How to improve driving safety

The method improves driving safety by validating lane transitions using multiple sensors and data sources, addressing false detections and integrity issues, ensuring reliable lane keeping and autonomous driving.

JP7583600B2Active Publication Date: 2024-11-14CONTINENTAL AUTONOMOUS MOBILITY GERMANY GMBH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
JP2020209121
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-03-09
Filing Date
2020-12-17
Publication Date
2024-11-14
Estimated Expiration
2040-12-17

AI Technical Summary

Technical Problem

Current driving assistance systems face safety issues due to false lane detections by cameras, especially in hands-off driving scenarios, and existing cameras fail to meet increased integrity requirements for higher automation levels, necessitating additional validation of detected lanes.

Method used

A method involving multiple sensors and data sources to validate lane transitions, including peripheral image capture, lane boundary detection, calculating lane transitions, verifying agreement, determining confidence values, and using HD maps or infrastructure for accurate lane change verification, ensuring reliable driving functions.

Benefits of technology

Enhances safety and robustness of driving functions by providing accurate lane transition validation, preventing erroneous steering interventions, and ensuring reliable operation in partially or fully autonomous vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007583600000001
    Figure 0007583600000001
  • Figure 0007583600000002
    Figure 0007583600000002
Patent Text Reader

Abstract

To provide a method for improving safety and robustness of a travel function.SOLUTION: In a method for improving safety of a travel function in a partial automated vehicle or a complete autonomous vehicle, it comprises the following steps. The method comprises: a step S1 of photographing a peripheral image or a series of peripheral images by using at least one periphery detection sensor; a step S2 of detecting lane lines in the peripheral image or the series of peripheral images; a step S3 of calculating lane transition on the basis of the detected lane lines; a step S4 of fetching further lane transition from a data source; a step S5 of confirming validity of the calculated lane transition by verifying coincidence of the lane transition; a step S6 of determining a degree of coincidence; a step S7 of determining a confidence value on the basis of the degree of coincidence; and a step S8 of determining whether or not the calculated lane transition is supplied to the travel function.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a method for improving the safety of driving functions. [Background technology]

[0002] In current driver assistance systems with steering intervention, the trajectory to be traveled according to the corresponding steering signal is determined on the basis of roadway markings. In this case, the markings are detected by sensors in the form of various camera systems. Here, erroneous detections can only be recognized by further camera systems. If only one camera is installed, erroneous detection of the markings leads to deviations from the planned trajectory. This represents a significant safety problem, especially when the driver is not steering himself ("hands off"). Summary of the Invention [Problem to be solved by the invention]

[0003] It is therefore an object of the present invention to provide a method for increasing the safety and robustness of driving functions.

[0004] The first consideration was that the validation of the detected lanes using an additional camera is not justified: common cause failures (common electronic / electrical failures, backlight, snow, rain, etc.) cause the cameras to be affected in the same way in certain situations. Apart from that, a second camera needs to be installed, which is a non-trivial cost factor.

[0005] The development of safety concepts for lane keeping, which may be used "hands off", leads to higher requirements. According to the safety target ASIL defined for the "hands on" function, an additional validation of the detected lane is required. Currently, cameras available on the market do not meet the increased integrity requirements. [Means for solving the problem]

[0006] Therefore, according to the present invention, a method for improving the safety of driving functions in a vehicle having an automaticity level of SAE L2 or higher in SAE J3016, comprising the following steps: - taking a surroundings image or a series of surroundings images using at least one surroundings detection sensor; - detecting lane lines in a surrounding image or a sequence of surrounding images; - calculating lane transitions based on the detected lane boundaries; - ingesting further lane transitions from a data source; and - validating the calculated lane transitions by verifying the agreement of the lane transitions; - determining the degree of match; - determining a confidence value based on the degree of match; - determining whether the calculated lane change is provided to a driving function or not.

[0007] The periphery detection sensor is preferably a monocular or stereo camera. Lane boundaries are understood in the context of the present invention to mean lane markings and guard rails or other raised roadway boundaries, for example construction site boundaries.

[0008] Lane transitions can be calculated from detection points of lane detection, for example, using keypoint regression.

[0009] The degree of agreement can be, for example, a percentage agreement of the lane transitions. Based on this percentage agreement, a confidence value of the calculated lane transition can be determined. This confidence value can then be compared with a previously determined confidence threshold. In this way, it can be determined whether the calculated lane transitions are sufficiently reliable, that a driving function, for example a lane keeping assistance, can function correctly or whether autonomous driving of the planned trajectory is possible.

[0010] In a preferred embodiment, the further lane transitions are retrieved from an HD map, a cloud or an infrastructure. The data source may thus be, for example, an HD map generated from various current sensor data of the ego vehicle and / or the further vehicle. The system may also already have a map based on previous trips and records of the same vehicle surroundings. It is also conceivable to retrieve the lane transitions via a cloud, which may be supplied with current lane transitions, for example, from other vehicles and / or infrastructure. Alternatively or additionally, direct V2X communication may also be performed using infrastructure elements, which may be, for example, located at the side of the lane and which may transmit the actual lane transitions to the corresponding traffic participants. This is advantageous, since in this way always up-to-date and accurate lane transitions are available for verifying the calculated lane transitions.

[0011] In a further preferred embodiment, the calculated lane transition is fed to the driving function if the confidence value exceeds a predefined confidence threshold, thus ensuring that only lane transitions that allow error-free driving function are fed.

[0012] Particularly preferably, when checking the coincidence of the lane transitions, the coincidence of the radii of curvature is checked. In particular, in the case of curves, an accurate recognition of the lane transitions is important in order to prevent erroneous steering interventions. Advantageously, therefore, an adversely recognized or incorrectly recognized curve is determined and the driving function is controlled accordingly.

[0013] Also preferably, if the confidence value is below a predefined confidence threshold, a warning is given to the driver. The warning may also indicate a takeover request to the driver. As an alternative or in addition to the warning, it is also conceivable to provide the captured lane transitions to the driving function instead of the calculated lane transitions. This is particularly advantageous for fully autonomous vehicles, since they can also be configured without a steering wheel, thus making human intervention impossible. In this way, a fallback level is generated, which contributes to an increase in the reliability of the entire system.

[0014] Further advantageous configurations are the subject of the drawings. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is a schematic flow diagram of one embodiment of the present method. [Diagram 2] FIG. 2 is a schematic diagram of one embodiment of the present system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] In Fig. 1, a schematic flow diagram of one embodiment of the method is shown. In a first step S1, a surrounding image or a series of surrounding images is taken using at least one surrounding detection sensor. In step S2, lane lines in the surrounding image or the series of surrounding images are detected. In step S3, lane transitions are calculated based on the detected lane lines. In a further step S4, further lane transitions are retrieved from a data source. In step S5, the calculated lane transitions are validated by verifying the match of the lane transitions. In a further step S6, a degree of match is determined. In step S7, a confidence value is determined based on the degree of match. Finally, in step S8, it is decided whether the calculated lane transitions are provided to a driving function.

[0017] 2 is a schematic diagram of one embodiment of the system 1. The system 1 comprises a perimeter detection sensor 2 and a data processing device 3. The data processing device 3 is connected to the perimeter detection sensor 2 via a data connection D. This application relates to the invention described in the claims, but also includes the following as other aspects. 1. 1. A method for improving safety of driving functions in a partially automated or fully autonomous vehicle, comprising the steps of: - taking (S1) an image or a series of images of the surroundings by means of at least one surrounding detection sensor (2), - detecting lane boundaries in said peripheral image or in said sequence of peripheral images (S2); - calculating a lane transition based on the detected lane boundary line (S3); - A step (S4) of retrieving further lane transitions from a data source; - validating the calculated lane transitions by verifying the coincidence of the lane transitions (S5); - determining the degree of match (S6); - determining (S7) a confidence value based on said degree of match; - determining (S8) whether the calculated lane change is provided to a driving function. 2. 2. The method of claim 1, wherein the further lane transitions are retrieved from an HD map, a cloud or infrastructure. 3. 2. The method of claim 1, wherein the calculated lane transition is provided to a navigation function if the confidence value exceeds a predefined confidence threshold. 4. 4. The method according to any one of 1 to 3 above, wherein, when verifying the match of the lane transition, the match of the radius of curvature is verified. 5. 5. The method of any one of 1 to 4 above, wherein if the confidence value is below a predetermined confidence threshold, a warning is given to the driver. 6. A system (1) for improving the safety of driving functions in a partially automated vehicle or a fully autonomous vehicle, comprising at least one surrounding detection sensor (2) and a data processing device (3), the data processing device (3) being configured to execute any one of the methods 1 to 5 above.

Claims

1. 1. A method for improving safety of driving functions in a partially automated or fully autonomous vehicle, comprising the steps of: - taking (S1) an image or a series of images of the surroundings by means of at least one surrounding detection sensor (2), - detecting lane boundaries in said peripheral image or in said sequence of peripheral images (S2); - calculating lane transitions based on the detected lane boundary lines (S3); - retrieving further lane movements from a data source (S4); - a step (S5) of validating the calculated lane transitions by verifying the coincidence of the lane transitions; - a step (S6) of determining the degree of match; - determining (S7) a confidence value based on said degree of match; - determining whether the calculated lane transition is provided to a driving function capable of accurately functioning lane keeping assistance, and if the reliability value exceeds a predetermined reliability threshold, the calculated lane transition is provided to the driving function, and if the reliability value is below the predetermined reliability threshold, the captured lane transition is provided to the driving function (S8).

2. The method of claim 1 , wherein the further lane transitions are retrieved from an HD map, a cloud or an infrastructure.

3. The method according to claim 1 or 2, wherein when verifying the match of the lane transitions, the match of the radius of curvature is verified.

4. A method according to any one of claims 1 to 3, wherein if the confidence value is below a predefined confidence threshold a warning is given to the driver.

5. A system (1) for improving the safety of driving functions in a partially automated or fully autonomous vehicle, comprising at least one surroundings detection sensor (2) and a data processing device (3), said data processing device (3) being configured to perform the method according to any one of claims 1 to 4.

Citation Information

Patent Citations

  • Travel control system

    JP2016099635A

  • Driving support system for vehicle

    JP2019175097A