DE102020216496A1 - Method for operating an at least partially automated vehicle - Google Patents

Abstract

Method for operating an at least partially automated vehicle, comprising the steps:- detecting an environment of the vehicle by means of sensors (10);- determining ambiguous localization data (L); and- Enabling a trajectory planning as a function of a check of the ambiguous localization data (L) using data from a digital map (50), the trajectory planning being carried out using the data from the digital map (50).

Description

The invention relates to a method for operating an at least partially automated vehicle. The invention also relates to a device for operating an at least partially automated vehicle. The invention also relates to a computer program product.

State of the art

A robust localization of the own vehicle (ego vehicle) in a high-precision digital HAD map (HAD: "highly automated driving") is considered indispensable for automatic driving (SAE levels 3 to 5). The estimated vehicle pose can be used to derive information on the road topology and the course of the lane relative to the ego vehicle from the HAD map, for example. This information can be used to implement automated driving functions. At low automation levels (SAE levels 2, 3), this can be lateral control within the lane to a preferred lane trajectory or an assisted lane change, as well as adjusting the driving speed to recommended speeds from the HAD map. With highly automated driving, trajectory planning can be based entirely on the HAD map.

A HAD map usually contains different layers. This includes one or more localization levels with environmental features and a planning level for trajectory planning. In today's AD vehicles, various environmental sensors (video, radar, lidar) are used to record environmental features while driving and to compare them with features from the localization level of the map. From this, the current vehicle pose (position and orientation) can be estimated. For example, 3D point clouds or semantic landmarks (road markings, traffic signs, lamp posts, etc.) can be used as localization features.

The accuracy requirement for the localization is typically very high (up to 10 cm max. localization error relative to the HAD map). In many practical cases, however, an unambiguous localization with this level of accuracy is hardly possible. Therefore, very high demands are placed on the sensors, especially for SAE levels 4 and 5, e.g. expensive high-resolution lidar scanners are used and satellite navigation systems with differential correction data from ground-based reference stations (e.g. D-GPS) are required.

US 10,209,718 B2 discloses a vehicle sensor system and a method of using it, proposing automated driving using visual analysis with the help of cameras.

An object of the invention is to provide an improved method for operating an at least partially automated vehicle.

• - Erfassen eines Umfelds des Fahrzeugs mittels Sensorik;
• - Ermitteln von mehrdeutigen Lokalisierungsdaten;
• - Durchführen einer Trajektorienplanung in Abhängigkeit von einer Prüfung der mehrdeutigen Lokalisierungsdaten mittels Daten aus einer digitalen Karte, wobei die Trajektorienplanung unter Verwendung der Daten der digitalen Karte durchgeführt wird; und
• - Freigeben und Durchführen einer automatisierten Fahrfunktion des automatisierten Fahrzeugs unter Verwendung der geplanten Trajektorie.

According to a first aspect, the object is achieved with a method for operating an at least partially automated vehicle, having the steps:

• - Detecting an environment of the vehicle by means of sensors;
• - detecting ambiguous localization data;
• - carrying out a trajectory planning depending on a check of the ambiguous localization data using data from a digital map, the trajectory planning being carried out using the data from the digital map; and
• - Enabling and performing an automated driving function of the automated vehicle using the planned trajectory.

In this way, a method is provided which, despite an ambiguous localization solution, is able to carry out trajectory planning as a function of a check of the ambiguous localization information using the digital map for the vehicle. If trajectory planning is not possible, a handover to the driver can be carried out, for example, so that he can carry out manual control of the vehicle.

The proposed method makes use of the fact that it is known from the map information on the digital map where the vehicle is currently located. It can thus be decided that, despite an ambiguous localization, a defined route can be continued automatically to the next intersection, for example. In this way, the proposed method offers added value compared to known methods, such as a known lane departure warning system.

• - eine Schnittstelle zum Einlesen von Daten einer Umfeldsensorik in ein Lokalisierungsmodul;
• - ein funktional mit dem Lokalisierungsmodul verbundener Trajektorienplaner; wobei
• - mittels des Lokalisierungsmoduls mehrdeutige Lokalisierungsdaten ermittelbar sind; und wobei
• - mittels des Trajektorienplaners eine Trajektorienplanung in Abhängigkeit von einer Prüfung der mehrdeutigen Lokalisierungsdaten mittels Daten aus einer digitalen Karte durchführbar ist, wobei die Trajektorienplanung unter Verwendung der Daten der digitalen Karte durchführbar ist, und wobei eine Freigabe und eine Durchführung einer automatisierten Fahrfunktion des automatisierten Fahrzeugs unter Verwendung der geplanten Trajektorie durchführbar ist.

According to a second aspect, the object is achieved with a device for operating an at least partially automated vehicle, having:

• - An interface for reading data from an environment sensor system into a localization module;
• - a trajectory planner functionally connected to the localization module; whereby
• - Ambiguous localization data can be determined by means of the localization module; and where
• - using the trajectory planner, trajectory planning can be carried out depending on a check of the ambiguous localization data using data from a digital map, the trajectory planning being able to be carried out using the data from the digital map, and a release and execution of an automated driving function of the automated vehicle being possible Use of the planned trajectory is feasible.

According to a third aspect, the object is achieved with a computer program comprising instructions which, when the computer program is executed by a computer, cause the latter to execute the proposed method for operating an at least partially automated vehicle.

Preferred developments of the method are the subject of dependent claims.

An advantageous development of the method provides for the ambiguous localization data to be discrete or continuous. In this way, the method can advantageously process different types of ambiguity in the localization information. An ambiguity in the localization data can result, for example, from driving between continuous lane markings or interrupted lane markings.

A further advantageous development of the method provides that the ambiguous localization data include a lateral position of the vehicle in relation to at least one marking of a lane. A specific type of ambiguity in the localization data can advantageously be processed in this way.

A further advantageous development of the method provides that at least one of the following types of environmental sensor data is provided by the sensor system: camera data, radar data, lidar data. The proposed method can thus advantageously process different environmental sensor data.

A further advantageous development of the method provides that the ambiguous localization data includes at least one of the following: particle filter data, multi-hypothesis tracking data. This advantageously supports the fact that the proposed method can process different types of ambiguities and is not limited to a single type of an ambiguous localization solution.

A further advantageous development of the method provides that data relating to a future lane course are used to enable the trajectory planning. For example, it can be output that there is still a defined route to the next junction. As a result, the information from the digital map is used specifically for the application of the method.

A further advantageous development of the method provides that the vehicle continues to be operated automatically in the event that a number of similar target trajectories are determined from the ambiguous localization data. As a result, automated operation of the vehicle is still possible because it was determined that in the future there will be an essentially homogeneous lane course and thus an ambiguous or ambiguous localization solution for automated driving operation of the vehicle can be accepted.

A further advantageous development of the method provides that the driver of the vehicle is informed if the trajectory planning is not enabled. In this case, a message, for example a visual and/or acoustic and/or haptic message, etc., can be output to the driver of the vehicle that manual driver takeover is now required. Advantageously, valuable time can be gained so that the driver is prepared to take over the driving manually.

The invention is described in detail below with further features and advantages on the basis of several figures. Elements that are the same or have the same function have the same reference numbers.

Disclosed method features result analogously from corresponding disclosed device features and vice versa. This means in particular that features, technical advantages and versions relating to the proposed methods result in an analogous manner from corresponding versions, features and advantages relating to the proposed device and vice versa.

• 1 eine prinzipielle Darstellung einer mehrdeutigen Lokalisierung eines Fahrzeugs während eines Fahrbetriebs;
• 2 eine prinzipielle Darstellung einer weiteren mehrdeutigen Lokalisierung eines Fahrzeugs während eines Fahrbetriebs;
• 3 eine prinzipielle Darstellung des vorgeschlagenen Verfahrens und der vorgeschlagenen Vorrichtung; und
• 4 eine prinzipielle Darstellung eines Ablaufs einer Ausführungsform des vorgeschlagenen Verfahrens.

In the figures shows:

• 1 a basic representation of an ambiguous localization of a vehicle while driving;
• 2 a basic representation of a further ambiguous localization of a vehicle while driving;
• 3 a basic representation of the proposed method and the proposed device; and
• 4 a basic representation of a sequence of an embodiment of the proposed method.

Description of Embodiments

A core concept of the present invention is in particular providing an improved method for operating an at least partially automated vehicle even when ambiguous localization data are present.

In the following, for the sake of simplicity, the term “automatically operated vehicle” is used synonymously in the meanings “semi-automatically operated vehicle”, “autonomously operated vehicle” and “semi-autonomously operated vehicle”. The following predominant use of the term automated vehicle thus does not exclude other of the vehicle types mentioned.

The proposed approach makes it possible to carry out automated driving functions with the aid of a high-resolution digital map 50 even when there is an ambiguous localization solution. The problem is addressed that in many situations a clear localization of the ego vehicle is not possible. A clear localization is often not possible, for example, if the accuracy of satellite navigation systems is low with an obscured view of the sky (e.g. in tunnels/parking garages or with a dense stock of houses/trees) or if localization is to be carried out without correction data for cost reasons. The solvability of the localization problem then depends above all on the availability of the localization features. On the other hand, ambiguous localization is often possible in many situations.

1 shows such a scenario: a roadway with a left lane marking 1a and a right lane marking 1b can be seen. Also recognizable is ambiguous localization information L or ambiguous localization data L, which represents a subset of localization data, by means of which an at least partially automated vehicle (not shown) drives in the y-direction and with only a lateral distance of the vehicle from the two lane markings 1a, 1b is known. In the direction of travel y, the position of the vehicle is therefore not exactly known while driving. As a result, this means a situation with an ambiguous localization solution because the vehicle position can only be estimated laterally on a substantially straight stretch with solid lane markings 1a, 1b. Although additional localization data can be provided by means of GPS data, these are generally not sufficiently precise for an automated driving function.

A longitudinal position of the vehicle in the y-direction depends on other features that are not always present, e.g. on an odometry of the vehicle, the error of which can accumulate over the journey.

An inaccurate localization of the vehicle can also be present with lidar or radar point clouds when driving between noise barriers, with the localization uncertainty behaving as in the aforementioned case.

In the presence of a dashed center line, a periodic situation of the ambiguous localization data L results, with possible vehicle positions being ambiguous, as in 2 implied. In this case, the ambiguous localization is caused due to dashed lane markings 1c. The vehicle position in a digital map cannot be determined unambiguously from the measurement of the single line position relative to the vehicle; instead, several possible localization solutions or data L result, with each localization solution being determined in relation to a lane marking 1c. The result is a kind of discontinuous localization of the vehicle with the periodically sensed (e.g. every 10 m) single line markings 1c.

Such a situation can also occur, for example, if radar reflections from supports of a crash barrier on the roadway are used as localization features.

On multi-lane roads with multiple dashed lines, lateral ambiguity can also result if the ego lane cannot be unambiguously determined.

A well-known approach to resolving the localization ambiguity mentioned is to use additional sensors (e.g. more complex odometry or a combination of environment sensors) or other localization features (e.g. dense lidar point clouds or video textures).

In contrast, the approach developed within the scope of this invention enables the implementation of an automatic driving function without an expanded sensor set, ie only with conventional sensors from the consumer segment. In this way, situations such as those described in 1 and 2 described, can be safely managed by an at least partially automated vehicle.

A system diagram of the proposed approach shows in principle 3 . An environment sensor 10 (eg camera, radar, lidar, etc.) detects environmental features and makes them available to a localization module 20 via an interface (not shown) of the proposed device 100 . The localization module 20 uses a localization level of a high-resolution digital map 50 in order to determine possible vehicle positions from the ambiguous localization data L. The usual ambiguous localization approaches such as particle filters or multi-hypothesis tracking can be used.

All plausible localization data L, not just the most probable localization data L, can advantageously be transferred to a trajectory planner 30 . The ambiguous localization data L can contain further information, such as the covariance matrix of the pose or a measure of a goodness of a feature match, etc.

The trajectory planning carried out by the trajectory planner 30 can calculate and evaluate a target trajectory for all plausible localization solutions L. Based on further information from the digital map 50, a decision is made as to whether the automatic driving function of the vehicle can be reliably activated despite ambiguous or ambiguous vehicle localization, or whether the vehicle must be handed over to the driver. The trajectory planner 30 transfers the data of the determined desired trajectory to a control device 40, which controls an automated driving function of the vehicle.

The proposed device 100 can be embodied, for example, as part of an electronic central control unit, in which central regulation of the vehicle is carried out. Alternatively, it is also conceivable for the proposed device 100 to be in the form of a driver assistance control unit. For low-end systems, for example, provision can also be made for the proposed device 100 to be designed within a sensor of the sensor system 10, for example within a camera of the sensor system 10.

• a) Wenn alle Lokalisierungslösungen (Ego-Fahrzeug-Posen) zu einer ähnlichen Solltrajektorie führen, kann das Fahrzeug weiter automatisiert betrieben werden. Ein einfaches Beispiel dafür ist das spurhaltende Fahren auf gerader Strecke, wobei eine hochgenaue Kenntnis der Fahrzeug-Längsposition nicht unbedingt erforderlich ist. Im Gegensatz zum an sich bekannten Spurhalte-Assistenten verwendet das vorgeschlagene Verfahren eine Lokalisierung und kann dadurch weitere Information aus der digitalen Karte 50 verwenden, z.B. die bevorzugte laterale Fahrtrajektorie inklusive typischer Längsgeschwindigkeiten innerhalb der aktuellen Fahrspur oder vorausliegende Spurzusammenführungen außerhalb einer Sensorreichweite (z.B. verdeckt durch vorausfahrende Fahrzeuge).
• b) Wenn die möglichen Lokalisierungslösungen zu unterschiedlichen Soll-Trajektorien führen, entscheidet der Trajektorienplaner 30, ob weiterhin automatisiert gefahren werden kann.

The ambiguous localization data L can be evaluated using various strategies:

• a) If all localization solutions (ego vehicle poses) lead to a similar target trajectory, the vehicle can continue to be operated automatically. A simple example of this is staying in lane on a straight stretch, where highly accurate knowledge of the vehicle's longitudinal position is not absolutely necessary. In contrast to the lane keeping assistant known per se, the proposed method uses a localization and can therefore use additional information from the digital map 50, e.g. the preferred lateral driving trajectory including typical longitudinal speeds within the current lane or upcoming lane junctions outside of a sensor range (e.g. covered by vehicles ahead Vehicles).
• b) If the possible localization solutions lead to different target trajectories, the trajectory planner 30 decides whether automated driving can continue.

Example: Due to heavy traffic, only a small sensor range can be used. An upcoming map situation (e.g. lane merger, tight curve, or the like) leads to different target trajectories depending on the ego vehicle localization (e.g. lane merger at a distance of 100m or 130m in front of the vehicle).

The trajectory planner 30 evaluates whether automated driving can continue, e.g. keeping to the lane and distance based on the current sensor measurements. Driving behavior can also be adjusted based on the map information, e.g. by reducing the planned speed along the trajectories, although the situation ahead cannot yet be detected by the sensors.

If the target trajectories of the various localization solutions L deviate too much from one another, the information can also be handed over to the driver as a precaution. The map information is very valuable here, since further information beyond the sensor range can be used. For example, it can be provided for this purpose that information (e.g. optically and/or acoustically and/or haptically, etc.) is output to the driver via a man-machine interface, so that he can promptly take over the vehicle manually can prepare.

Digital map 50 can also be used to estimate when a position can be clearly determined again, for example using a clear landmark (eg traffic sign, gantry sign, tunnel exit, etc.) along the route at a distance outside the current sensor range. The required range within which the various target trajectories have to be compared can thus be restricted.

• Wenn nur eine Kamera oder ein Radar-Sensor für die Lokalisierung verwendet wird und die Lokalisierungsmerkmale periodisch sind, kann ein herkömmliches System keine Trajektorienplanung über die Sensorreichweite hinaus durchführen. Dieser Testfall kann insbesondere bei einem Systemstart bei verdecktem Himmel (hohe Gebäude, dichtes Blätterdach, Tunnel) geprüft werden, wenn GPS nur in schlechter Qualität verfügbar ist. Die verwendeten Lokalisierungsmerkmale können z.B. auf einem CAN-Bus des Fahrzeugs oder aus der Karte 50 zur Analyse entnommen werden.

The proposed method advantageously provides an automatic driving function:

• If only one camera or radar sensor is used for localization and the localization features are periodic, a conventional system cannot perform trajectory planning beyond the sensor range. This test case can be checked in particular when the system is started under an obscured sky (tall buildings, dense canopy, tunnel) if the GPS quality is poor. The localization features used can be taken from a CAN bus of the vehicle or from the card 50 for analysis, for example.

4 shows a basic sequence of the proposed method.

In a step 200, the surroundings of the vehicle are detected using sensors 10.

In a step 210, ambiguous localization data L is determined.

In a step 220, trajectory planning is carried out as a function of a check of the ambiguous localization data L using data from a digital map 50, the trajectory planning being carried out using the data from the digital map 50, and an automated driving function of the automated vehicle being enabled and carried out vehicle using the planned trajectory.

The proposed method and the proposed device 100 are at least partially preferably in the form of a computer program which has program code means for executing the method or which is stored on a computer-readable storage medium.

Although the invention has been described above on the basis of specific exemplary embodiments, the person skilled in the art can also implement embodiments that are not disclosed or only partially disclosed without departing from the core of the invention.

QUOTES INCLUDED IN DESCRIPTION

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

Patent Literature Cited

• US10209718B2 [0005]

Claims (10)

Method for operating an at least partially automated vehicle, comprising the steps: - Detecting an environment of the vehicle by means of sensors (10); - determining ambiguous localization data (L); - Carrying out a trajectory planning as a function of a check of the ambiguous localization data (L) using data from a digital map (50), the trajectory planning being carried out using the data from the digital map (50); and - Enabling and performing an automated driving function of the automated vehicle using the planned trajectory. procedure after claim 1 , where the ambiguous location data (L) is discrete or continuous. procedure after claim 1 or 2 , wherein the ambiguous localization data (L) comprises a lateral position of the vehicle in relation to at least one marking (1a... 1n) of a traffic lane. Method according to one of the preceding claims, wherein at least one of the following types of environmental sensor data is provided by the sensor system (10): camera data, radar data, lidar data. Method according to one of the preceding claims, wherein the ambiguous localization data (L) comprises at least one of the following: particle filter data, multi-hypothesis tracking data. Method according to one of the preceding claims, data relating to a future lane course being used to enable the trajectory planning. Method according to one of the preceding claims, wherein if several similar target trajectories are determined from the ambiguous localization data (L), the vehicle continues to be operated automatically. Method according to one of the preceding claims, wherein in the event that the trajectory planning is not enabled, the driver of the vehicle is informed. Apparatus (100) for operating an at least partially automated vehicle, comprising: - An interface for reading data from an environment sensor system (10) into a localization module (20); - a trajectory planner (30) functionally connected to the localization module (20); whereby - Ambiguous localization data (L) can be determined by means of the localization module; and where - using the trajectory planner (30), trajectory planning can be carried out as a function of a check of the ambiguous localization data (L) using data from a digital map (50), the trajectory planning being able to be carried out using the data from the digital map (50), and wherein a release and an execution of an automated driving function of the automated vehicle can be carried out using the planned trajectory. Computer program comprising instructions which, when the computer program is executed by a computer, cause it to carry out a method according to one of Claims 1 until 8th executed or stored on a computer-readable storage medium.

Images