Method for creating a map with collision probability

Abstract

The present invention relates to a method for creating a map with collision probabilities for an area, where a number of vehicles traveling in the area are detected, movement data are determined for each vehicle, at least one path is predicted for each of the vehicles based on the movement data, and a collision probability is calculated based on said paths, whereby the collision probabilities can be stored in the map.

Description

[Technical field]

[0001] The present invention relates to a method for creating a map with collision probabilities of an area. [Background technology]

[0002] Intersections or other locations on public transport routes inherently have some risk of accidents. An accident can occur, for example, when two vehicles collide. Information about this can be collected using environmental sensors, for example, allowing infrastructure operators, such as cities or municipalities, to analyze accident hotspots. Summary of the Invention [Problem to be solved by the invention]

[0003] It would be desirable to provide a method for creating a map with collision probabilities, which has an alternative or better design than known designs. This is achieved according to the invention by the method as claimed in claim 1. Advantageous configurations can be gleaned, for example, from the dependent claims. The content of the claims is expressly incorporated into the content of this description by reference. [Means for solving the problem]

[0004] The present invention relates to a method for generating a map with collision probabilities for an area, the method comprising the steps of: - detecting one or more vehicles traveling within the area and determining movement data for each vehicle; - predicting, for each of the vehicles, at least one trajectory based on the movement data; - calculating a collision probability based on the trajectory; - storing the collision probability in a map; The present invention relates to a method having the following structure:

[0005] Such a method can be used to create a map with collision probabilities that is based on actually captured movement data and that may also be based on computational models already used, for example, for vehicle control, such computational models typically not executed by each vehicle in the methods described herein but by an infrastructure that may be specially set up to generate such a map with, for example, collision probabilities.

[0006] The detection of the vehicles and the determination of the movement data can be performed using suitable sensors, such as cameras or motion sensors, but also via data obtained, for example, during V2X communication. For example, for prediction, deterministic algorithms and / or statistical methods and / or models based on artificial intelligence can be used. For example, a collision probability can be calculated to ascertain the probability that trajectories overlap or that areas around trajectories overlap.

[0007] The map may for example be an electronically stored map that may for example be stored in a central unit, which map may then be used for example to assess accident black spots and identify possible ways to improve road safety.

[0008] For example, the movement data can be repeatedly determined while each vehicle moves through the area, and at least one trajectory can be predicted in each case based on the movement data, thereby improving the map since it is possible to rely on a wider range of potential data. However, the corresponding data can also be used for another map.

[0009] In particular, the movement data may be determined at predefined time intervals, which allows for a simple implementation.

[0010] According to one embodiment, multiple trajectories with their associated probabilities are always or at least partially predicted, which applies in particular to each vehicle, so that it is possible to predict how the vehicle will move with what probability, and in particular to assign a probability to each possible movement sequence, which makes it easier to calculate the collision probability.

[0011] The movement data may in particular be detected and / or determined by information received wirelessly from the vehicle. For example, V2X communications may be used for this purpose. However, it is also possible to use roadside sensors such as cameras, radars, lidar sensors, etc.

[0012] In particular, the area may include an intersection, a junction, a curve or a T-junction. Such locations are typically accident hotspots. However, other areas can also be used.

[0013] In particular, the collision probability may be normalized to a baseline value, so that the map may be performed to indicate not absolute probabilities but relative probabilities compared to the baseline value.

[0014] For example, collision probabilities can be stored in a manner that is aggregated to predefined subdivisions of the area, allowing for a proper division of the map to avoid overly detailed design, allowing for specific evaluation in an aggregated form.

[0015] In particular, the prediction uncertainties and / or error bounds in determining the movement data may be taken into account when predicting the trajectory and associated probabilities, which may further improve the calculations. In particular, multiple trajectories with respective probabilities based on the uncertainties and / or error bounds may be calculated.

[0016] In particular, the collision probabilities of multiple vehicle pairs can be stored in an aggregated manner. A pair can be understood to mean, in particular, two vehicles that are close enough together that at least some probability of collision occurs. In order to realize the aggregated evaluation, an aggregated storage can also be used.

[0017] If only one vehicle is considered, then the collision probability, especially with fixed obstacles, may be taken into account, in which case one or more trajectories starting from a single vehicle are usually sufficient.

[0018] In particular, the collision probabilities can be stored in a map such that the map only considers collision probabilities from a predefined time window. As a result, the map can be created to allow, for example, an evaluation of road safety improvements at a certain time, where traffic volumes are typically different at different times. A sliding window function can also be implemented so that the map is always created for a predefined period of time in the past.

[0019] According to one embodiment, one or more maps are generated and only collision probabilities that satisfy one or more predefined conditions are considered in each map. This makes it possible, for example, to generate maps with different characteristics. In the following, some examples are given, particularly with regard to the conditions. - Different maps depending on the forecast horizon, e.g. one map for forecast times of 1 second, 2 seconds, 3 seconds, etc. - maps for different times, e.g. 6a.m.~10a.m., 10a.m.~3p.m., 3p.m.~7p.m., 7p.m.~10p.m., etc., and / or one map for a particular day of the week, - Maps for only certain combinations of objects, e.g. one map for vehicle-vehicle, vehicle-pedestrian, bicycle-pedestrian, bicycle-car, truck-VRU, etc. - A map that does not show the collision probability, but shows the location of the object involved, if the collision probability exceeds a certain threshold. This can be particularly advantageous when determining where the object involved came from or where the structural reasons for the collision risk may be. - a map for different traffic light phases or time until the traffic light phase changes; - maps according to the density of objects, possibly also distinguished by object type, e.g. "very many pedestrians", e.g. one map for a few objects, a normal number of objects, a very large number of objects and an extremely large number of objects in the viewing area, - Maps as deviations from the norm: for example, a map describing the baseline state can first be created, then further maps can be created that show the differences from this baseline state. This can be particularly useful when the consequences of a change are to be shown.

[0020] The method may be particularly implemented such that one or more pre-crash events are determined based on the fact that no vehicle collisions have occurred in locations where the probability of a collision between two vehicles is high, and such pre-crash events are particularly valuable for improving black spot areas with respect to road safety, since unlike real accidents, they cannot be determined based on real events.

[0021] For example, when retrieving collision probabilities from a map, each retrieved collision probability may be assigned to one of a number of predefined areas, and this area may be output in each case. This may in particular mean that the map has a coarser readout than is practically possible, allowing for aggregated views and simplification of evaluation.

[0022] The invention further relates to a computing module configured to carry out the methods described herein.The invention further relates to a non-volatile computer-readable storage medium on which a program code is stored, during the execution of which a processor carries out the methods described herein.For this method, in each case reference may be made in this specification to all the embodiments and variants described herein.

[0023] For example, infrastructure equipment having at least one environmental sensor (e.g., radar, camera, lidar, ultrasonic, etc.) and / or V2X communication module may be considered as a basis. A movement prediction may be created for each detected object. A check is then made to determine whether the movement predictions of two or more objects overlap and therefore there is a risk of collision. Ideally, but not necessarily, both the movement prediction and the collision risk detection are performed implicitly taking into account both the detection error and the inaccuracy of the prediction.

[0024] An example is given below: A vehicle is detected and its position is accurately detected to within ±0.5m, its speed to within ±1m / s and its direction of movement to within ±1°. Now the prediction is made as a kind of sector of the movement range with the most probable path in the center (assuming no errors) and an outer boundary assuming detection errors and changes in driving dynamics during the prediction time.

[0025] The collision risks thus determined may be recorded on a map, which may for example be in the form of a "heat map". For each location and each object combination, a collision risk ranging from 0% to 100% may be added to the other collision risks.

[0026] For better evaluation of the heatmap or map, a grid can be used as the locations for the evaluation, i.e. the collision probability is added up only for locations at a distance of, for example, 10 cm or another distance.

[0027] If not only the absolute collision probability but also the relative collision probability is of interest, i.e. it is sought where an accident is most likely to occur, the map or heat map can also be normalized: for this purpose, the summed collision probability is divided by the maximum collision probability within a given viewing area.

[0028] The collision probabilities can also be summed as a sliding window: only the collision probabilities for the past x seconds, or x minutes, or x hours are summed.

[0029] For differentiation analysis, multiple maps or heat maps may also be created, the possible differences already being described in detail above.

[0030] In particular, the view can be simplified if only clusters of collision probabilities are considered instead of collision probabilities. Thus, collision probabilities can be divided into clusters of, for example, less than 50%, 50%-75%, 75%-90%, and more than 90%. Thus, it is possible, for example, to count the frequency with which each cluster is reached (a dedicated heatmap for each cluster) or each cluster receives an evaluation point and these are summed up (for example, in the above example, this could be 1, 3, 7, 15).

[0031] The map or heat map can also be used to identify so-called "near misses", in particular when a high collision probability is determined at a short prediction time, but no collision occurs. To identify further near misses, i.e. near-crash events, the minimum space-time distance (distance of a four-dimensional space-time vector) for each combination of vehicle trajectories of a sector of driving range that exceeds a certain minimum probability with respect to the collision probability can be calculated. This space-time distance can then be weighted, for example, by the probability of the trajectory pair and summed up. When a threshold value is reached, this weighted space-time distance can be evaluated as a near miss and entered again into the heat map at the position of the minimum distance. The advantage of this second approach is that even a narrow pass at a very clear speed and direction, where the collision probability is not large, is recognized as a near miss.

[0032] In addition to or instead of the heat map, the collision probability can also be provided as an additional function or device of the system. This can be done, for example, in the form of raw data or as a trigger when the collision probability exceeds a certain value. Hotspots and near misses can be identified based on relatively well-known methods.

[0033] It is also possible for drivers to identify situations or locations that are unpleasant or difficult to deal with. This can be used to make structural changes or adjust traffic flow control before an accident occurs.

[0034] In the following, the invention will be explained with reference to the drawings. [Brief description of the drawings]

[0035] [Figure 1] Shows a situation where there are two vehicles ahead at an intersection. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0036] Fig. 1 shows purely diagrammatically a first vehicle 10 and a second vehicle 20. The first vehicle 10 moves on a first road S1 and the second vehicle 20 moves on a second road S2. Both vehicles 10, 20 are moving on the roads S1, S2 towards a junction K where the two roads S1, S2 intersect. The first vehicle 10 has a V2X communication module 15 fitted with an antenna 17. The second vehicle 20 has a V2X communication module 25 fitted with an antenna 27. This allows the two vehicles 10, 20 to take part in V2X communication.

[0037] A roadside V2X communication module 45 having an antenna 47 is located beside the roads S1, S2. This also enables the vehicles 10, 20 to communicate with the roadside infrastructure. A computing unit 30 is located beside the roads S1, S2 and can be used to create maps.

[0038] Furthermore, a camera 50 is placed beside the roads S1, S2, which is shown here diagrammatically and is capable of capturing the two vehicles 10, 20. The camera 50 is an infrastructure-side environmental sensor.

[0039] When the vehicles 10, 20 approach the intersection K, they are captured via the camera 50 and V2X communication. The data collected in this process is sent to the computing unit 30. The mentioned mechanisms are also used to capture the position, course and speed of the vehicles 10, 20 together with their respective errors. The computing unit 30 is designed to generate trajectory predictions and associated probabilities for each of the vehicles 10, 20 at several times when they approach the intersection K. In this case, the computing unit 30 calculates multiple trajectories of each vehicle starting from each time point at which the corresponding measurements are made and assigns a specific probability to each trajectory. Then, based on these trajectories, the collision probability at the intersection K is calculated, i.e. it is calculated how likely a collision is to occur at any location. This can be used to generate a heat map, i.e. an electronic map, showing the respective collision probabilities at specific points of the intersection K. The map can be normalized if necessary or can be created based only on specific data, for example only on data recorded at a specific time. Such maps can help planners identify accident hotspots and optimize them to improve road safety.

[0040] It is to be noted that in general, V2X communication is understood to mean in particular direct communication between vehicles and / or between vehicles and infrastructure devices. Thus, by way of example, this can be vehicle-to-vehicle or vehicle-to-infrastructure communication. When vehicle-to-vehicle communication is mentioned in the present application, said communication can for example essentially take place as part of vehicle-to-vehicle communication, which is typically implemented without switching by a mobile radio network or a similar external infrastructure and must therefore be distinguished from other solutions, for example based on a mobile radio network. By way of example, V2X communication can be implemented using the IEEE 802.11p or IEEE 1609.4 standard. Examples of other communication technologies include LTE-V2X, 5G-V2X, C-V2X, WLAN, WiMax, UWB or Bluetooth. V2X communication can also be referred to as C2X communication. The subareas can be referred to as C2C (vehicle-to-vehicle) or C2I (vehicle-to-infrastructure). However, the present invention does not explicitly exclude V2X communication, for example with switching via a mobile radio network.

[0041] The above-mentioned steps of the method according to the invention can be carried out in the order shown. However, if technically feasible, they can also be carried out in a different order. For example, in one of its embodiments having a specific combination of steps, the method according to the invention can be carried out such that no other steps are carried out. However, in principle, further steps can also be carried out, including steps not mentioned.

[0042] For example, for ease of understanding, features may be described in combination in the claims and in the description of this specification, but it should be noted that they may also be used separately from one another. Those skilled in the art will recognize that such features may also be combined with other features or combinations of features independently of one another.

[0043] The references in the dependent claims may characterize preferred combinations of the respective features but do not exclude other combinations of features. [Explanation of symbols]

[0044] K intersection S1 First Road S2 Second Road 10 First Car 15 V2X communication module 17 Antenna 20 Second Car 25 V2X communication module 27 Antenna 30 Computing Units 45 Roadside V2X communication module 47 Antenna 50 Cameras / Environmental Sensors

Claims

1. 1. A method of generating a map having a collision probability of an area, comprising the steps of: - detecting one or more vehicles (10, 20) traveling within said area and determining movement data relating to each of said vehicles (10, 20); - predicting, for each of said vehicles (10, 20), at least one trajectory based on said movement data; - calculating a collision probability based on said trajectory; storing said collision probabilities in said map; The method according to claim 1,

2. 2. The method of claim 1, wherein the movement data is determined repeatedly while each of the vehicles (10, 20) moves through the area, and at least one trajectory is predicted in each case based on the movement data.

3. The method of claim 2 , wherein the movement data is determined at predetermined time intervals.

4. The method according to any one of claims 1 to 3, wherein a plurality of trajectories with their associated probabilities are always or at least partially predicted.

5. The method according to any one of claims 1 to 4, wherein the movement data is detected and / or determined by one or more roadside environmental sensors (50).

6. The method according to any one of the preceding claims, wherein movement data is detected and / or determined by information received wirelessly from the vehicle (10, 20).

7. The method according to any one of claims 1 to 6, wherein the area comprises a crossroads (K), a junction, a curve or a T-junction.

8. The method of any one of claims 1 to 7, wherein the collision probability is normalized to a reference value.

9. A method according to any one of claims 1 to 8, wherein the collision probabilities within predefined subdivisions of the area are stored in an aggregated manner.

10. The method according to any one of claims 1 to 9, wherein prediction uncertainties and / or error bounds in determining the movement data are taken into account when predicting the trajectory and associated probabilities.

11. The method of any one of claims 1 to 10, wherein the collision probabilities for multiple vehicle pairs are stored in an aggregated manner.

12. A method according to any one of claims 1 to 11, wherein the collision probabilities are stored in a map such that the map only considers collision probabilities from a predefined time window.

13. One or more maps are generated; A method according to any one of the preceding claims, wherein only collision probabilities which satisfy one or more predefined conditions are considered for each map.

14. The method according to any one of the preceding claims, wherein the one or more near-collision events are determined based on the fact that a collision of the vehicles (10, 20) did not occur in a location where there is a high probability of a collision between the two vehicles.

15. A method according to any one of the preceding claims, wherein when retrieving collision probabilities from the map, each retrieved collision probability is assigned to one of a number of predefined regions, and this region is output in each case.

Images