assessed as inconsistent. The predetermined limit value can be 1.2, for example.

In the present example of the traffic light image 7th the traffic light cluster t _{4 represents} an inconsistent traffic light cluster.

In a step S203, the phase profiles of all traffic light objects n _c , especially those of the traffic light clusters assessed as consistent, checked for plausibility. Here are those traffic light objects n _c discarded for the following steps of the procedure, the phases of which have the following sequences of traffic light phases within their phase: yellow phase to red phase, red-yellow phase to green phase, green phase or red-yellow phase to a yellow Phase, red phase or a yellow phase to a red-yellow phase. Implausible traffic light objects n _c are discarded

In a step S204, all traffic light pair combinations are formed from the consistent traffic light clusters t ₁ , t ₂ , t _3. In the present case, the traffic light pair combinations t ₁ / t ₂ , t ₁ / t ₃ , t ₂ / t _{3 are} formed.

Subsequently, in step S205, an overlap duration Tt and a coincidence duration Tm between all traffic light objects are determined across the crossing n _{c, t} assigned to the first traffic light cluster of the traffic light pair and all traffic light objects n _{c, t} , which are assigned to the second traffic light cluster of the traffic light pair, summed up and from this determine a correlation factor ρ _AP between the first and the second traffic light of the traffic light pair. In reference to 7th the terms “overlap duration” and “match duration” are described in more detail later.

für das Ampelpaar aus Ampelcluster t₁ und Ampelcluster t₂ gilt folgende Formel: $$T{t}_{t_1,t_2}={\displaystyle \sum _{c=1}^C{\displaystyle \sum _{n_{c,t_1}=1}^{N_{c,t_1}}{\displaystyle \sum _{n_{c,t_2}=1}^{N_{c,t_2}}T{t}_{n_{c,t_1},n_{c,t_2}}}}}$$

Step S205 is described as an example for the pair of traffic lights t ₁ / t ₂ . For the calculation of the overlap duration across the crossing $\mathrm{<figure-callout\; id="1"\; label="T\; t\; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">T</figure-callout>}{t}_{t_{}}{{}_1,t_2}_{}\left({\text{Tt}}_{\text{AP}}\right)$

The following formula applies to the pair of traffic lights made up of traffic light cluster t ₁ and traffic light cluster t _2: $$\mathrm{<figure-callout\; id="1"\; label="T\; t\; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">T</figure-callout>}{t}_{t_{}}{{}_1,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}_{}={\displaystyle \sum _{c=1}^{\mathrm{C.}}{\displaystyle \sum _{n_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}=1}^{N_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}}{\displaystyle \sum _{n_{c,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}=1}^{N_{c,t_2}}T{t}_{n_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1},n_{c,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}}}}}$$

(Tt_n) steht für die überfahrtspezifische Überlappungsdauer zwischen einem dem Ampelcluster t₁ zugehörigen Ampelobjekt n_{c,t 1} und einem dem Ampelcluster t₂ zugehörigen Ampelobjekt n_{c,t 2} .Here, c stands for the respective crossing from c = 1 ... C and n _{c, t 1} and n _{c, t 2} for the respective traffic light object of the traffic light cluster t ₁ or traffic light cluster t ₂ , where N _{c, t 1} and N _{c, t 2} the number of traffic light objects n _{c, t} represents that were created / recorded from a crossing c for the corresponding traffic light tp ₁ or traffic light tp _2. $T{t}_{n_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1},n_{c,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}}$

(Tt _n ) stands for the crossing-specific overlap duration between a traffic light object n _{c, t belonging to the traffic light cluster t 1 1} and a traffic light object n _{c, t} belonging to the traffic light cluster t _{2 2} .

For example, is from the exemplary traffic light image 7 ' out 4th It can be determined that for the crossing c = 1 there is only one (circular) _{traffic light object n 1,1 for the first traffic light tp 1} and two (circular) _{traffic light objects n 1,2} for the second traffic light tp ₂ . For the second crossing c = 2, there are two (triangular) _{traffic light objects n 2.1 for the first traffic light tp 1} , while only one (triangular) _{traffic light object n 2.2} is present for the second traffic light tp _2. For the crossing c = 3, a (star-shaped) traffic light object n _3.1 , n _3.2 is present for both traffic lights t ₁ , t ₂ .

für das Ampelpaar aus Ampelcluster t₁ und Ampelcluster t₂ gilt folgende Formel: $$T{m}_{t_1,t_2}={\displaystyle \sum _{c=1}^C{\displaystyle \sum _{n_{c,t_1}=1}^{N_{c,t_1}}{\displaystyle \sum _{n_{c,t_2}=1}^{N_{c,t_2}}T{m}_{n_{c,t_1},n_{c,t_2}}}}}$$

$T{m}_{n_{c,t_1},n_{c,t_2}}\left({\text{Tm}}_{\text{n}}\right)$

steht für die überfahrtspezifische Übereinstimmungsdauer zwischen einem dem Ampelcluster t₁ zugehörigen Ampelobjekt n_{c,t 1} und einem dem Ampelcluster t₂ zugehörigen Ampelobjekt n_{c,t 2} .For calculating the duration of the match $\mathrm{<figure-callout\; id="1"\; label="T\; m\; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">T</figure-callout>}{m}_{t_{}}{{}_1,t_2}_{}\left({\text{Tm}}_{\text{AP}}\right)$

The following formula applies to the pair of traffic lights made up of traffic light cluster t ₁ and traffic light cluster t _2: $$\mathrm{<figure-callout\; id="1"\; label="T\; m\; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">T</figure-callout>}{m}_{t_{}}{{}_1,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}_{}={\displaystyle \sum _{c=1}^{\mathrm{C.}}{\displaystyle \sum _{n_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}=1}^{N_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}}{\displaystyle \sum _{n_{c,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}=1}^{N_{c,t_2}}T{m}_{n_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1},n_{c,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}}}}}$$

$T{m}_{n_{c,{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">t</figure-callout>}}_1},n_{c,t_2}}\left({\text{Tm}}_{\text{n}}\right)$

stands for the crossing-specific period of correspondence between a traffic light object n _{c, t belonging to the traffic light cluster t 1 1} and a traffic light object n _{c, t} belonging to the traffic light cluster t _{2 2} .

(allgemein ρ_AP für das Ampelpaar t₁/t₂ gemäß folgender ermittelt werden: $${\rho }_{t_1,t_2}=\frac{T{m}_{t_1,t_2}}{T{t}_{t_1,t_2}}$$

A correlation factor can then then be used ${\mathrm{<figure-callout\; id="1"\; label="\rho \; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">\rho </figure-callout>}}_{t_{}}$

(In general, ρ _AP for the pair of traffic lights t ₁ / t ₂ can be determined according to the following: $${\mathrm{<figure-callout\; id="1"\; label="\rho \; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">\rho </figure-callout>}}_{t_{}}=\frac{\mathrm{<figure-callout\; id="1"\; label="T\; m\; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">T</figure-callout>}{m}_{t_{}}{{}_1,t_2}_{}}{\mathrm{<figure-callout\; id="1"\; label="T\; t\; t"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">T</figure-callout>}{t}_{t_{}}{{}_1,{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}_{}}$$

In step S206, all consistent traffic light clusters are grouped based on the correlation factors of all traffic light pair combinations. The traffic lights are grouped in such a way that all traffic lights within a group have the same phase progression. The grouping takes place using appropriate clustering methods, such as the hierarchical clustering algorithm. The resulting groups, so-called traffic light groups, are referenced with an index g = 1 ... G. Each consistent traffic light cluster is assigned to one of these traffic light groups.

und eine Übereinstimmungsdauer $T{m}_{n_{c,t*},g}$

für ein Ampelobjekt n_c,t* in einem inkonsistenten Ampelcluster t* hinsichtlich aller Ampelobjekte n_{c,t g} in einer Ampelgruppe g ermittelt.In step S207, the traffic light clusters t * recognized as inconsistent are assigned to a traffic light group g. There is an overlap duration for each passage c $T{t}_{n_{c,t*},G}$

and a match duration $T{m}_{n_{c,t*},G}$

for a traffic light object n _{c, t *} in an inconsistent traffic light cluster t * with regard to all traffic light objects n _{c, t G} determined in a traffic light group g.

zwischen dem Ampelobjekt n_c,t* und allen Ampelobjekten n_c,tg aus der Ampelgruppe g wie folgt ermittelt: $$T{t}_{n_{c,t*},g}={\displaystyle \sum _{t_g=1}^{T_g}{\displaystyle \sum _{n_{c,t_g}=1}^{N_{c,t_g}}T{t}_{n_{c,t*},n_{c,t_g}}}}$$

The overlap duration specific to the crossing is used $T{t}_{n_{c,t*},G}$

between the traffic light object n _{c, t *} and all traffic light objects n _{c, tg} from the traffic light group g determined as follows: $$T{t}_{n_{c,t*},G}={\displaystyle \sum _{t_G=1}^{T_G}{\displaystyle \sum _{n_{c,t_G}=1}^{N_{c,t_G}}T{t}_{n_{c,t*},n_{c,t_G}}}}$$

steht für die überfahrtspezifische Überlappungsdauer zwischen einem der inkonsistenten Ampel t* zugehörigen Ampelobjekt n_c,t* und einem dem Ampelcluster t_g zugehörigen Ampelobjekt n_{c,t g} .Here, T _g stands for the number of traffic light clusters in the respective traffic light group g. Furthermore, n _{c, t stands G} for the respective traffic light object _{assigned to the traffic light cluster t g} , where N _{c, t G} represents the number of traffic light objects within a crossing c for the corresponding traffic light cluster t _g . $T{t}_{n_{c,t*},n_{c,t_G}}$

represents the crossing specific overlap duration between one of the inconsistent traffic signal t * corresponding traffic light object n _{c t *} and a traffic light, the traffic light associated cluster t _g object n _{c, t G} .

pro Überfahrt c wird wie folgt ermittelt: $$T{m}_{n_{c,t*},g}={\displaystyle \sum _{t_g=1}^{T_g}{\displaystyle \sum _{n_{c,t_g}=1}^{N_{c,t_g}}T{m}_{n_{c,t*},n_{c,t_g}}}}$$

$T{m}_{n_{c,t*},n_{c,t_g}}$

steht für die überfahrtspezifische Übereinstimmungsdauer zwischen einem dem inkonsistenten Ampelcluster t* zugehörigen Ampelobjekt n_c,t* und allen der Ampelgruppe g zugehörigen Ampelobjekten n_{c,t g} .The crossing-specific overlap duration $T{m}_{n_{c,t*},G}$

per pass c is determined as follows: $$T{m}_{n_{c,t*},G}={\displaystyle \sum _{t_G=1}^{T_G}{\displaystyle \sum _{n_{c,t_G}=1}^{N_{c,t_G}}T{m}_{n_{c,t*},n_{c,t_G}}}}$$

$T{m}_{n_{c,t*},n_{c,t_G}}$

stands for the crossing-specific period of agreement between a traffic light object n _{c, t * belonging to the inconsistent traffic light cluster t * and all traffic light objects n c, t} belonging to the traffic light group g _G .

Then for each inconsistent traffic light t * and for each traffic light group g formed, a cross-crossing _{record Γ g, t *} with all _{traffic light objects n c, t * of} the inconsistent traffic light t * is created according to the following formula: $${\Gamma }_{G,t*}={\displaystyle \underset{c=1}{\overset{\mathrm{C.}}{\cup }}\left\{n_{c,t*}|T{t}_{n_{c,t*},G}>0\wedge \frac{T{m}_{n_{c,t*},G}}{T{t}_{n_{c,t*},G}}>0.99\right\}}$$

von 0.99 aufweisen. Der Wert des Mindestkorrelationsfaktors ist hier beispielhaft und kann angepasst werden.It can therefore be seen that the _{traffic light objects belonging to the set Γ g, t * of} the inconsistent traffic light t * have a minimum correlation factor for the traffic light objects n _{c, g of the traffic light group g} $\frac{T{m}_{n_{c,t*},G}}{T{t}_{n_{c,t*},G}}$

of 0.99. The value of the minimum correlation factor is an example here and can be adjusted.

$$T{m}_{t*,g}={\displaystyle \sum _{c=1}^C{\displaystyle \sum _{n_{c,t*}=1}^{N_{c,t*}}T{m}_{n_{c,t*},g}}}$$

$$\text{mit }{n}_{c,t*}\in {\Gamma }_{g,t*}$$

For the traffic light objects from the set Γ _{g, t * , an overlap period Tt t *, g} and a correspondence period Tm _{t *, g are determined} according to the following formulas, across crossing and traffic light objects. $$T{t}_{t*,G}={\displaystyle \sum _{c=1}^{\mathrm{C.}}{\displaystyle \sum _{n_{c,t*}=1}^{N_{c,t*}}T{t}_{n_{c,t*},G}}}$$

$$T{m}_{t*,G}={\displaystyle \sum _{c=1}^{\mathrm{C.}}{\displaystyle \sum _{n_{c,t*}=1}^{N_{c,t*}}T{m}_{n_{c,t*},G}}}$$

$$\text{With}{n}_{c,t*}\in {\Gamma }_{G,t*}$$

In the above two formulas, n _{c, t *} ∈ Γ _{g, t *} also applies. Thus, only _{traffic light objects n c, t * of} the inconsistent traffic light t * which have or exceed the above-mentioned minimum correlation factor for traffic light group g are added up.

ermittelt, so dass die Anzahl der Ampelobjekte Γ_g,t* der inkonsistenten Ampel t* am höchsten ist, wobei die Ampelobjekte n_c,t* ∈ Γ_g,t* mit der Phase der entsprechenden Gruppe $g_{t*}^{\ast }$

einen hohen Korrelationsfaktor $\left(\frac{T{m}_{n_{c,t*},g}}{T{t}_{n_{c,t*},g}}>\mathrm{0,99}\right)$

aufweisen.For each inconsistent traffic light t * there is a corresponding traffic light group $G_{t*}^{\ast }$

determined so that the number of _{traffic light objects Γ g, t * of} the inconsistent traffic light t * is highest, with the traffic light objects n _{c, t *} ∈ Γ _{g, t *} with the phase of the corresponding group $G_{t*}^{\ast }$

a high correlation factor $\left(\frac{T{m}_{n_{c,t*},G}}{T{t}_{n_{c,t*},G}}>0.99\right)$

exhibit.

The following formula is used for this: $$G_{t*}^{\ast }=\underset{G}{{\displaystyle \text{argmax}}}\left\{\left|{\Gamma }_{G,t*}\right|\right\}$$

zugeordnet, wenn die Anzahl der korrelierten Ampelobjekte $\left|{\Gamma }_{g_{t*}^{\ast },t*}\right|$

aus den Ampelobjekten n_c,t* der inkonsistenten Ampel t* und Ampelobjekten $n_{c,g_{t*}^{\ast }}$

der entsprechenden Ampelgruppe $g_{t*}^{\ast }$

gleich der oder größer als ein vorbestimmter Bruchteil aller (korrelierten und nicht-korrelierten) Ampelobjekte n_c,t* der inkonsistenten Ampel t* ist. Dabei kann der vorbestimmte Bruchteil bspw. ein Drittel sein.The inconsistent traffic light t * becomes this traffic light group $G_{t*}^{\ast }$

assigned when the number of correlated traffic light objects $\left|{\Gamma }_{G_{t*}^{\ast },t*}\right|$

from the _{traffic light objects n c, t * of} the inconsistent traffic light t * and traffic light objects $n_{c,G_{t*}^{\ast }}$

the corresponding traffic light group $G_{t*}^{\ast }$

equal to or greater than a predetermined fraction of all (correlated and non-correlated) _{traffic light objects n c, t * of} the inconsistent traffic light t *. The predetermined fraction can be a third, for example.

zugeordnet. Die übrigen (nicht-korrelierten) Ampelobjekte n_c,t* ∉ Γ_g,t* entsprechen mit hoher Wahrscheinlichkeit (Phasen-)Sichtungen der Abbiegerampel t₅, die nicht weiter interpretierbar sind.Only the correlated traffic light objects n _{c, t *} ∈ Γ _{g, t * of} the inconsistent traffic light t * of the traffic light group are used $G_{t*}^{\ast }$

assigned. The remaining (non-correlated) traffic light objects n _{c, t * g, t *} correspond with a high probability to (phase) sightings of the turning traffic light _{t 5} , which cannot be further interpreted.

In 6th is a method for determining a traffic light phase of a traffic light or a traffic light group of a traffic light system. The following is the procedure for a traffic light from the traffic lights tp ₁ -tp ₅ or a traffic light group from traffic light groups g is carried out.

For example, the information represented by the traffic light objects can be used for the method. In alternatives, the information required for the method can also be provided / represented in a different form than traffic light objects.

In particular, the method is carried out when the vehicle 1 just before or at the stop line of the traffic lights A. (corresponds to the stop line at an intersection). The stop line can be known, for example from map information. Alternatively, the stop line can be used as a median value over all positions along the course of the road section (or along the trajectory of the vehicle 1 ) can be determined at which the vehicle 1 drives past the traffic light objects.

In any case, the vehicle is there 1 that is, in a position in which the vehicle-side detection device 3 the traffic lights A. and can no longer see the traffic lights. This is primarily due to a limited opening angle of a camera device of the detection device 3 . However, it is precisely the point in time when driving over or stopping at the stop line that is relevant for a traffic light lane allocation.

In a step S301, a point in time T _event at which the vehicle 1 either crosses the stop line or stands at it. This can be seen from the vehicle's detection system 3 derive and especially in combination with information from the traffic light objects n _{c, t} .

When the vehicle 1 Passing this stop line at a speed greater than a predetermined speed, for example 3 km / h, is interpreted as crossing the stop line. When the vehicle 1 If the speed drops below 1 km / h for the first time at the stop line, this is interpreted as a stop at the stop line. "At the stop line" includes a predetermined distance from the vehicle 1 (in the direction of travel) to the stop line of up to at least 25m, preferably 15m. Here, the distance from a front end of the vehicle 1 be meant up to the stop line. Other predetermined distances are also conceivable, provided they are suitable for indicating that the vehicle is stopping at the stop line.

In a step S302, a first detection time T _{i is} retrieved at which the vehicle 1 a traffic light is detected for the first time.

A time interval ΔT _i between the point in time T _{i of} the first sighting of the traffic light and the event point in time T _event can be determined.

In a step S303, a last detection time T _{e is} retrieved at which the vehicle detects the traffic light for the last time.

A time interval ΔT _e between the time T _{e of} the last sighting of the traffic light and the event time T _event can be determined. The time interval ΔT _e is set to 0 if the traffic light is still visible _{at the event time T event.}

In a step S304, a phase curve of the traffic light is called up, which extends from the first detection time T _i to the last detection time T _e . A traffic light phase C _{g of} the traffic light which is present at time T _i can be determined from the retrieved phase profile. As a rule, the traffic light phase C _{g can have} at least the following values: green, yellow, red as well as red and yellow. The following traffic light phases C _g can therefore occur: red phase, yellow phase, red phase, green phase and red-yellow phase

In the event that a traffic light phase C _{g of} a traffic light group is called up, traffic lights in this traffic light group may assume / reproduce contradicting traffic light phases instead of a common traffic light phase. Then the traffic light phase C _{g is} set to "invalid" for this traffic light group.

In a step S305, a phase change point in time T _{switch is} determined or retrieved at which a phase change of the traffic light takes place or was detected. In other words, the traffic light phase C _{g of} the traffic light _{changes at the point in time T switch.} The phase change time T _switch is before the last acquisition time T _e .

A time interval ΔT _switch between the time T _{switch of} the phase change of the traffic light and the event time T _event can be determined.

If there is no phase change or if there are contradictions between two or more traffic lights in a traffic light group, the time T _switch and thus also the time interval Δt _{switch are set} to "invalid". If there is no phase change in the retrieved phase profile, the phase change point in time (as described later) can be determined.

In a step S306, a traffic light ^{phase C Δ- is determined} before the phase change time T _{switch of} the traffic light and a traffic light ^{phase C Δ +} after the time T _{switch of} the phase change of the traffic light. The traffic light phases C ^Δ- and C ^{Δ +} thus indicate those traffic light phases that exist before or after the point in time T _switch . The traffic light phases C ^Δ- and C ^{Δ +} can assume the traffic light phases described in step S304.

In a step S307 it is determined which traffic light phase C _{g is present} at the event time T _event . In particular, a probability P (C _g = red) is determined that the traffic light has a traffic light phase of “red” _{at time T event.} In other words, it is determined with what probability the traffic light is red (or indicates a red traffic light phase) when the vehicle 1 stops at the traffic lights or drives past them. The following applies to the probability P (C _g = red): $$\mathrm{P.}\left({\mathrm{C.}}_G=rOt\right)=\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=c,\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\text{C.}}^{\mathrm{\Delta }+}={\text{c}}^{\mathrm{\Delta }+},{\text{C.}}^{\mathrm{\Delta }-}={\text{c}}^{\mathrm{\Delta }-}\right)$$

The probability P (C _g = red) is therefore a conditional probability that takes into account _{the variables ΔT e} , ΔT _i , C _g , ΔT _switch , C ^Δ- and C ^{Δ +.}

The following applies to the probability P (C _g = green) that the traffic light shows a traffic light phase C _g of "green" _{at time T event:} $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=Gr\ddot{u}n\right)=\\ =\mathrm{P.}\left({\mathrm{C.}}_G=Gr\ddot{u}n|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=c,\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\text{C.}}^{\mathrm{\Delta }+}={\text{c}}^{\mathrm{\Delta }+},{\text{C.}}^{\mathrm{\Delta }-}={\text{c}}^{\mathrm{\Delta }-}\right)=\\ =1-\mathrm{P.}\left({\mathrm{C.}}_G=rOt\right)\end{array}$$

The probability P (C _g = green) is also a conditional probability that takes into account _{the variables ΔT i} , ΔT _e , C _g , ΔT _switch , C ^Δ- and C ^{Δ +.}

To calculate the probabilities P (C _g = red) or P (C _g = green) there can be three cases I, II, III. These cases depend on the time intervals ΔT _e and / or ΔT _switch .

In case I, ΔT _e = 0s. The detection device 3 saw the traffic light at time T _event . Correspondingly, the traffic light phase C _{g of} the traffic light is known, so that the following applies to the probability P (C _g = red): $$\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{0,}{\mathrm{C.}}_G=c\right)=\{\begin{array}{ll}\mathrm{1,}\hfill & c=rO\mathrm{<figure-callout\; id="0"\; label="t"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">t</figure-callout>}\hfill \\ \mathrm{0,}\hfill & c=Gr\ddot{u}n\hfill \\ \mathrm{0.8,}\hfill & c=Gelb\hfill \\ \mathrm{0.7,}\hfill & c=Gelb+rOt\hfill \end{array}$$

The probability P (C _g = red) of 0.8 for the traffic light phase c = yellow allows a yellow traffic light to be interpreted as green or red, this value of 0.8 being dependent on the driver. The same applies to the probability P (C _g = red) of 0.7 for c = yellow + red. The probabilities for c = yellow and c = yellow + red are exemplary here. In alternatives, these values can be determined specifically for the driver. In other words, the probabilities for c = yellow and c = yellow + red can be determined / ascertained as a function of a driver's driving behavior.

In case II, the time interval ΔT _{e is} greater than 0s and the time interval ΔT _switch has a valid value (and thus a value greater than 0s). Thus, the traffic light _{phase C g} at time T _{event can be} derived from the last detected phase change at time T _switch .

For this purpose, phase durations of the traffic light phases, which are usually unknown, are taken into account by modeling the phase durations as random variables. A reference traffic light cycle is therefore modeled. In the model, T cycle stands for a circulation time of the reference _{traffic light cycle.} A cycle time corresponds to the time it takes for a traffic light to go through a complete switching process. For example, a complete switching process can be a traffic light phase sequence with the traffic light phases in the order “green”, “yellow”, “red” and “yellow and red”.

The following applies to the cycle time T _cycle: $$T_{cycle}\sim U\left(t_{cycle\_min},t_{cycle\_max}\right)$$

Here, t _{cycle_min is} a minimum cycle time and t _{cycle_max is} the maximum cycle time.

_{For the reference traffic light cycle} , it is assumed that a uniform distribution of the cycle time in the above value range [t cycle_min; t _{cycle_max} ] is present.

For example, t _{cycle_min} can be 30s and t _{cycle_max} 120s. These values come from the guideline for traffic light systems, which is a set of rules valid for Germany. For other countries, other values can be specified accordingly.

For a phase duration X _red in which the reference traffic light cycle shows red, the following applies: $$X_{rOt}\sim U\left(rO{t}_{min},rO{t}_{max}\right)$$

Here, too, it is assumed for the reference traffic light cycle that the phase duration X _{rot is} in the value range [rot _min ; rot _max ] is evenly distributed. As a rule, red _min and red _max indicate the minimum or maximum fraction of the cycle time T _{cycle at which the reference traffic light cycle} shows “red”. For example, red _min = 0.3 and red _max = 0.7. Alternatively, the range of values for the phase duration X _{red can} also be specified by absolute minimum and maximum times in seconds.

In case II the following applies to the probability P (C _g = red): $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\text{C.}}^{\mathrm{\Delta }+}={\text{c}}^{\mathrm{\Delta }+},T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)=\\ =\{\begin{array}{ll}min\left\{\frac{\mathrm{\Delta }{t}_{switcH}^{\text{'}}}{t_{trans}}\mathrm{,1}\right\},\hfill & {\text{c}}^{\mathrm{\Delta }+}=rOt\vee Gelb,\mathrm{\Delta }{t}_{switcH}^{\text{'}}<x_{rOt}\ast t_{cycle}\hfill \\ \mathrm{0,}\hfill & {\text{c}}^{\mathrm{\Delta }+}=rOt\vee Gelb,\mathrm{\Delta }{t}_{switcH}^{\text{'}}\ge x_{rOt}\ast t_{cycle}\hfill \\ \mathrm{0,}\hfill & {\text{c}}^{\mathrm{\Delta }+}=Gr\ddot{u}n\vee \left(rOt+Gelb\right),\mathrm{\Delta }{t}_{switcH}^{\text{'}}<\left(1-x_{rOt}\right)\ast t_{cycle}\hfill \\ min\left\{\frac{\mathrm{\Delta }{t}_{switcH}^{\text{'}}-\left(1-x_{rOt}\right)\ast t_{cycle}}{t_{trans}}\mathrm{,1}\right\},\hfill & {\text{c}}^{\mathrm{\Delta }+}=Gr\ddot{u}n\vee \left(rOt+Gelb\right),\mathrm{\Delta }{t}_{switcH}^{\text{'}}\ge \left(1-x_{rOt}\right)\ast t_{cycle}\hfill \end{array}\end{array}$$

das zykluskorrigierte Zeitintervall, das seit dem letzten Phasenwechsel verlaufen ist. Für $\mathrm{\Delta }{t}_{switch}^{\text{'}}$

gilt: $$\mathrm{\Delta }{t}_{switch}^{\text{'}}=\mathrm{\Delta }{t}_{switch}mod{t}_{cycle}$$

Here is $\mathrm{\Delta }{t}_{switcH}^{\text{'}}$

the cycle-corrected time interval that has elapsed since the last phase change. For $\mathrm{\Delta }{t}_{switcH}^{\text{'}}$

applies: $$\mathrm{\Delta }{t}_{switcH}^{\text{'}}=\mathrm{\Delta }{t}_{switcH}mOd{t}_{cycle}$$

verwendet, um den Fall abzufangen, dass mehr als eine komplette Umlaufzeit der Ampel nach dem Zeitpunkt T_e verstrichen ist. Es ist auch möglich, lediglich das (nicht korrigierte) Zeitintervall Δt_switch in für den Fall II zu verwenden.In particular, the cycle-corrected time interval is $\mathrm{\Delta }{t}_{switcH}^{\text{'}}$

used to intercept the case that more than one complete cycle time of the traffic light has elapsed _{after the point in time T e.} It is also possible to use only the (uncorrected) time interval Δt _switch in for case II.

Furthermore, t _{trans is} a predetermined, in particular fixed, parameter which corresponds to a transition time after the traffic light has changed to “red”. A driver of the vehicle behaves in this transition time t _trans 1 as if the traffic light was still "green". During this transition time t _trans there is a probability other than 0 that the current phase of the traffic light is "interpreted" as green. The transition time t _trans usually has a value of a few seconds. For example, the transition time t _trans can assume a value between 0s and 5s and, in particular, 3s. Furthermore, a further transition period after a phase change to “yellow” can be taken into account. The further transition time is generally longer than the transition time t _trans and the probability that a driver will cross the traffic light in the further transition time is higher.

The distinctions made in Case II are explained below.

mit der roten Phasendauer X_rot des Referenzampelzyklus verglichen. Hier wird die rote Phasendauer X_rot des Referenzampelzyklus als Bruchteil X_rot von der Umlaufzeit t_cycle angegeben.If the traffic light phase C ^{Δ +} present after the phase change is a red or yellow traffic light _{phase, the time interval ΔT switch} (or the cycle-corrected time interval $\mathrm{\Delta }{t}_{switcH}^{\text{'}})$

compared with the red phase duration X _{red of} the reference traffic light cycle. Here the red phase duration X _{rot of the reference traffic light cycle is} specified as a fraction X _rot of the cycle time t cycle.

annimmt. Es wird also angenommen, dass die Ampel nach dem letzten Erfassungszeitpunkt T_e nicht auf die nächste Ampelphase weitergeschalten hat, so dass eine rote Ampelphase zum Event-Zeitpunkt T_event vorliegt.If the time interval ΔT _{switch is} smaller (shorter) than the red phase duration X _{rot of} the reference traffic light cycle, it is assumed that the red traffic light phase is present with a probability P (C _g = red), the probability P (C _g = red) being the smaller value of $\left\{\frac{\mathrm{\Delta }{t}_{switcH}^{\text{'}}}{t_{trans}}\mathrm{,1}\right\}$

accepts. It is therefore assumed that the traffic light has not switched to the next traffic light phase _{after the last detection time T e , so that a red traffic light phase} is present at the event time T event.

However, if the time interval ΔT _{switch is} greater (longer) or equal to the red traffic light phase X _{red of} the reference traffic light cycle, it is assumed that the traffic light has switched to the next (i.e. green) traffic light phase _{between the last detection time T e} and the event time T _event so that there is a green traffic light phase at the event time T _event . The probability P (C _g = red) is then 0 accordingly.

The same applies to the other case distinctions in Case II.

It is noted that the reference traffic light _cycle does not have to take into account the transition time t trans. For example, the transition time t _{trans can be} ignored, so that a probability of 1 always results for the first and fourth case distinctions from case II. The following then applies: $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\text{C.}}^{\mathrm{\Delta }+}={\text{c}}^{\mathrm{\Delta }+},T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)=\\ =\{\begin{array}{ll}\mathrm{1,}\hfill & {\text{c}}^{\mathrm{\Delta }+}=rOt\vee Gelb,\mathrm{\Delta }{t}_{switcH}^{\text{'}}<x_{rOt}\ast t_{cy\mathrm{<figure-callout\; id="0"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">c</figure-callout>}le}\hfill \\ \mathrm{0,}\hfill & {\text{c}}^{\mathrm{\Delta }+}=rOt\vee Gelb,\mathrm{\Delta }{t}_{switcH}^{\text{'}}\ge x_{rOt}\ast t_{cy\mathrm{<figure-callout\; id="0"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">c</figure-callout>}le}\hfill \\ \mathrm{0,}\hfill & {\text{c}}^{\mathrm{\Delta }+}=Gr\ddot{u}n\vee \left(rOt+Gelb\right),\mathrm{\Delta }{t}_{switcH}^{\text{'}}<\left(1-x_{rOt}\right)\ast t_{cy\mathrm{<figure-callout\; id="1"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">c</figure-callout>}le}\hfill \\ \mathrm{1,}\hfill & {\text{c}}^{\mathrm{\Delta }+}=Gr\ddot{u}n\vee \left(rOt+Gelb\right),\mathrm{\Delta }{t}_{switcH}^{\text{'}}\ge \left(1-x_{rOt}\right)\ast t_{cycle}\hfill \end{array}\end{array}$$

The probability for P (C _g = rot) given for case II can be calculated for all possible cycle times T _cycle and all possible phase durations X _rot . Thus, an overall probability P (C _g = rot) can then be determined over all cycle times T _cycle and phase durations X _rot as follows: $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\text{C.}}^{\mathrm{\Delta }+}={\text{c}}^{\mathrm{\Delta }+}\right)=\\ ={\displaystyle {\int }_{t_{cycle\_min}}^{t_{cycle\_max}}{\displaystyle {\int }_{\text{Max}\left\{rO{t}_{min};\frac{\mathrm{\Delta }{t}_{switcH}^{\text{'}}-\mathrm{\Delta }{t}_e}{t_{cycle}}\right\}}^{rO{t}_{max}}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\text{C.}}^{\mathrm{\Delta }+}={\text{c}}^{\mathrm{\Delta }+},T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)}}\\ \ast \mathrm{P.}\left(X_{rOt}=x_{rOt}\right)\ast \mathrm{P.}\left(T_{cycle}=t_{cycle}\right)d{x}_{rOt}d{t}_{cycle}\end{array}$$

The term P (C _g = red | ΔT _switch = Δt _switch , C ^{Δ +} = c ^{Δ +} , T _cycle = t _cycle , X _red = x _red ) can be determined as stated above.

P (X _rot = x _rot ) is the probability that the phase duration X _{rot has} a certain value x _rot from the above-mentioned value range [rot _min ; rot _max ] for the phase duration. P (T _cycle = t _cycle ) is the probability that the cycle time T _{cycle has} a certain value t _cycle from the above-mentioned value range [t _{cycle_min} ; t _{cycle_max} ] for the cycle time. In the case of a uniform distribution of the cycle times and phase durations modeled as random variables, as is the case here, the probability for each value from the value ranges is the same.

In order to calculate the overall probability, the integrals can be approximated, for example, by means of sums. Other approximation methods are also possible.

In case III, neither of the two previous cases I and II apply. This means that in case III only the last seen / recorded traffic light phase of the traffic light is known. Accordingly, it is not known how long this traffic light phase has been around.

The probability P (C _g = red) is therefore determined taking into account a relative duration of the green and red phases and the time elapsed between time T _i and time T _e as a function of the phase last seen. This results in four subcases III.a, III.b, III.c and Ill.d. The above explanations for case II apply accordingly.

For case III.a, in which the last seen phase C _g = red, the following then applies to the probability P (C _g = red): $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt\right)=\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=rOt,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ ={\displaystyle {\int }_{\mathrm{\Delta }{t}_i}^{\mathrm{\Delta }{t}_e+t_{cycle}}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\mathrm{C.}}_G=rOt,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)}\\ \ast \mathrm{P.}\left(\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH}|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=rOt,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)d\mathrm{\Delta }{t}_{switcH}\end{array}$$

The following also applies: $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\text{Δ}{T}_{switcH}=\text{Δ}{t}_{switcH},{\mathrm{C.}}_G=rOt,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ =\{\begin{array}{cc}\mathrm{1,}& \text{Δ}{t}_{switcH}\text{mod}{t}_{cycle}<x_{rOt}\ast t_{cy\mathrm{<figure-callout\; id="0"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">c</figure-callout>}le}\\ \mathrm{0,}& \text{Δ}{t}_{switcH}\text{mod}{t}_{cycle}\ge x_{rOt}\ast t_{cycle}\end{array}\end{array}$$

The phase change time T _switch and, accordingly, the time interval ΔT _switch are unknown and are therefore modeled. For this purpose, it is assumed that the phase change time T _{switch had to take place} between a cycle time T _cycle before the first detection time T _i and the last detection time T _e . Therefore, the following applies to the probability that the phase change time T _{switch occurs} at a specific time t _switch or that the time interval ΔT _{switch has} a specific value _{Δt switch:} $$\begin{array}{l}\mathrm{P.}\left(\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH}|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=rOt,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ =\frac{1}{\mathrm{\Delta }{t}_e+\mathrm{\Delta }{t}_{cycle}-\mathrm{\Delta }{t}_i}\end{array}$$

In case III.b, the last seen phase C _g = green. This corresponds to case III.a, only that the green phase replaces the red phase in case III.a. Hence: $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt\right)=\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=Gr\ddot{u}n,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ ={\displaystyle {\int }_{\mathrm{\Delta }{t}_i}^{\mathrm{\Delta }{t}_e+t_{cycle}}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\mathrm{C.}}_G=Gr\ddot{u}n,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)}\\ \ast \mathrm{P.}\left(\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH}|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=Gr\ddot{u}n,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)d\mathrm{\Delta }{t}_{switcH}\end{array}$$

The following applies to the probability that the red traffic light phase is present at the event time T _event if the green traffic light phase was last seen: $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH},{\mathrm{C.}}_G=Gr\ddot{u}n,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ =\{\begin{array}{cc}\mathrm{1,}& \mathrm{\Delta }{t}_{switcH}\text{mod}{t}_{cycle}<\left(1-x_{rOt}\right)\ast t_{cy\mathrm{<figure-callout\; id="0"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">c</figure-callout>}le}\\ \mathrm{0,}& \mathrm{\Delta }{t}_{switcH}\text{mod}{t}_{cycle}\ge \left(1-x_{rOt}\right)\ast t_{cycle}\end{array}\end{array}$$

For the probability that the phase change point in time T _{switch occurs} at a certain point in time t _switch or that the time interval ΔT _{switch has} a certain _{value Δt switch} , the following applies analogously to case III.a: $$\begin{array}{l}\mathrm{P.}\left(\mathrm{\Delta }{T}_{switcH}=\mathrm{\Delta }{t}_{switcH}|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=Gr\ddot{u}n,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ =\frac{1}{\mathrm{\Delta }{t}_e+\mathrm{\Delta }{t}_{cycle}-\mathrm{\Delta }{t}_i}\end{array}$$

There is the same probability P (ΔT _switch = Δt _switch | ...) for cases III.a and III.b, since the circulation time T _{cycle of} the reference traffic light cycle does not change.

In the case Ill.c, the last seen traffic light phase is yellow. It is assumed here that time T _e corresponds to time T _switch , that is, T _e = T _switch . The same applies to the time intervals Δt _e = Δt _switch . In this case the following applies to the probability P (C _g = red): $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=Gelb,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ =\{\begin{array}{cc}\mathrm{1,}& \mathrm{\Delta }{t}_e\le x_{rOt}\ast t_{cy\mathrm{<figure-callout\; id="0"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0072.png,DE102019211098B4\_0073.png"\; state="\{\{state\}\}">c</figure-callout>}le}\\ \mathrm{0,}& \mathrm{\Delta }{t}_e>x_{rOt}\ast t_{cycle}\end{array}\end{array}$$

In the case of Ill.d, the last seen traffic light phase is “red + yellow”. Here, too, it is assumed that the time T _e corresponds to the time T _switch , that is, T _e = T _switch . Correspondingly, the following also applies to the time intervals Δt _e = Δt _switch . $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=rOt+Gelb,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)\\ =\{\begin{array}{cc}\mathrm{0,}& \mathrm{\Delta }{t}_e\le \left(1-x_{rOt}\right)\ast t_{cy\mathrm{<figure-callout\; id="1"\; label="c\; l\; e"\; filenames="DE102019211098B4\_0070.png,DE102019211098B4\_0072.png"\; state="\{\{state\}\}">c</figure-callout>}le}\\ \mathrm{1,}& \mathrm{\Delta }{t}_e>\left(1-x_{rOt}\right)\ast t_{cycle}\end{array}\end{array}$$

The probabilities P (C _g = rot) for cases III.a to III.d can be calculated for all possible cycle times T _cycle and all possible phase durations X _rot , similar to case II. The following applies to the overall probability: $$\begin{array}{l}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=c\right)=\\ ={\displaystyle {\int }_{t_{cycle\_min}}^{t_{cycle\_max}}{\displaystyle {\int }_{\text{Max}\left\{rO{t}_{min};\frac{\mathrm{\Delta }{t}_i-\mathrm{\Delta }{t}_e}{t_{cycle}}\right\}}^{rO{t}_{max}}\mathrm{P.}\left({\mathrm{C.}}_G=rOt|\mathrm{\Delta }{T}_e=\mathrm{\Delta }{t}_e,\mathrm{\Delta }{T}_i=\mathrm{\Delta }{t}_i,{\mathrm{C.}}_G=c,T_{cycle}=t_{cycle},X_{rOt}=x_{rOt}\right)}}\\ \ast \mathrm{P.}\left(X_{rOt}=x_{rOt}\right)\ast \mathrm{P.}\left(T_{cycle}=t_{cycle}\right)d{x}_{rOt}d{t}_{cycle}\end{array}$$

Here, too, the integrals can be approximated, for example by means of sums, in order to calculate the overall probability. Other approximation methods are also possible.

In 7th are schematic and exemplary two phase courses 9 , 9 ' represented by two traffic light objects. The phases 9 , 9 ' show the green traffic light phase G , the red light phase R. , the yellow traffic light phase Y and the red-yellow traffic light phase R + Y .

The combination of the solid lines results in the coincidence duration Tm of the two phase courses 9 , 9 ' in which they have the same traffic light phase (simultaneously). The combination of the solid lines with the dashed line results in the overlap duration Tt of the two phase courses 9 , 9 ' , in which the phase progressions 9 , 9 ' overlap in time (regardless of their traffic light phases).

The duration of the overlap Tt thus indicates the duration of the two phases 9 , 9 ' are present at the same time, ie overlap in time. The period of correspondence Tm indicates that period of time (within the overlap period Tt) at which the phase progressions 9 , 9 ' have the same traffic light phase.

Furthermore, in 7th corresponding first acquisition _{times T i} , last acquisition _{times T e} and phase change times T _{switch of} the phase profiles 9 , 9 ' shown.

List of reference symbols

A.

Traffic lights ( A. )

F1-F3

Lanes

Fb

Lane section

G

green traffic light phase

tp1-tp5

traffic light

nc

Traffic light objects

nc, t

traffic light object assigned to a cluster t

R.

red light phase

R + Y

red-yellow traffic light phase

t1-t5

(Traffic light / traffic light object) cluster

tp1-tp5

physical traffic light

Y

yellow traffic light phase

1

vehicle

3

Detection device

5

Radio interface

7th

Traffic light image

7 '

Traffic light image

9

Phase progression

9 '

Phase progression

S ...

Procedural steps

Claims (15)

Method for determining a traffic light phase of a traffic light of a traffic light system, the traffic light being on a route of a vehicle, with: - (S301) calling up an event time (T _event ) at which the vehicle is at the traffic light; - (S302) Retrieving a first detection time (T _i ) at which the vehicle (1) detects the traffic light for the first time; - (S303) retrieving a last detection time (T _e ) at which the vehicle (1) detects the traffic light for the last time; - (S304) retrieving a phase curve of the traffic light which extends from the first detection time (T _i ) to the last detection time (T _e ); - (S307) determining a traffic light phase of the traffic light at the event time (T _event ) by comparing the retrieved phase profile with a reference traffic light cycle, the reference traffic light cycle having a predetermined cycle time and phase durations (X _red ) of the reference traffic light cycle being modeled as random variables. Procedure according to Claim 1 , wherein the predetermined round trip time is modeled as a random variable. Method according to one of the preceding claims, wherein phase durations (X _red ) of the reference traffic light cycle are dependent on the predetermined cycle time. Method according to one of the preceding claims, wherein the reference traffic light cycle consists of a red phase and a green phase. Method according to one of the preceding claims, wherein the determination of the traffic light phase further comprises: - (S305) determining a phase change time (T _switch ) at which a phase change of the traffic light takes place before the last detection time (T _e ); and - (S307) determining a probability that the particular traffic light _{phase is present at the event time (T event} ) by comparing a time interval (ΔT _switch ) that differs from the phase change time (T _switch ) and at the event time ( T _event ), with the duration (X _red ) of a corresponding traffic light phase of the reference traffic light cycle, the corresponding traffic light phase corresponding to a traffic light phase present due to the phase change. Procedure according to Claim 5 , where the phase change time (T _switch ) lies within the retrieved phase curve. Procedure according to Claim 5 , wherein the phase change time (T _switch ) is before the first detection time and the determination of the traffic light phase further comprises: - Carrying out one of the following steps depending on the traffic light phase of the retrieved phase profile: - (III.c, III.d) Use the last one Detection time (T _e ) as phase change time (T _switch ) for determining the probability that the particular traffic light phase is present _{at the event time (T event);} or - (III.a, III.b) Modeling a phase change point in time (T _switch ) as a random variable, the modeled phase change point in time (T _switch ) lying before the first detection point in time. Procedure according to Claim 7 , wherein the modeled phase change point in time (T _switch ) lies in a predetermined time interval, which extends in particular from a point in time that is one cycle time before the last detection point in time (T _e ) to the first detection point in time (T _i ). Method according to one of the preceding claims, wherein, if there are at least three phase change times in the retrieved phase profile, at least one can be derived from the cycle time (T _cycle ) of the traffic light and phase duration of the traffic light, taking into account the at least three phase change times. Method according to one of the preceding claims, wherein the traffic light _{phase at the event time (T event} ) corresponds to a green phase of the traffic light or a red phase of the traffic light. Method according to one of the preceding claims, wherein the reference _{traffic light cycle takes into account a transition time (t trans} ) which follows a phase change to the red phase and in which the red phase is regarded as a green phase. Method according to one of the preceding claims, wherein the event time (T _event ), the first acquisition time (T _i ), the last acquisition time (T _e ), the retrieved phase profile and the reference traffic light cycle are stored in a database. Method according to one of the preceding claims, wherein the event time (T _event ), the first detection time (T _i ), the last detection time (T _e ) and the retrieved phase profile are detected by a vehicle-side detection device (3). Device for determining a traffic light phase of a traffic light of a traffic light system, the traffic light lying on a route of a vehicle, the device being designed and set up, a method according to one of the Claims 1 to 13th to execute. Computer program for determining a traffic light phase of a traffic light of a traffic light system, the traffic light lying on a route of a vehicle, the computer program being designed to cause a data processing device when it is executed, a method according to one of the Claims 1 to 13th to execute.

Images