Method and control device for fully automated or autonomous driving of a vehicle

Abstract

The invention relates to a method (100) for fully automated or autonomous driving of a vehicle (F), wherein the method (100) comprises the following steps: - Receiving sensor data relating to a vehicle environment of the vehicle (F) (101), - Evaluating a traffic situation based on sensor data (102), - Planning and executing a driving maneuver for the vehicle (F) based on the evaluated traffic situation (103), - Analysis of the executed driving maneuver for a driving error of the own vehicle based on the sensor data recorded during the executed driving maneuver (104), and - if a driving error has been analyzed, the analysis result is stored internally in the vehicle and / or made available for transmission to an external server unit (105).

Description

[0001] The invention relates to a method, in particular a computer-implemented method for fully automated or autonomous driving of a vehicle, a control device, a computer program and a computer program product on which the computer program according to the invention is stored.

[0002] Many modern vehicles are equipped with driving functions that, based on the current driving scenario, allow the vehicle to move at least semi-autonomously, meaning, for example, without steering input from the driver and / or without the driver operating the accelerator or brake pedals. Well-known driving functions include lane keeping assist, emergency braking assist, and adaptive cruise control. Such vehicles have a sensor unit with one or more sensors to detect the vehicle's surroundings. The sensors used include, for example, radar sensors, lidar sensors, ultrasonic sensors, or camera sensors. Based on the sensor data acquired by these sensors, conclusions can be drawn about the vehicle's environment, which can then be used to plan and control the driving process.

[0003] By implementing such driving functions, the autonomy of vehicle control increasingly transfers to the vehicle itself. According to the current SAE (Society of Automotive Engineers) standard, a distinction is made between five levels. System Level 4 refers to fully automated driving, where the driver may be prompted to intervene. SAE Level 5 is autonomous driving, where the vehicle is controlled without the influence of a human driver.

[0004] To execute an automated driving maneuver, a trajectory is typically calculated based on the current vehicle state and information about the vehicle's surroundings, along which the vehicle then moves automatically. Examples of such driving maneuvers include reducing speed when approaching a slower vehicle ahead, overtaking maneuvers, turning, obstacles, and many more.

[0005] Fully automated and autonomous driving requires a very high level of safety. Real and virtual vehicle tests can examine a large number of different traffic scenarios, but these are resource-intensive and time-consuming. Furthermore, it is practically impossible to consider the almost unlimited variety of situations.

[0006] Based on this, the present invention aims to improve safety in automated driving of a vehicle.

[0007] This problem is solved by the method according to claim 1, a control device according to claim 9, the computer program according to claim 10 and the computer program product according to claim 11.

[0008] According to a first aspect of the present invention, the problem underlying the invention is solved by a method, in particular a computer-implemented method, for the fully automated or autonomous operation of a vehicle. Fully automated operation of the vehicle corresponds to automation level 4. Autonomous operation of the vehicle corresponds to automation level 5. Accordingly, in this driving state, the vehicle is driven independently of the driver by the automatic control system. The fully automated or autonomous operation of the vehicle can be activated by a request from the driver. Alternatively, the fully automated or autonomous operation can be activated automatically, i.e., without prior input from the driver. The fully automated or autonomous vehicle can be a passenger car, a commercial vehicle, or an agricultural machine.The procedure includes the following steps:

[0009] Sensor data relating to the vehicle's surroundings is received. A wide variety of sensors, particularly various environmental sensors, can be used to acquire this data. In this respect, various ADAS sensors are especially suitable, for example, one or more cameras, radar sensors, ultrasonic sensors, and / or lidar sensors. The sensor data differs depending on the sensors used—for example, in the case of one or more cameras, the sensor data is image data. A single sensor or several identical or different sensors can be used, whose sensor data can be evaluated independently or in combination. Preferably, the sensors capture the entire surrounding area of ​​the vehicle (360° coverage).

[0010] In the next step, the traffic situation is evaluated using the sensor data. The traffic situation refers specifically to the traffic situation in which the vehicle is currently located. The evaluated traffic situation describes the static environment and the intentions of road users in the vehicle's vicinity. An example of an evaluated traffic situation is driving on a street in a built-up area and noticing that a pedestrian with the right-of-way is crossing a crosswalk ahead. Another example is driving on a street in a built-up area with a traffic jam ahead, identified, for example, by activated brake lights and / or flashing hazard warning lights. The sensor data for evaluating the traffic situation includes, for example, static and dynamic objects as well as infrastructure elements such as...The sensor data includes information on road users, traffic lights, lane markings, traffic signs, and pedestrian crossings. Furthermore, it can encompass features characterizing the traffic situation, such as the hazard warning lights of a vehicle stopped in the lane, indicating temporary parking. Additionally, the sensor data can be used to determine the position, orientation, absolute and / or relative speeds, and / or accelerations of the vehicle and other road users. Thus, the sensor data represents the current and expected traffic scenario.

[0011] Optionally, in addition to sensor data, the traffic situation can be analyzed using map information from a digital map. Examples of map information include road classification, number of lanes, speed limits, intersections, and the presence of traffic lights and pedestrian crossings. This digital map information can be provided, for example, by the vehicle's navigation system.

[0012] In the next step, a driving maneuver for the vehicle is planned based on the traffic situation evaluated using sensor data. The driving maneuver is preferably a sequence of movements tailored to the evaluated traffic situation. At least one trajectory is preferably evaluated when planning the driving maneuver. Trajectory planning comprises several steps: First, a route is planned from a starting point to a desired destination. Such route planning can be carried out using an existing road network, analogous to current navigation systems. Using appropriate behavior planning, the abstract behavior of the vehicle and other road users is also planned, taking into account the vehicle's surroundings, other road users, static obstacles, lane markings, and traffic signs.Various behavior classes can be defined here, such as "staying in lane," "changing lanes," "turning," or "stopping." Trajectory planning adapts the planned abstract behavior to the current situation and environment.

[0013] The planned driving maneuver, i.e., the planned sequence of movements, is then executed. For this purpose, a drive and / or brake unit and / or a steering system of the vehicle is controlled in such a way that the vehicle adheres to the planned driving maneuver. The executed driving maneuver is at least temporarily stored or retained for subsequent analysis.

[0014] In the next step, the executed driving maneuver is analyzed for any driving errors made by the vehicle. This analysis is performed, for example, by a processing unit, specifically an electronic control unit (ECU). The analysis is based on sensor data recorded during the maneuver, optionally supplemented by information from a digital map. A driving error is defined as driving behavior that does not comply with traffic regulations, specifically disregarding the rules of the road. The behavioral analysis of the vehicle during the maneuver is used to determine, in particular, whether the assessed traffic situation was based on a misinterpretation of the sensor data. In other words, it is specifically determined whether an incorrect evaluation of the actual traffic situation occurred, leading to an inappropriately planned and executed driving maneuver.For example, if the vehicle crosses an intersection, the computing unit can analyze the recorded sensor data, captured by a rear-facing sensor unit such as a vehicle camera, after crossing to determine whether the right-of-way rules or traffic light signals were correctly recognized.

[0015] If a driving error is analyzed during a maneuver, the analysis result is stored internally in the vehicle, for example in a database, and / or made available for transmission to an external server. The analysis result includes at least the identified driving error of the vehicle itself. Optionally, it may also include road and / or location information related to the detected driving error and / or the sensor data on which the planned and executed maneuver was based. The analysis result can be stored, for example, on an internal storage unit. Based on the stored analysis result, the vehicle's fully automated or autonomous driving system can continuously monitor its own driving behavior and, if driving errors are detected, correct them, for example, using an AI-based learning process.

[0016] By providing the analysis results, detected driving errors can also be transmitted to the vehicle-external server unit, which can centrally collect the analysis results from various vehicles during continuous operation. Based on the collected analysis results, troubleshooting can be carried out centrally. This centralized collection is suitable for identifying recurring driving errors. The troubleshooting can be made available, for example, via a software update provided by the server unit, which can then be accessed by a control unit of the automated vehicle.

[0017] The stored or provided analysis results regarding driving errors are used to detect and document situation-dependent wrong-way driving scenarios. This documentation allows for measures to be taken to complete and adapt the system's assessment of driving scenarios. In this way, the quality of the fully automated or autonomous driving system can be improved.

[0018] For the method according to the invention, it is advantageous to define a fixed time range for storing the executed, previously performed driving maneuver. By way of example, 10 seconds could be considered a fixed time range. Reducing the data collection to a minimum enables a particularly space-saving and therefore resource-efficient implementation of the driving maneuver analysis.

[0019] Driving errors are often not immediately apparent during behavioral analysis. For example, the detection of a disregarded regulation, such as a right-of-way or speed limit, resulting from a traffic sign that is obscured and therefore not detectable by the vehicle's sensors, cannot be directly determined even after the fact. Against this background, it is preferable that, for the analysis of faulty driving behavior, at least one or more indicators of a failed driving maneuver be identified during the executed maneuver. A driving error is preferably identified if at least one or two indicators of a failed driving maneuver are found. Identifying these indicators is a helpful tool for verifying that driving maneuvers were performed correctly.

[0020] For example, a flashing headlights and / or an audible horn from another road user can be an indication of inappropriate driving behavior. The flashing headlights or horn are strong indicators that the traffic situation was not compliant with traffic regulations, and in particular, that it was a conflict situation. Ideally, the flashing headlights or horn should be recorded as an indication of inappropriate driving behavior if the distance between the driver's vehicle and the honking road user is within a specified maximum distance. This maximum distance is intended to ensure that horn signals are only taken into account if they were directed at the driver's own vehicle.

[0021] Another common indicator of poor driving behavior is failing to maintain a prescribed minimum distance between one's vehicle and another road user. In one possible implementation, the prescribed minimum distance to another road user is correlated with the vehicle's own speed and / or the relative speed of the vehicle and the other road user. Specifically, the minimum distance is reduced the lower the speed of both the vehicle and the other road user. Setting the minimum distance limit based on speed allows for situational assessment of traffic situations. For example, if a vehicle and a cyclist are close together in stop-and-go traffic, a smaller minimum distance is permitted than when driving on a rural road at the permitted speed.

[0022] Another common indicator of poor driving behavior is failing to maintain a minimum required following distance to a crossing vehicle. Such a situation suggests that the crossing vehicle is attempting to avoid a collision, and analyzing the maneuver can reveal potential errors in the driver's own vehicle.

[0023] Determining the described signs enables reliable detection of a traffic situation that has been incorrectly assessed by the system.

[0024] According to a preferred embodiment, when faulty driving behavior is detected, it is added to a digital map stored on a processing unit of the vehicle and taken into account when the same section of road is subsequently traversed during further automatic vehicle control. For example, when traversing the same section of road—that is, the section where the maneuver with the detected faulty driving behavior occurred—an alternative driving maneuver is calculated. In this way, the risk of a recurring driving error can be reduced.

[0025] Another aspect of the invention relates to a control device, in particular a control unit, which is configured to carry out the method according to the preceding description.

[0026] The control device may in particular include a microcontroller or microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) and the like, as well as software for carrying out the corresponding process steps.

[0027] The present invention can therefore be implemented in digital electronic circuits, computer hardware, firmware or software.

[0028] Furthermore, the present invention also includes a computer program with instructions which, when the computer program is executed by a computer, cause the computer to execute the method according to the preceding description.

[0029] Furthermore, the present invention comprises a computer program product on which the computer program according to the invention is stored.

[0030] The invention and some of its advantageous embodiments are explained in more detail with reference to the following figures. All features described and / or illustrated are, individually or in any combination, fundamentally the subject matter of the invention, irrespective of their compilation in the claims or their cross-reference. The content of the claims is also incorporated into the description.

[0031] They show: Fig. 1 a schematic top view of a vehicle with a control device for carrying out the method according to the invention; Fig. 2a, Fig. 2b an exemplary traffic scenario with the vehicle from Fig. 1 during automated guiding using the method according to the invention; Fig. 3 a schematic flowchart for carrying out the method according to the invention.

[0032] In the figures, identical elements are always labelled with the same reference symbol.

[0033] Fig. Figure 1 shows a vehicle F from a bird's-eye view. The vehicle F has a control device 1 which is designed to drive the vehicle F fully automatically or autonomously. The vehicle F is thus equipped with an automated driving system of SAE Level 4 or 5, which takes over driving independently, i.e., without supervision or even the physical presence of the driver.

[0034] The control device 1 comprises a sensor unit 2, which is configured to detect a portion or the entire area (360° detection) of the vehicle F's surroundings. The sensor unit 2 can include any type of sensor, for example, ultrasonic sensors, one or more radar sensors, one or more cameras, and / or one or more lidar sensors.

[0035] Sensor unit 2 is coupled to a computing unit 3, which has at least one processor and one memory unit. Computing unit 3 is designed to evaluate the traffic situation, i.e., the current traffic scenario, based on the sensor data received from sensor unit 2. In the Fig. In the illustrated example 2a, the traffic situation being evaluated is an intersection with a first road user V1, who enters the intersection and has the right of way over the vehicle F. The second road user V2, who wants to enter the intersection from an exit, was also detected by sensor unit 2.

[0036] The processing unit 3 is designed to plan a driving maneuver for vehicle F based on the traffic situation evaluated from the sensor data. In the case of the Fig. In the illustrated example 2a, the driving maneuver is a turning operation at an intersection, taking into account the first road user V1, who must yield the right-of-way. The second road user V2 is not considered in the driving maneuver planning because the area where the second road user V2 is located is not evaluated as an exit and the second road user V2 is evaluated as a parked vehicle. Subsequently, the planned driving maneuver is executed by controlling at least one vehicle actuator.

[0037] Furthermore, the processing unit 3 is designed to analyze the executed driving maneuver for any driving errors by the vehicle F following the maneuver. This analysis is based on sensor data acquired during the maneuver and, optionally, additionally on map data. If the processing unit 3 identifies a driving error, it stores the error as an analysis result within the vehicle, for example, on the processing unit 3's storage device. For instance, the processing unit 3 is designed to take detected driving errors into account in future driving scenarios involving the same or similar scenarios. The ability to independently learn unknown driving scenarios and to detect errors in maneuver planning due to misinterpretations of object detection ensures reliable quality control and, in particular, improvement of the fully automated or autonomous driving system.Alternatively or optionally in addition to in-vehicle storage, the analysis result can be made available for transmission to an external server unit.

[0038] At the in Fig. In the illustrated example 2b, the second road user, V2, enters the intersection, thus losing its right-of-way to vehicle F. To evaluate this driving error, indicators of a failed maneuver are preferably identified. These indicators include, for example, the vehicle F and the second road user, V2, failing to maintain a minimum distance, squealing tires from a vehicle F in the vicinity, or a horn signal activated by the second road user, V2. By evaluating the driving error and the misinterpretation of the surrounding area, the processing unit can, for example, reinterpret this and other similar traffic situations. In this case, it can recognize the area occupied by the second road user, V2, as an exit and take V2 into account when planning the maneuver.

[0039] Fig.Figure 3 shows a schematic flowchart illustrating the processes of the procedure for fully automated or autonomous driving of a vehicle F.

[0040] According to the inventive method, sensor data relating to the vehicle environment of vehicle F are received 101. Based on the sensor data, a traffic situation of vehicle F is evaluated 102. Based on the evaluated traffic situation, a driving maneuver is planned and executed 103. In a next process step, the executed driving maneuver is analyzed for driving errors of the vehicle itself, based on the sensor data 104 acquired during the executed driving maneuver. If a driving error has been analyzed, an analysis result is stored internally in the vehicle and / or the analysis result is made available for transmission to an external server unit 105.

[0041] The invention has been described above using exemplary embodiments. It is understood that numerous modifications and adaptations are possible without thereby departing from the scope of protection defined by the patent claims.

Claims

[1] Method (100), in particular a computer-implemented method, for the fully automated or autonomous driving of a vehicle (F), the method (100) comprising the following steps: - receiving sensor data relating to a vehicle environment of the vehicle (F) (101), - evaluating a traffic situation based on the sensor data (102), - Planning and executing a driving maneuver for the vehicle (F) based on the evaluated traffic situation (103), - analysis of the driving maneuver performed for a driving error of the own vehicle based on the sensor data (104) recorded during the driving maneuver performed, and - if a driving error has been analyzed, storing an analysis result within the vehicle and / or making the analysis result available for transmission to a server unit external to the vehicle (105). [2] Method (100) according to claim 1, characterized bythat the analysis of the driving maneuver carried out is also carried out using map information from a digital map. [3] Method (100) according to claim 1 or 2, characterized by that in order to analyse the driving manoeuvre carried out for a driving error, signs of a failed driving manoeuvre are determined. [4] Method (100) according to claim 2, characterized by that an indication of unsuccessful driving behaviour is an activated headlight flash and / or an acoustic horn of a road user (V1, V2). [5] Method (100) according to claim 2, characterized by that an indication of unsuccessful driving behaviour is an under-run of the specified minimum distance between the own vehicle (F) and a road user (V1, V2). [6] Method (100) according to claim 5, characterized bythat the specified minimum distance to a road user (V1, V2) is set in correlation with the vehicle's own speed (F) and / or the relative speed of the vehicle (F) and road user (V1, V2). [7] Method (100) according to one of the preceding claims 2 to 5, characterized by that an indication of unsuccessful driving behaviour is if the specified minimum rear distance of the own vehicle (F) in the longitudinal direction to a crossing road user (V1, V2) is not maintained. [8] Method (100) according to one of the preceding claims, characterized by that a detected driving error is added to a digital map of the vehicle and taken into account in the further automatic guidance of the vehicle (F) when driving through the same road section again. [9] Control device which is arranged to carry out the method according to one of claims 1 to 8. [10] A computer program comprising instructions which, when the computer program is executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 8. [11] A computer program product on which a computer program according to claim 10 is stored.

Images