EP-4020406-B1 - VEHICLE-BASED MEASUREMENT OF SIGNAL OBJECT INTEGRITY

Inventors

• POHL, DANIEL
• TANRIOVER, Cagri
• ALVAREZ, Ignacio J.
• FOX, MAIK

Dates

Publication Date

20260506

Application Date

20210920

Claims (11)

1. A method for vehicle-based measurement of signal object integrity, the method comprising: obtaining, by processing circuitry of a vehicle, map data indicating a presence of a signal object at a corresponding location, wherein the signal object is a traffic sign, a light signal, or a road surface marking; obtaining, by the processing circuitry of the vehicle, sensor data representing an environment of the vehicle, the sensor data originating from a sensor mounted to the vehicle, wherein the environment of the vehicle comprises a region which is expected to comprise the signal object at the corresponding location; determining a bound based on the map data, the bound comprising a portion of the sensor data corresponding to the region of the environment; determining a result using a classifier which is configured to output the result responsive to inputting the bound into the classifier, wherein the bound defines an area within the sensor data for input to the classifier, and wherein the result indicates a degradation of a signal ability of the signal object, the signal ability representing an integrity of the signal object; and communicate, by a transmitter of the vehicle, the result of the classifier.
2. The method of claim 1, wherein the traffic sign is a stop sign, a yield sign, a speed limit sign, a street sign, or an exit sign.
3. The method of claim 1 or 2, wherein the result comprises a confidence score below a threshold.
4. The method of claim 3, wherein the classifier is a first classifier, the method further comprising: determining a type of the degradation using another classifier which is configured to output the type of the degradation responsive to inputting the bound into the second classifier.
5. The method of claim 4, wherein the second classifier is further configured to output a degree of the degradation.
6. The method of any of claims 4-5, wherein the degradation is at least one of occlusion, deformation, degradation of surface markings, absence of surface markings, or additional markings.
7. The method of any of claims 1-6, wherein the result includes: a representation of the bound with respect to the sensor data; a type of signal object detected; and a confidence score for the signal object detected.
8. The method of claim 7, wherein the representation of the bound includes a bounding box for two-dimensional data.
9. The method of any of claims 7-8, wherein the confidence score indicates the degradation of the signal ability of the signal object, and wherein the result includes: a classification of the degradation of the signal ability; or a compressed representation of the sensor data within the bound.
10. At least one machine-readable medium including instructions that, when executed by a machine, cause the machine to perform any method of claims 1-9.
11. A system comprising means to perform any method of claims 1-9.

Description

TECHNICAL FIELD Embodiments described herein generally relate to automated vehicle sensing and more specifically to vehicle-based measurement of signal object integrity. BACKGROUND Automated vehicles include technologies to perform autonomous or semi-autonomous travel, often referred to as "self-driving" or "assisted-driving" in reference to automobiles. These systems use an array of sensors to continuously observe the vehicle's motion and surroundings. A variety of sensor technologies may be used to observe the vehicle's surroundings, such as the road surface and boundaries, other vehicles, pedestrians, objects and hazards, signal objects (e.g., signage or road markings), and other relevant items. Automated vehicles also generally include processing circuitry to process the sensor data and may include actuators to control a vehicle. Image-capture sensors that are implemented with one or more cameras are particularly useful for object detection and recognition, such as reading signal objects like signs (e.g., stop signs, speed limits signs, street signs, etc.) or road markings (e.g., dividing lines, turn arrows, etc.). Such image capture sensors generally cover several fields of view (FOV) around the vehicle. Relevant prior art includes: J. Guo et al., Detection of Occluded Road Signs on Autonomous Driving Vehicles, 2019 IEEE International Conference on Multimedia and Expo (ICME), Shanghai, China, 2019, pp. 856-861, doi: 10.1109/ICME.2019.00152, relating to the detection of occluded road signs from autonomous driving vehicles;P. Hienonen et al., Framework for Machine Vision Based Traffic Sign Inventory, In: Sharma, P., Bianchi, F. (eds) Image Analysis, SCIA 2017, Lecture Notes in Computer Science, vol 10269, relating to the detection of traffic signs and assessing the condition of the recognized signs. BRIEF DESCRIPTION OF THE DRAWINGS In the drawings, which are not necessarily drawn to scale, like numerals may describe similar components in different views. Like numerals having different letter suffixes may represent different instances of similar components. The drawings illustrate generally, by way of example, but not by way of limitation, various embodiments discussed in the present document. FIG. 1 illustrates an example of an environment including a vehicle with a system for vehicle-based measurement of signal object integrity, according to an embodiment.FIG. 2 illustrates an example of a vehicle with a system for vehicle-based measurement of signal object integrity, according to an embodiment.FIG. 3. Illustrates several signal-object integrity scenarios and corresponding communications from a vehicle, according to an embodiment.FIG. 4 illustrates a flow diagram of an example of a method for vehicle-based measurement of signal object integrity, according to an embodiment.FIG. 5 is a block diagram illustrating an example of a machine upon which one or more embodiments may be implemented. DETAILED DESCRIPTION Signal objects are often important to provide effective communication of roadway traffic flow and navigation, increasing safety concerns in an increasing traffic flow on roadways. However, because signal objects are physical objects, signal objects may be damaged over time. Even if the signal object itself is undamaged, its ability to perform its signal function may be damaged, as occurs when a stop sign is hidden behind foliage. Due to the usually large number of signal objects, detecting when signal object integrity is too low can be a challenge. Thus, safety or efficiency of roadways may be compromised. Current attempts to address signal object integrity monitoring often involve reporting by the public to an entity responsible for the signal object (e.g., citizen complaints to a city), observation after an incident has occurred (e.g., a traffic collision), or refurbishment to an area of roadway for other reasons. These activities are generally labor intensive and unpredictable, with some (such as post-collision analysis) being untimely. The prevalence of semi-autonomous automated vehicles, and the emergence of fully autonomous automated vehicles, provides an opportunity to address current issues with signal object integrity measurement. Instead of manually checking signal objects to determine whether signal integrity is acceptable, automated vehicles may be used to detect and report such issues. Automated vehicles generally include some facility for recognizing signal objects. This facility may be extended to identify damage in signal objects, prioritize signal objects, and communicate the results. For example, signal object maintenance may be prioritized based on the recognition confidence crowdsourced from several automated vehicles over time. Further, map data (e.g., high-definition or high-resolution maps) may be compared to automated vehicle observations to recognize signal objects. If there are discrepancies-e.g., because a sign is damaged or missing-a report may be automatically generated