Automatic driving vehicle zebra crossing comity method, device and equipment and storage medium

Abstract

The invention provides an automatic driving vehicle zebra crossing comity method, device and equipment and a storage medium, and the method comprises the steps: querying the nearest zebra crossing information according to the planned path and map information of an automatic driving vehicle; obtaining the minimum observation range of the zebra crossing according to the nearest zebra crossing information; when a shielding obstacle exists in the minimum observation range of the zebra crossing, obstacle movement information of the shielding obstacle is collected; calculating the avoidance confidence coefficient of the shielded obstacle based on the obstacle motion information; when the yielding confidence coefficient is greater than a yielding threshold value, the automatic driving vehicle slows down or stops to give way to pedestrians; and when the yielding confidence coefficient is smaller than a yielding threshold value, the automatic driving vehicle passes by following the vehicle. According to the method, the avoidance confidence is obtained by analyzing the shielded condition, the potential shielded pedestrians can be subjected to commission, higher safety is achieved, normal driving can be referred to under the conditions that normal driving is shielded by obstacles and no avoidance intention exists, and the running time efficiency can be guaranteed to a large extent.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle driving, and particularly relates to a method, device, equipment and storage medium for an autonomous vehicle to yield to pedestrians at a zebra crossing. Background Art

[0002] Currently, from the perspective of pedestrian safety, it is required that motor vehicles yield to pedestrians in front of a zebra crossing. Among them, the yielding behavior can be understood as: when a motor vehicle driver discovers that a pedestrian is performing an act of passing through the zebra crossing, the vehicle needs to be decelerated or stopped to ensure the safe passage of the pedestrian. Although the yielding behavior is relatively easy to understand, in actual situations, even experienced drivers find it difficult to accurately judge the intention of pedestrians. Because pedestrians or non-motor vehicle pushers, etc., as important participants in the traffic scene, have a relatively large randomness in their running trajectories and their own movement intentions. Therefore, it is also difficult to yield according to their predicted trajectories. In addition, obstacles such as pedestrians and bicycles are small in size and are easily blocked by other larger obstacles, thus forming a "ghost emergence" scenario.

[0003] All of the above series of problems are the difficulties in judging reasonable yielding behaviors for autonomous vehicles in front of a zebra crossing.

[0004] Based on this, there are two common types of existing solutions for autonomous vehicles to yield to pedestrians, which are as follows:

[0005] 1. One is to determine the starting position of the crosswalk and the influence range of the crosswalk in the planned driving path; according to the influence range of the crosswalk, obtain the set of pedestrians intending to pass through the crosswalk and the movement information of each pedestrian in the pedestrian set to obtain pedestrian reference information; according to the pedestrian reference information and the driving information of the host vehicle, determine the risk assessment value of each pedestrian colliding with the host vehicle; if the risk assessment value of any pedestrian colliding with the host vehicle is greater than a preset risk value threshold, a virtual wall is generated at the starting position of the crosswalk.

[0006] 2. Another method for yielding to pedestrians is similar to the above method, the difference being that: instead of performing individual detection on pedestrians within the zebra crossing range, a clustering algorithm is used to output the results of the clustered pedestrian flow and non-motor vehicle flow as the basis for judging whether to yield.

[0007] Both of the above two solutions require an unobstructed view of the host vehicle, and the courtesy is carried out on the basis that the perception module can normally perceive obstacles such as pedestrians and bicycles crossing in front. However, in real life, there are often occluded scenarios, such as: there are stationary social vehicles parked in front of the zebra crossing, causing a certain degree of occlusion to non-motor vehicles or pedestrians crossing; or in the case of multiple lanes, the social vehicles moving in the side front of the adjacent lane of the host vehicle will also cause a certain degree of occlusion to non-motor vehicles or pedestrians crossing. In the above situations and other similar scenarios, non-motor vehicles or pedestrians are occluded to a certain extent, and the perception ability of the perception module is limited. Often, when the host vehicle is relatively close to pedestrians or non-motor vehicles, a sudden perception of the risk of collision with them occurs, and it is very difficult to brake for safe avoidance at this time, so it is easy to cause collision accidents.

[0008] In summary, in view of the above problems existing in the prior art, a more comprehensive courtesy solution needs to be proposed, that is, the improved solution can maximize the courtesy to potentially occluded obstacles, ensuring the timeliness of autonomous driving vehicles while improving safety. Summary of the Invention

[0009] In view of the problems existing in the prior art, the present invention provides a method, device, equipment and storage medium for more comprehensive zebra crossing courtesy of autonomous driving vehicles.

[0010] To achieve the above object, the technical solutions adopted by the present invention are as follows:

[0011] In a first aspect, the present invention provides a method for zebra crossing courtesy of an autonomous driving vehicle, which mainly includes the following steps:

[0012] Query the nearest zebra crossing information according to the planned path and map information of the autonomous driving vehicle; obtain the minimum observation range of the zebra crossing according to the nearest zebra crossing information;

[0013] When there are no occluding obstacles within the minimum observation range of the zebra crossing, obtain the movement information of individuals or groups of pedestrians intending to pass through the nearest zebra crossing, and calculate the collision risk value between the pedestrians and the autonomous driving vehicle based on the pedestrian movement information; when the collision risk value is greater than the collision threshold, the autonomous driving vehicle decelerates or stops to give way to pedestrians;

[0014] When there are occluding obstacles within the minimum observation range of the zebra crossing, collect the movement information of the occluding obstacles;

[0015] Calculate the yielding confidence of the occluding obstacle based on the obstacle movement information; when the yielding confidence is greater than the yielding threshold, the autonomous driving vehicle decelerates or stops to give way to pedestrians; when the yielding confidence is less than the yielding threshold, the autonomous driving vehicle follows the vehicle through.

[0016] Optionally, when the distance between the autonomous vehicle and the nearest zebra crossing is less than or equal to the observation distance threshold, the autonomous vehicle acquires pedestrian and occluding obstacle information within the minimum observation range of the zebra crossing through multiple sensors.

[0017] Optionally, both ends of the minimum observation range of the zebra crossing are the two ends of the nearest zebra crossing, and the observation range covers the nearest zebra crossing.

[0018] Optionally, the occluding obstacles are classified into dynamic obstacles and static obstacles based on the obstacle motion information and / or the image recognition result;

[0019] Calculate the yielding confidence of the occluding obstacle based on the motion information of the dynamic obstacle.

[0020] Optionally, the calculation method of the yielding confidence includes:

[0021] Based on the motion information of the dynamic obstacle, calculate the acceleration and / or jerk of the dynamic obstacle within the observation time period;

[0022] Compare the acceleration and / or jerk of the dynamic obstacle with the statistical speed change range of yielding behaviors, and calculate the yielding confidence.

[0023] Optionally, the calculation method of the yielding confidence includes:

[0024] Statistically analyze the speed change data in vehicle yielding behaviors, and obtain a speed data set based on time series arrangement;

[0025] Train a machine learning model based on the speed data set, and obtain a yielding behavior judgment model;

[0026] Substitute the motion information of the dynamic obstacle into the yielding behavior judgment model to obtain the yielding confidence.

[0027] Optionally, when the occlusion range of the occluding obstacle is greater than or equal to the occlusion threshold, the autonomous vehicle decelerates or stops and waits until the occlusion range is less than the occlusion threshold;

[0028] When the yielding confidence is less than the yielding threshold, the autonomous vehicle follows the vehicle in front.

[0029] In a second aspect, the present invention further provides a device for an autonomous vehicle to yield at a zebra crossing. The device includes a sensing module and a calculation module;

[0030] The sensing module is configured to: acquire pedestrian motion information, obstacle motion information, and the positioning information of the autonomous vehicle;

[0031] The calculation module is configured to: calculate the yielding confidence based on the information obtained by the perception module, and calculate vehicle control parameters based on the yielding confidence.

[0032] In a third aspect, the present invention further provides an electronic device, which includes: a processor, a memory, and a bus. The memory stores readable instructions executable by the processor. The processor communicates with the memory through the bus. The readable instructions are used for the processor to execute the method for a self-driving vehicle to yield at a zebra crossing as described in the first aspect.

[0033] In a fourth aspect, the present invention further provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program runs on a processor, it is used to execute the method for a self-driving vehicle to yield at a zebra crossing as described in the first aspect.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] The present invention analyzes the occluded situation to obtain the yielding confidence, and can yield to potential occluded pedestrians or non-motor vehicles. It not only has higher safety, but also can drive normally when there is no intention to yield in the case of normal driving through the occluding obstacle, which can ensure the operation efficiency to a great extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0037] Figure 1 It is a flowchart of the method in a specific embodiment of the present invention;

[0038] Figure 2 It is a schematic diagram of the observation range in a specific embodiment of the present invention;

[0039] Figure 3 It is a device diagram in a specific embodiment of the present invention;

[0040] Figure 4 It is an electronic device diagram in a specific embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0042] It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0043] It is worth noting that the methods used in the present invention are all conventional methods unless otherwise specified; the raw materials and devices used are all conventional commercially available products unless otherwise specified, and their sources are not specifically limited.

[0044] It should also be noted that in the specific embodiments of the present invention, for the convenience of understanding, the method steps are described in a certain order. However, those skilled in the art can adjust the order of the steps according to actual needs. Therefore, this cannot be used as a limiting condition.

[0045] As Figure 1 shown, this embodiment provides a method for an autonomous vehicle to yield to pedestrians at a zebra crossing, which includes the following steps:

[0046] S1. Locate the nearest zebra crossing;

[0047] All autonomous vehicles are equipped with a path planning module. Combining with a high-precision map pre-set or cached online, the corresponding planned path can be obtained under the drive of task requirements. Since there are various technical solutions in the field of path planning technology for unmanned vehicles in the prior art, they will not be elaborated here.

[0048] Query the zebra crossings passed in sequence according to the planned path and map information of the known autonomous vehicle, and select the zebra crossing with the shortest distance along the path to the current vehicle as the nearest zebra crossing. Here, it should be noted that, as mentioned above, during the selection of the zebra crossing, in strict accordance with the principle that the zebra crossing intersects the planned path, the situation of being adjacent to the path but not passing through is excluded, that is, it is necessary to first calculate whether the path intersects the zebra crossing area in the high-precision map. A more accurate method is to connect adjacent path points into line segments and determine by the method of intersection of the line segment and the polygon (the zebra crossing area is not all a regular rectangle). Thus, record the road position, endpoint coordinates, etc. of the nearest zebra crossing as the nearest zebra crossing information.

[0049] Obtain the minimum observation range of the zebra crossing according to the nearest zebra crossing information. Further, asFigure 2 As shown, in this embodiment, the minimum observation range of the zebra crossing needs to be understood as a spatial range, that is, a three-dimensional environment. The straight-line distance between the vehicle and the center line of the zebra crossing is the observation distance threshold. The two lateral ends are respectively the two ends of the zebra crossing. The overall range is a cone-like structure with the vehicle sensor as the origin connecting the edges. This can make the observation range cover the nearest zebra crossing and reduce the total amount of collected information. Further, the observation distance threshold needs to be determined by combining statistical data and experimental analysis, and it should be formulated for different scenarios. The scenarios include, but are not limited to, the vehicle turning through the zebra crossing and the vehicle going straight through the zebra crossing.

[0050] S2. Collect information on occluding obstacles;

[0051] When the system calculates the distance between the current vehicle and the nearest zebra crossing in real time based on the current vehicle position and the road position of the nearest zebra crossing, when the distance between the autonomous vehicle and the nearest zebra crossing is less than or equal to the observation distance threshold, the autonomous vehicle collects information on pedestrians and occluding obstacles within the minimum observation range of the zebra crossing through multiple sensors.

[0052] S3. Judge the behavior of occluding obstacles;

[0053] When there are occluding obstacles within the minimum observation range of the zebra crossing, collect the movement information of the occluding obstacles; A simplified scenario is as Figure 2 shown. The pedestrians include those passing on both sides, and the occluding obstacles include the social vehicles on the adjacent side.

[0054] Thus, based on the movement information of the obstacles, calculate the yielding confidence of the occluding obstacles; Optionally, in this embodiment, the occluding obstacles are divided into dynamic obstacles and static obstacles based on the combination of the movement information of the obstacles and the image recognition results. Among them, the sensor (such as lidar) collects the movement data of the occluding obstacles for a period of time. When its relative ground speed is 0, it is initially judged as a static obstacle; Further, the image sensor extracts the categories of the occluding obstacles through an image recognition algorithm, including motor vehicles, non-motor vehicles, stalls, goods, etc. The system makes a secondary judgment based on the image recognition results to confirm whether the occluding obstacle is a static obstacle, so as to avoid misjudging the vehicles waiting to pass as static obstacles near the intersection.

[0055] After excluding the static obstacles, the system calculates the movement information of the dynamic obstacles to obtain the yielding confidence of the occluding obstacles.

[0056] Optionally, a simplified calculation method is as follows: Based on the motion information of dynamic obstacles, calculate the acceleration and jerk of dynamic obstacles within the observation time period; compare the acceleration and / or jerk of dynamic obstacles with the statistical speed change range of yielding behaviors to calculate the yielding confidence. Specifically, since yielding is accompanied by a rapid speed change and is generally decelerating to yield, the yielding intention of dynamic obstacles is judged by relying on the first derivative of speed (acceleration) and the second derivative of speed (jerk). First, collect the speed change data of motor vehicles or some non-motor vehicles in common yielding behaviors. Since the speed change data is a discrete data set based on time series, the maximum values of acceleration and jerk are respectively selected from it to form the maximum acceleration set and the maximum jerk set. Then, statistically calculate the data ranges of 99.7% of the maximum acceleration and maximum jerk. Since such data satisfies the normal distribution characteristics, the probability density functions are calculated respectively through the data ranges of the maximum acceleration and maximum jerk, including the maximum acceleration probability density function and the maximum jerk probability density function. Thus, substitute the acceleration and jerk of dynamic obstacles into the above two functions respectively to obtain the maximum acceleration confidence and the maximum jerk confidence. Finally, take the average of the maximum acceleration confidence and the maximum jerk confidence as the yielding confidence.

[0057] Optionally, another calculation method is as follows: Statistically calculate the speed change data in vehicle yielding behaviors and obtain a speed data set based on time series arrangement. Train a machine learning model based on the speed data set and obtain a yielding behavior judgment model; specifically, the machine learning model can be selected as a Long-Short Term Memory (LSTM) network, which is mainly used to process sequence data. Especially in the scenario of predicting speed change behaviors, speed change is usually a time series (such as the speed change of a vehicle over a period of time). LSTM can effectively capture the long-term dependencies in the sequence and control the flow and memory of information through its designed gate structures (input gate, forget gate, and output gate), so as to better predict future behaviors based on past speed changes. Substitute the motion information of dynamic obstacles into the above yielding behavior judgment model to obtain the yielding confidence.

[0058] S4. Control yielding;

[0059] In this embodiment, when there are no occluding obstacles within the minimum observation range of the zebra crossing, the pedestrian information on the zebra crossing can be completely obtained by collecting data through sensors at this time. Then, by analyzing the speed of the pedestrian movement data (the pedestrian speed has a component in the passing direction of the zebra crossing and is accelerating or moving at a constant speed), the movement information of individuals or groups of pedestrians who intend to pass through the nearest zebra crossing is obtained, and the collision risk value between the pedestrians and the autonomous driving vehicle is calculated based on the pedestrian movement information. Specifically, the collision risk value can be judged through a machine learning model or it can be determined whether the movement paths of the person and the vehicle intersect in the future for a period of time based on the movement trend. When the collision risk value is greater than the collision threshold, the autonomous driving vehicle decelerates or stops to give way to pedestrians.

[0060] When there are dynamic obstacles with occlusion problems within the minimum observation range of the zebra crossing and the yielding confidence level corresponding to the occluding obstacles is greater than the yielding threshold, the autonomous driving vehicle decelerates or stops to give way to pedestrians; when the yielding confidence level is less than the yielding threshold, the autonomous driving vehicle follows the vehicle in front.

[0061] When the occlusion range of the dynamic obstacle is greater than or equal to the occlusion threshold, the autonomous driving vehicle decelerates and waits for the occlusion range to be less than the occlusion threshold.

[0062] The method of this embodiment can not only give way to pedestrians and the like normally perceived on the zebra crossing, but also give way to potentially occluded pedestrians in the case of occlusion. Specifically, by analyzing the behavioral intention of the occluding obstacles, when the occluding obstacles have the intention of yielding, the host vehicle also yields; when the occluding obstacles are driving normally and have no intention of yielding, the host vehicle drives normally. While ensuring safety, the running timeliness is guaranteed at the maximum speed limit. At the same time, while detecting the yielding intention of the occluding obstacles, the host vehicle also calculates the occlusion degree of the occluding obstacles. The greater the occlusion degree, the lower the speed of the host vehicle is restricted. In this way, not only dynamic obstacles can be handled, but also static obstacles can be effectively handled. Thus, the autonomous driving vehicle can show excellent safety and good timeliness in front of the zebra crossing in the cases of no occlusion, with occlusion, and different types of occluding obstacles.

[0063] This embodiment also provides a device for an autonomous driving vehicle to yield at the zebra crossing, as Figure 3 shown, which mainly includes a perception module and a calculation module.

[0064] Among them, the perception module is used to: obtain pedestrian movement information, obstacle movement information, and the positioning information of the autonomous driving vehicle;

[0065] The calculation module is used to: calculate the yielding confidence level based on the information obtained by the perception module, and calculate the vehicle control parameters based on the yielding confidence level.

[0066] This embodiment also provides an electronic device, as Figure 4As shown, it mainly includes: a processor, a memory, and a bus. The memory stores readable instructions executable by the processor. The processor communicates with the memory through the bus, and specific networking and other devices are connected through interfaces 1 - n. The readable instructions are used for the processor to execute the method of a self-driving vehicle yielding at a zebra crossing as described above.

[0067] This embodiment also provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program runs on the processor, it is used to execute the method of a self-driving vehicle yielding at a zebra crossing as described above.

[0068] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than limiting the protection scope of the present invention. Any simple modification or equivalent replacement of the technical solution of the present invention by those of ordinary skill in the art shall not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A method for autonomous driving vehicles to give way to pedestrians at zebra crossings, characterized in that: The steps include: Query the nearest zebra crossing information according to the planned path and map information of the autonomous driving vehicle; obtain the minimum observation range of the zebra crossing according to the nearest zebra crossing information; When there is an occluding obstacle within the minimum observation range of the zebra crossing, collecting obstacle motion information of the occluding obstacle; The yield confidence of the obstructing obstacle is calculated based on the obstacle motion information; when the yield confidence is greater than the yield threshold, the autonomous driving vehicle slows down or stops to yield to pedestrians; when the yield confidence is less than the yield threshold, the autonomous driving vehicle follows the vehicle and passes.

2. The method for giving way to pedestrians at zebra crossings according to claim 1, characterized in that: When the distance between the autonomous driving vehicle and the nearest zebra crossing is less than or equal to the observation distance threshold, the autonomous driving vehicle collects information on pedestrians and obstructions within the minimum observation range of the zebra crossing through multiple sensors.

3. The method for giving way to pedestrians at zebra crossings according to claim 1 or 2, characterized in that: The two ends of the minimum observation range of the zebra crossing are the two ends of the nearest zebra crossing, and the observation range covers the nearest zebra crossing.

4. The method for giving way to pedestrians at zebra crossings according to claim 1, characterized in that: The blocking obstacles are divided into dynamic obstacles and static obstacles based on the obstacle motion information and / or image recognition results; The yield confidence of the blocking obstacle is calculated based on the motion information of the dynamic obstacle.

5. The method for giving way to pedestrians at zebra crossings according to claim 4, characterized in that: The calculation method of the yield confidence includes: Calculating the acceleration and / or jerk of the dynamic obstacle during an observation period based on the motion information of the dynamic obstacle; The acceleration and / or jerk of the dynamic obstacle is compared with the speed variation range of the statistical yielding behavior to calculate the yielding confidence.

6. The method for giving way to pedestrians at zebra crossings according to claim 4, characterized in that: The calculation method of the yield confidence includes: Collect statistics on speed change data during vehicle courtesy behavior and obtain speed data sets based on time series arrangement; Training a machine learning model based on the speed data set and obtaining a courtesy behavior judgment model; The motion information of the dynamic obstacle is substituted into the courtesy behavior judgment model to obtain the yielding confidence.

7. The method for giving way to pedestrians at zebra crossings according to claim 1, characterized in that: When the occlusion range of the occlusion obstacle is greater than or equal to the occlusion threshold, the autonomous driving vehicle decelerates or stops to wait until the occlusion range is less than the occlusion threshold; When the yield confidence is less than the yield threshold, the automatic driving vehicle follows the vehicle.

8. A device for autonomous driving vehicles to give way to pedestrians at zebra crossings, characterized in that: The device includes a perception module and a calculation module; The perception module is used to obtain pedestrian movement information, obstacle movement information and positioning information of the autonomous driving vehicle; The calculation module is used to calculate the yield confidence based on the information obtained by the perception module, and calculate the vehicle control parameters based on the yield confidence.

9. An electronic device, characterized in that: The electronic device includes: a processor, a memory and a bus, the memory stores readable instructions executable by the processor, the processor and the memory communicate via the bus, and the readable instructions are used by the processor to execute the method for zebra crossing courtesy of an autonomous driving vehicle as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the computer program is used to execute the method for zebra crossing courtesy of an autonomous driving vehicle as described in any one of claims 1-7.

Images