Method for determining elevated trains, related operating systems and vehicles

Abstract

The present invention relates to a method (8) for determining elevated trains (6.3) for a driving system (2). According to the method (8), sensor data is received from sensors (3,4) including at least one radar detector (4). An object (6) is then detected and classified based on the received sensor data. It is determined whether a candidate object (6) is classified as an elevated train-compatible object and whether the candidate object (6) was detected based solely on radar data. If both conditions are true, the candidate object (6) is labeled as potentially an elevated train. Next, it is determined whether at least one other candidate object (6) is labeled as potentially an elevated train. If yes, it is determined whether the candidate object (6) and the other candidate object (6) have features that make them compatible with elevated trains (6.3). If yes, the candidate object (6) is classified as likely to be an elevated train, and information regarding the likely elevated train object (6) is provided to the function of the driving system (2).

Description

[Technical Field]

[0001] The present invention relates to object detection for driving assistance systems and / or autonomous driving, and in particular to a method for determining elevated trains for driving systems, more specifically advanced driving assistance systems and / or autonomous driving systems, a driving system configured to determine elevated trains, and a vehicle comprising said driving system. [Background Art]

[0002] Object detection is an important element of driving assistance systems and / or autonomous driving systems, and many functions of said driving assistance systems and / or autonomous driving systems depend on the results of object detection.

[0003] When an elevated train is present in a scene and the detection of said object is performed by a radar detector, a specific problem relating to object detection arises as follows. Since radar detectors cannot accurately measure the height of an object from the ground, elevated trains may be misinterpreted as objects on the road even though they are at different altitude levels and do not affect vehicles on the road. Such misinterpretation may cause the driving assistance system and / or the autonomous driving system to take incorrect actions (such as incorrect automatic emergency braking). [Summary of the Invention] [Problem to be Solved by the Invention]

[0004] Accordingly, it is an object of the present invention to provide an object detection method that solves the above-described problems. In particular, it is an object of the present invention to provide a method for determining an elevated train, a related driving system, and a vehicle comprising such a driving system. [Means for Solving the Problem]

[0005] The object of the present invention is resolved by the subject matter of the independent claim, and further embodiments are incorporated into the dependent claims.

[0006] According to an aspect of the present invention, a method for determining elevated trains for an operating system is provided.

[0007] This method may be performed by a computer. Specifically, this method may be performed by a computing unit, in particular by a computing unit of an operating system. The computing unit may include one or more processors, and this method may be performed using several processors by parallel computing and / or distributed computing.

[0008] An elevated train is a train that operates at a higher elevation than ground level. In this context, ground level refers to the altitude at which the train travels. In other words, an elevated train is at a higher elevation than the train itself. More specifically, an elevated train is positioned at a height that allows cars to pass underneath without colliding with it.

[0009] The driving system may be a driver assistance system, in particular an advanced driver assistance system, and / or an autonomous driving system.

[0010] According to this method, sensor data is received from sensors in a vehicle's driving system. In the context of this application, a vehicle may be a road vehicle, particularly an automobile, bus, truck, or motorcycle. The sensors include at least one radar detector and may further include a plurality of other sensors such as cameras (RGB cameras, monochrome cameras, infrared cameras, etc.) and LiDAR detectors. The sensors may acquire radar data and / or image data, and may provide such data to the processing unit of the driving system. According to this method, the sensor data includes radar data.

[0011] Objects are detected based on received sensor data. When using image data, object detection may be performed, for example, by edge detection and subsequent shape analysis, or by artificial intelligence such as a deep neural network. When using radar data, object detection may be performed, for example, by shape recognition technology or artificial intelligence. If sensor data from multiple sensors is available, the sensor data may be combined to obtain more accurate results. When analyzing multiple sets of sensor data acquired at different times, it may be further confirmed whether the detected object remained for a predetermined length of time. If it did not remain for the predetermined length of time, it may be considered a false positive. The distance to the object, the object's position, and the object's size may be determined using a radar detector.

[0012] Next, the detected objects are classified. This classification may be performed, for example, based on the size and / or shape of the objects, or it may be performed for each of the detected objects. The classification classes may include, for example, pedestrians, bicycles, motorcycles, automobiles, trucks, or unclassified objects.

[0013] For each candidate object among the detected objects, the following steps are performed. In this context, the term "candidate object" refers to any one of the detected objects and is selected to identify the object to which the steps are performed.

[0014] This determines whether a candidate object is classified as an elevated train-compatible object, that is, whether the candidate object could be part of an elevated train. Considering the examples of classification classes mentioned above, objects classified as pedestrians, bicycles, motorcycles, or automobiles would be too small to be part of an elevated train. Therefore, only trucks or unclassified objects are elevated train-compatible objects.

[0015] Furthermore, it is determined whether the candidate object was detected based solely on radar data. That is, it is checked which sensor data was used to detect the candidate object, more specifically, whether only radar data was used, or whether other sensor data was also used.

[0016] If a candidate object is classified as an elevated train-compatible object and it is determined that the candidate object was detected based solely on radar data, the following steps are performed. On the other hand, if a candidate object is not classified as an elevated train-compatible object, or if the candidate object is detected based on sensor data other than radar data, the following steps to determine whether the candidate object may be part of an elevated train are not necessary. If a candidate object is classified as not being elevated train-compatible, it is unlikely that the candidate object is an elevated train. On the other hand, if sensor data other than radar data, especially camera data, is also used, the height of objects above ground level can be measured, which already indicates whether or not the candidate object is part of an elevated train.

[0017] Next, the candidate objects are labeled as potentially being elevated trains. For example, individual flags may be set for each candidate object.

[0018] Furthermore, it is determined whether at least one other candidate object can be labeled as a potential elevated train. This is preferably done after performing the above steps for all candidate objects, as it clarifies which of the detected objects could be elevated trains.

[0019] If at least one other candidate object could be an elevated train, determine whether the candidate object and the other candidate objects have features that make them fit an elevated train. This determination may be repeated for each of the other candidate objects that could be an elevated train. If the candidate object and at least one other candidate object have features that make them fit an elevated train, the candidate object is classified as likely to be an elevated train, that is, it is determined that the candidate object is likely to be an elevated train. In this context, features that make an elevated train fit mean that the candidate object and at least one other candidate object have features that allow them to be considered as two cars of an elevated train, in particular two adjacent cars.

[0020] Finally, information regarding objects that are likely to be elevated trains is provided to at least one other function of the driving system. Thus, the other function of this driving system has information about which objects are likely to be elevated trains and can take action according to that information. This improves the performance of the driving system, increases the reliability of the driving system, and enhances the safety of the vehicle.

[0021] According to one embodiment, which can be combined with the embodiments described above or any further embodiments described later, the elevated train is an elevated monorail or a suspended railway. Both of these elevated trains are often found near roads and / or intersecting roads and are at a height that makes them easily detectable by radar. Therefore, these two types of elevated trains are most relevant to the present method.

[0022] According to one embodiment, which can be combined with the embodiments described above or any further embodiments described later, the radar detector is a millimeter-wave radar detector. Since this is a standard detector used in many cars, this method can also be used to improve the driving system of an existing car, for example, by simply updating the software.

[0023] According to one embodiment, which can be combined with the embodiments described above or any further embodiments described later, the radar detector is a short-range radar detector. In this context, a short-range radar detector may be a radar detector with a wide detection range but low height resolution. Furthermore, a short-range radar detector may have a maximum range of 100m, more specifically 50m, and even more specifically 30m. Low height resolution of a short-range radar detector may make it impossible to distinguish between elevated trains and ground-level vehicles. Therefore, the method described above is particularly useful for such radar detectors.

[0024] According to one embodiment, which can be combined with the embodiments described above or any further embodiments described later, elevated train-compatible objects include trucks and / or unclassified objects. These are objects that may resemble elevated train vehicles. On the other hand, other objects such as pedestrians, bicycles, motorcycles, or automobiles do not resemble elevated train vehicles. If there are more classification classes, elevated train-compatible objects are determined based on the likelihood that elevated train vehicles can be identified as objects of a given classification class.

[0025] According to one embodiment that can be combined with the embodiments described above or any further embodiments described later, a candidate object and other candidate objects having features suitable for elevated trains include the size of the candidate object and the size of the other candidate object matching within a predetermined size uncertainty range. This size uncertainty may depend on the resolution of the radar detector, but may be, for example, 2 m in the horizontal direction. This is a distinctive feature because the individual cars of an elevated train are usually the same size. Alternatively or in addition to this, a candidate object and other candidate objects having features suitable for elevated trains include the distance between the candidate object and the other candidate object being within a predetermined range. Here, an absolute distance, a horizontal distance, or a depth distance, i.e., the difference in distance from the car, may be selected. Such a horizontal distance may be, for example, within 10 m, and a depth distance may be, for example, within 1.5 m. This is also a distinctive feature because the individual cars of an elevated train are usually arranged side by side. Alternatively or in addition to this, a candidate object and other candidate objects having features suitable for elevated trains include the speed of the candidate object and the speed of the other candidate object matching within a predetermined speed uncertainty range. Here too, when the elevated train is moving, especially with respect to adjacent cars, the speeds of individual cars are the same, and their velocities are very similar. Therefore, the velocity uncertainty may be, for example, 1.5 m / s along the direction of movement of the object and 0.5 m / s perpendicular to the direction of movement of the object. The coincidence of velocities is also a notable feature.

[0026] According to an embodiment that can be combined with the above-described embodiment or any further embodiment described below, after detecting an object based on received sensor data, the method further comprises determining whether the detected object is likely to intersect with the path of the vehicle, and performing the remaining steps of the method only for objects that are likely to intersect with the path of the vehicle. Since determining whether an object is an elevated train is most critical for objects that may actually collide with the host vehicle, performing the method only on such objects can save calculation time and resources.

[0027] According to an embodiment that can be combined with the above-described embodiment or any further embodiment described below, before providing information about an object that is highly likely to be an elevated train to a function of a driving system, the method further comprises increasing the elevated train probability of each candidate object classified as highly likely to be an elevated train by a predetermined increment, and decreasing the elevated train probability of each candidate object not classified as highly likely to be an elevated train by a predetermined decrement. Here, a newly detected object may be initialized with an elevated train probability of 0. Then, the information about the object that is highly likely to be an elevated train provided to the function of the driving system includes the elevated train probability of the object. The increase and decrease of the elevated train probability may each be limited to a predetermined value, for example, 10, 0, etc. As an example, the elevated train probability of the object is continuously increased based on the stability of confirmation until a predetermined threshold is reached and the object is confirmed as highly likely to be an elevated train by setting an individual flag. Similarly, if confirmation is unstable, the elevated train probability of the object is continuously decreased. Taking multiple radar data into account improves the determination of whether an object is part of an elevated train.

[0028] According to an embodiment that can be combined with the above-described embodiment or any further embodiment described below, the function of the driving system is an emergency braking function, and based on the determination that an object is an elevated train or a part of an elevated train, said function suppresses automatic emergency braking for said object. Since the elevated train is located above the vehicle, a collision between the elevated train and the vehicle cannot occur. Suppressing automatic emergency braking for elevated trains reduces the number of erroneous emergency brakings, improves user convenience, and also significantly enhances safety.

[0029] According to another aspect of the present invention, a driving system is provided. The driving system includes at least one radar detector and an arithmetic unit, and is configured to operate in accordance with the above description. Advantages and further embodiments correspond to those provided in the above description.

[0030] According to an embodiment that can be combined with the above-described embodiment, the driving system is a driving assistance system and / or an automated driving system. In any case, determining an object as an elevated train improves the operation of the driving system.

[0031] According to yet another aspect of the present invention, a vehicle is provided. The vehicle includes the driving system according to the above description. Therefore, advantages and further embodiments correspond to those provided in the above description. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] These and other aspects of the present invention will become apparent from the embodiments described as examples in the following description and from the accompanying drawings below, and will be further clarified with reference to them. [Figure 1] A schematic top view of an embodiment of a vehicle provided with a driving system is shown. [Figure 2] A schematic example of a traffic scene is shown. [Figure 3] A flowchart of an embodiment of a method for determining an elevated train for a driving system is shown.

[0033] In the figures, elements corresponding to elements already described may have the same reference numeral. Examples, embodiments, or any features, whether indicated as non-limiting or not, should not be construed as limiting the claimed invention. [Modes for carrying out the invention]

[0034] Figure 1 shows a schematic top view of an embodiment of a vehicle 1 equipped with a driving system 2. The driving system 2 has a camera 3, four short-range radar detectors 4, and a computing unit 5, with the camera 3 and radar detectors 4 connected to the computing unit 5. In the figure, the above connections are shown as wired connections, but wireless connections are also possible. The coverage angles of the radar detectors 4 are shown by dashed lines. The driving system 2 may further include additional cameras, long-range radar detectors, or other sensors such as LiDAR, but these are not shown here for clarity. The driving system 2 may also be a driver assistance system, in particular an advanced driver assistance system, and / or an autonomous driving system. The computing unit 5 receives sensor data from sensors 3 and 4 and processes the sensor data. Typically, object detection is performed based on the sensor data, the detected objects are tracked, and assistance functions, emergency functions, and / or autonomous driving functions are activated based on the tracked objects.

[0035] Figure 2 shows a schematic example of a traffic scene as observed by vehicle 1, including three objects 6: a car 6.1, a truck 6.2, and an elevated train 6.3 (shown here as a suspended railway). Cars 6.1 and truck 6.2 travel at ground level 7, similar to vehicle 1 observing the traffic scene, while the elevated train 6.3 travels at an elevated level higher than ground level 7. As can be seen from the figure, there is no risk of collision between the elevated train 6.3 and any of the cars at ground level 7. However, because the short-range radar detector 4 has low vertical resolution, it is not easy to distinguish between the elevated train 6.3 and the objects 6 at ground level 7 using only short-range radar data.

[0036] To overcome this problem, a method 8 for determining elevated trains 6.3 for the driving system 2 can be used, as shown in Figure 3.

[0037] According to Method 8, sensor data is received from sensors 3 and 4 of the driving system 2 (9). Based on the received (9) sensor data, object 6 is detected (10). Then, object 6 is classified (11).

[0038] Next, it is determined whether candidate object 6 is classified as an elevated train compatible object (12). For example, pedestrians, bicycles, motorcycles, and automobiles 6.1 are too small to be considered vehicles of an elevated train 6.3, so candidate object 6 classified into any of these classification classes is not elevated train compatible. In that case, it is determined that object 6 is not an elevated train 6.3 (13).

[0039] If candidate object 6 is determined to be suitable for an elevated train (12), and is classified as, for example, track 6.2 or an unclassified object, it is further determined whether candidate object 6 was detected based solely on radar data (14). If it was not detected based solely on radar data, i.e., if object 6 was detected using other data as well, it is assumed that height information can be obtained from other sensor data, and based on that other sensor data, it is determined whether object 6 is part of an elevated train 6.3 (15).

[0040] If object 6 is detected based solely on radar data, object 6 is labeled as a possible elevated train (16). Next, it is determined whether at least one other candidate object 6 can be labeled as a possible elevated train (17). If none of the other object 6 can be labeled as a possible elevated train, object 6 is determined not to be an elevated train 6.3 (13).

[0041] If it is determined that there is at least one other candidate object 6 that has been labeled as potentially being an elevated train (17), then it is determined whether candidate object 6 and the other candidate object 6 have features that conform to elevated train 6.3, such as being approximately the same size, being located next to each other, and / or having approximately the same speed (18). If it is determined that the features of candidate object 6 and the other candidate object 6 do not conform to elevated train 6.3 (18), then it is determined that object 6 is not an elevated train 6.3 (13).

[0042] On the other hand, if candidate object 6 and other candidate objects 6 have features that match elevated train 6.3, candidate object 6 is classified as having a high probability of being an elevated train (19).

[0043] Finally, information regarding object 6, which is highly likely to be an elevated train, is provided to the functions of the driving system (20). Furthermore, information regarding object 6, which has been determined not to be an elevated train 6.3 (13), is provided to the functions of the driving system 2 (20), and the result of determining whether object 6 is part of an elevated train 6.3 based on other sensor data (15) is provided to the functions of the driving system 2 (20).

[0044] Having such information regarding whether or not object 6 is part of elevated train 6.3 allows the operation of the driving system 2 to provide better and more reliable results, thereby improving the convenience and safety of operation.

[0045] Other variations of the disclosed embodiments can be understood and achieved by those skilled in the art in carrying out the claimed invention, based on a review of the drawings, this disclosure, and the appended claims. In the claims, the word “comprising” does not exclude other elements or processes, and the indefinite article “a” or “an” does not exclude the plural. The mere fact that certain means are described in different dependent claims does not mean that combinations of these means cannot be used advantageously. Reference numerals in the claims should not be construed as limiting the scope of the claims. [Explanation of Symbols]

[0046] 1: car 2: Driving System 3: Camera 4: Radar detector 5: Processing Unit 6: Object 6.1: Automobiles 6.2: Trucks 6.3:Elevated train 7: Ground level 8: Method 9: Received 10: Detection 11:Classification 12: Judgment 13: Judgment 14: Judgment 15: Judgment 16: Labeling 17: Judgment 18: Judgment 19:Classification 20: Provided

Claims

1. A method (8) for determining an elevated train (6.3) for an operating system (2), The process involves receiving sensor data from sensors (3, 4) of the driving system (2) of a vehicle (1), wherein the sensors (3, 4) include at least one radar detector (4), and the sensor data includes radar data. Detecting an object (6) based on the received sensor data, Classifying the detected objects (6), For each candidate object (6) among the detected objects (6), Determine whether the above candidate object (6) is classified as an elevated train compatible object. To determine whether the above candidate object (6) was detected based solely on radar data, Based on the determination that the above candidate object (6) is classified as an elevated train compatible object, and that the above candidate object (6) was detected based solely on radar data, Label the above candidate object (6) as potentially being an elevated train. Determine whether at least one other candidate object (6) can be labeled as a potentially elevated train. Based on the determination that the above other candidate object (6) is labeled as potentially being an elevated train, To determine whether the above candidate object (6) and the above other candidate object (6) have features that are suitable for an elevated train (6.3), and Based on the determination that the above candidate object (6) and the above other candidate object (6) have characteristics that make them suitable for an elevated train (6.3), The above candidate object (6) is classified as having a high probability of being an elevated train, and To provide the function of the above-mentioned driving system (2) with information regarding object (6) which is highly likely to be an elevated train. Method (8) including the following.

2. The method according to claim 1 (8), wherein the elevated train (6.3) is an elevated monorail or a suspended railway.

3. The method according to claim 1 or 2 (8), wherein the radar detector (4) is a millimeter-wave radar detector.

4. The method according to any one of claims 1 to 3 (8), wherein the radar detector (4) is a short-range radar detector.

5. The method according to any one of claims 1 to 4 (8), wherein the elevated train-compatible object includes a track (6.2) and / or an unclassified object.

6. The above candidate object (6) and the above other candidate object (6) having features suitable for elevated trains (6.3) are, The size of the above candidate object (6) and the size of the above other candidate object (6) must match within a predetermined size uncertainty range. The distance between the above candidate object (6) and the above other candidate object (6) is within a predetermined range, and / or The velocity of the above candidate object (6) and the velocity of the above other candidate object (6) coincide within a predetermined range of velocity uncertainty. The method according to any one of claims 1 to 5, including (8).

7. After the step of detecting an object (6) based on the received sensor data, To determine whether the detected object (6) is likely to intersect with the path of the vehicle (1), and Perform the remaining steps of this method only for objects (6) that are likely to intersect with the path of the vehicle (1) mentioned above. The method according to any one of claims 1 to 6, further comprising (8).

8. Before the step of providing information about an object (6) that is highly likely to be an elevated train to the functions of the driving system (2), The probability of each candidate object (6) classified as likely to be an elevated train is increased by a predetermined increment, and The probability of each candidate object (6) that was not classified as likely to be an elevated train is reduced by a predetermined amount. It further includes, The information regarding object (6), which is highly likely to be an elevated train, includes the probability that object (6) is an elevated train. The method according to any one of claims 1 to 7 (8).

9. The method according to any one of claims 1 to 8 (8), wherein the function of the above-mentioned driving system (2) is an emergency braking function, and based on the determination that the object (6) is an elevated train (6.3) or part of an elevated train (6.3), the above function suppresses automatic emergency braking on the object (6).

10. An operating system (2) comprising at least one radar detector (4) and a computing unit (5), the operating system (2) is configured to operate according to the method (8) of any one of claims 1 to 9.

11. The driving system (2) described above is a driving assistance system and / or an automated driving system, according to claim 10.

12. A vehicle (1) equipped with the driving system (2) according to claim 10 or 11.

Images