Search

DE-102024132590-A1 - Computer-implemented method for operating at least one vehicle

DE102024132590A1DE 102024132590 A1DE102024132590 A1DE 102024132590A1DE-102024132590-A1

Abstract

The invention relates to a computer-implemented method (100) for operating at least one vehicle (60), wherein the at least one vehicle (60) has at least one image acquisition unit (62) for providing at least one digital image of an environment arranged in front of the at least one vehicle (60), comprising at least the following steps: determining (102) first estimated position data of at least two different first points (24, 26) of a first horizon line (28) in a first digital image (22) from the at least one image acquisition unit (62) by means of at least one trained machine learning algorithm (16); determining (104) second estimated position data of at least two different second points (34, 36) of a second horizon line (38) in a second digital image (32) that was determined after the first digital image (22) from the at least one image acquisition unit (62) by means of at least one trained machine learning algorithm (16); Checking (106) whether a visible horizon (20) is approaching in front of the vehicle (60), at least based on the first horizon line (28) and the second horizon line (38); and providing (108) at least one control signal to adjust the speed of the vehicle (60) when a visible horizon (20) is approaching in front of the vehicle (60). The method (100) has increased accuracy for the detection of a horizon line.

Inventors

  • Jan Berka

Assignees

  • DR. ING. H.C. F. PORSCHE AKTIENGESELLSCHAFT

Dates

Publication Date
20260513
Application Date
20241108

Claims (10)

  1. A computer-implemented method (100) for operating at least one vehicle (60), wherein the at least one vehicle (60) has at least one image acquisition unit (62) for providing at least one digital image of an environment located in front of the at least one vehicle (60), comprising at least the following steps: a. Determining (102) first estimated position data of at least two different first points (24, 26) of a first horizon line (28) in a first digital image (22) from the at least one image acquisition unit (62) using at least one trained machine learning algorithm (16); b. Determining (104) second estimated position data of at least two different second points (34, 36) of a second horizon line (38) in a second digital image (32) acquired after the first digital image (22) from the at least one image acquisition unit (62) using at least one trained machine learning algorithm (16); c. Check (106) whether a visible horizon (20) is approaching in front of the vehicle (60), at least based on the first horizon line (28) and the second horizon line (38); and d. Provide (108) at least one control signal to adjust the speed of the vehicle (60) when a visible horizon (20) is approaching in front of the vehicle (60).
  2. Computer-implemented method (100) according to Claim 1 , characterized in that in step (106) checking whether a visible horizon (20) is approaching in front of the vehicle (60), an approach of the visible horizon (20) is output as a result if the second horizon line (38) is located below the first horizon line (28).
  3. Computer-implemented method (100) according to Claim 1 or 2 , characterized in that in step: Check (106) whether a visible horizon (20) is approaching in front of the vehicle (60), an approach of the visible horizon (20) is output as a result if the second horizon line (38) exceeds a predefined distance to the first horizon line (28).
  4. Computer-implemented method (100) according to one of the preceding claims, characterized in that the step: Checking (106) whether a visible horizon (20) is approaching in front of the vehicle (60), an approach of the visible horizon (20) is output as a result, is additionally based on map data.
  5. Computer-implemented method (100) according to one of the preceding claims, characterized in that the method (100) further comprises at least the following step between the step: determining (102) of first estimated position data, and the step: determining (104) of second estimated position data: a. determining (110) of third estimated position data of at least two different third points (44, 46) of a third horizon line (48) in a third digital image (42) that was determined after the first digital image (22) and before the second digital image (32) by the at least one image acquisition unit (62), using at least one trained machine learning algorithm (16).
  6. Computer-implemented method (100) according to Claim 5 , characterized in that in step (106) checking whether a visible horizon (20) is approaching in front of the vehicle (60), an approach of the visible horizon (20) is output as a result if, in addition, the third horizon line (48) is arranged below the first horizon line (28) and the second horizon line (38) is arranged below the third horizon line (48).
  7. Computer-implemented method (100) according to one of the preceding claims, characterized in that at least one digital image of the at least one image acquisition unit (62) is processed by means of an image preprocessing function before estimated position data for the at least one image are determined.
  8. Computer-implemented method (100) according to one of the preceding claims, characterized in that the at least one trained learning algorithm (16) comprises at least one trained neural network.
  9. Computer program product, comprising instructions which, when executed by a computer, cause the computer to perform the steps of procedure (100) according to one of the Claims 1 until 8 to execute.
  10. Vehicle (60) comprising at least one image acquisition unit (62) for capturing an environment in front of the vehicle (60) and at least one control unit (70) for receiving image data from the at least one image acquisition unit (62) and for performing the steps of the procedure (100) according to one of the Claims 1 until 8 is trained.

Description

The invention relates to a computer-implemented method for operating at least one vehicle. Assistance and/or automated driving functions can include, for example, adaptive cruise control. Adaptive cruise control can automatically brake the vehicle in the event of obstacles on the road, such as vehicles ahead, speed limit signs, and/or right-of-way situations. Out of WO 2022/083208 A1 Furthermore, a method for operating a vehicle is known in which a horizon line is estimated and the speed is adjusted accordingly, for example, to brake before hilltops. The horizon line is adjusted based on digital map data. The object of the invention is to provide a computer-implemented method that has increased accuracy for the detection of a horizon line. The problem is solved by the features of the independent claims. Advantageous further developments are the subject of the dependent claims and the following description. According to a first aspect, a computer-implemented method for operating at least one vehicle is described, wherein the at least one vehicle has at least one image acquisition unit for providing at least one digital image of an environment located in front of the at least one vehicle, comprising at least the following steps: determining first estimated position data of at least two different first points of a first horizon line in a first digital image from the at least one image acquisition unit using at least one trained machine learning algorithm; determining second estimated position data of at least two different second points of a second horizon line in a second digital image obtained after the first digital image from the at least one image acquisition unit using at least one trained machine learning algorithm; checking whether a visible horizon is approaching in front of the vehicle, at least based on the first horizon line and the second horizon line; and providing at least one control signal for adjusting the speed of the vehicle when a visible horizon is approaching in front of the vehicle. The computer-implemented method estimates the position of the horizon line in a digital image using a trained, machine learning algorithm. This algorithm can provide estimated position data for at least two points spaced apart along the horizon line. The horizon line is assumed to be a straight line. An image acquisition unit in the vehicle can provide an initial digital image for analysis by the computer-implemented method. Using the trained algorithm, the initial position data for at least two different points along the horizon line in this first digital image is estimated. This process is repeated for a second digital image, acquired by the image acquisition unit after the first. Based on these two horizon lines, the system then checks whether the vehicle is approaching a visible horizon ahead. Approaching the visible horizon occurs, for example, when a vehicle approaches a hilltop. This is a situation where it is advisable to slow the vehicle, as visibility beyond the visible horizon is not possible. Obstacles that may be located immediately behind or below the visible horizon can therefore be outside the field of vision of cameras or the driver. If the assessment indicates that the vehicle is approaching the visible horizon, at least one control signal is issued, which can be used to adjust the vehicle's speed. By using the trained machine learning algorithm, the determination of a horizon line in an image can be performed with increased accuracy, unlike when relying solely on map data. Furthermore, the trained machine learning algorithm can determine the position of the horizon line at a higher speed. According to some embodiments, it is conceivable that in the step: Checking whether a visible horizon is approaching in front of the vehicle, an approach of the visible horizon can be output as a result if the second horizon line is arranged below the first horizon line. Since at least one image capture unit typically maintains a nearly constant distance from the roadway, the visible horizon can descend as the vehicle approaches a hilltop. This means that if, in the second digital image (acquired after the first), the second horizon line is positioned below the first horizon line of the first digital image, a [missing information] can be assumed. The downward slope of the visible horizon can be deduced. Then, solely through the analysis of digital images, an approximation of a visible horizon can be determined. The use of map data providing information about the contour lines of the route is therefore unnecessary. According to some embodiments, it is conceivable that in the step: Checking whether a visible horizon is approaching in front of the vehicle, an approach of the visible horizon can be output as a result if the second horizon line exceeds a predefined distance to the first horizon line. The predefined distance to the first horizon line can be understood as a threshold. When the distance between t