Automatic driving vehicle takeover method and system and computer storage medium

Abstract

The invention discloses an automatic driving vehicle takeover method and system and a computer storage medium, and the method comprises the steps: 1, obtaining the position information of an automatic driving vehicle and a planned driving path, and determining a takeover detection interval in front of the automatic driving vehicle based on the position information and the planned driving path; and 2, judging whether a collision risk exists in the takeover detection interval or not based on the road condition information in front of the autonomous vehicle, if so, generating vehicle takeover prompt information, and sending the vehicle takeover prompt information to the autonomous vehicle, and if not, obtaining the position information again, and executing the step 1. According to the technical scheme, the road condition information sensed by the smart city is combined with the planned driving path of the vehicle, so that the reliability of switching the vehicle from the automatic driving mode to the manual driving mode in a complex road environment is improved, and the driving safety of the vehicle is ensured.

Description

Technical Field

[0001] The present application relates to the technical field of autonomous driving, and more specifically, to an autonomous driving vehicle takeover method, an autonomous driving vehicle takeover system, and a computer storage medium. Background Art

[0002] With the continuous development of vehicle autonomous driving technology, the vehicle's autonomous driving mode is gradually replacing manual driving on closed roads or roads with relatively simple scenes. However, it is inevitable that in some cases with more complex road conditions, the vehicle still needs to be switched from autonomous driving mode to manual driving, and the driver will take over the driving vehicle.

[0003] In the prior art, the decision on whether to switch the driving mode is usually based on the vehicle itself (such as vehicle deceleration) and the road environment (such as curves and lane changes). This switching method still has obvious shortcomings when dealing with emergencies on the road, such as sudden braking and stopping, and even the phenomenon of failure to switch normally, leading to traffic accidents. Summary of the invention

[0004] The purpose of this application is to improve the reliability of switching the vehicle from automatic driving mode to manual driving mode in complex road environments and ensure vehicle driving safety.

[0005] The technical solution of the first aspect of the present application is: a method for taking over an autonomous driving vehicle is provided, the method comprising: step 1: obtaining the position information and planned driving path of the autonomous driving vehicle, and determining the takeover detection interval in front of the autonomous driving vehicle based on the position information and the planned driving path; step 2: based on the road condition information in front of the autonomous driving vehicle, determining whether there is a collision risk in the takeover detection interval, if so, generating vehicle takeover prompt information, and sending the vehicle takeover prompt information to the autonomous driving vehicle, if not, re-obtaining the position information and executing step 1; wherein the road condition information is obtained by the cloud server through environmental perception devices deployed on both sides of the road.

[0006] In any of the above technical solutions, further, step 2 also includes: step 201: obtaining a disparity map within the takeover detection interval, and based on the disparity value, performing target segmentation on the disparity map received at any time; step 202: dividing the target segmentation result of the disparity map based on a preset screening size to obtain a segmentation target, wherein the segmentation target includes at least one of a first segmentation target and a second segmentation target, and the size of the second segmentation target is smaller than that of the first segmentation target; step 203: based on the disparity value, respectively calculating the spatial position of the segmentation target in the world coordinate system; step 204: fitting the motion trajectory of each segmentation target based on the spatial position in time sequence, and recording the motion trajectory as road condition information.

[0007] In any of the above technical solutions, further, the environmental perception device is a binocular camera, the disparity map is generated by the binocular camera, the binocular camera is installed on a smart lamp pole, and multiple smart lamp poles are arranged in sequence on both sides of the road. In step 2, it also includes: step 211: labeling the disparity maps generated by multiple binocular cameras received at the same time, wherein the labels are determined by the numbers of the smart lamp poles installed with the binocular cameras, and the numbers increase successively along the driving direction of the vehicle, and the smart lamp poles are located in the takeover detection interval; step 212: based on the disparity value, target segmentation is performed on the labeled disparity map to generate a third segmentation target, and the spatial position and target frame size of each segmented third segmentation target in the world coordinate system are calculated; step 213: when it is determined that the difference between the spatial positions of any two third segmentation targets in two frames of disparity maps with adjacent labels is less than or equal to the distance threshold, and the difference between the target frame sizes is less than or equal to the size threshold, the third segmentation target in the disparity map with a large label is filtered out, and the target segmentation result of the disparity map is updated.

[0008] In any of the above technical solutions, further, the segmented target includes a first segmented target, and the collision risk includes at least an abnormal behavior collision risk. In step 2, it specifically includes: based on the motion trajectory, calculating the motion state information of the first segmented target in the takeover detection interval, wherein the motion state information includes a trajectory deviation angle and a predicted acceleration; when the trajectory deviation angle is greater than an angle threshold, or the absolute value of the predicted acceleration is greater than an acceleration threshold, marking the first segmented target as an abnormal behavior target, and determining that there is an abnormal behavior collision risk.

[0009] In any of the above technical solutions, further, the segmented target includes a second segmented target, and the collision risk includes at least a spilled object collision risk. In step 2, it specifically includes: based on the disparity map, fitting the ground plane within the takeover detection interval; when the motion trajectory of the second segmented target and the fitted ground plane have an intersection, or, according to a preset time interval, when the distance between the second segmented target and the ground plane decreases and is less than a preset distance threshold, marking the second segmented target as a spilled object, and determining that there is a spilled object collision risk.

[0010] In any of the above technical solutions, further, the collision risk at least includes: road congestion collision risk. In step 2, it specifically includes: when the motion trajectory is a stationary trajectory and the number of segmented targets is greater than the safe number, the segmented targets are marked as congested targets, and it is determined that there is a road congestion collision risk.

[0011] In any of the above technical solutions, further, in step 1, it specifically includes: determining the nth environmental perception device in front of the autonomous driving vehicle based on the position information of the autonomous driving vehicle; starting from the nth environmental perception device, selecting a preset number of environmental perception devices, and recording the perception range of the selected environmental perception device as the takeover detection interval.

[0012] The technical solution of the second aspect of the present application is: a system for taking over an autonomous driving vehicle is provided, the system comprising: a determination unit, the determination unit is configured to obtain the position information and planned driving path of the autonomous driving vehicle, and determine the takeover detection interval in front of the autonomous driving vehicle based on the position information and the planned driving path; a judgment unit, the judgment unit is configured to judge whether there is a collision risk in the takeover detection interval based on the road condition information in front of the autonomous driving vehicle, if so, generate vehicle takeover prompt information, and send the vehicle takeover prompt information to the autonomous driving vehicle, if not, re-acquire the position information.

[0013] In any of the above technical solutions, further, the judgment unit is also configured to: obtain a disparity map within the takeover detection interval, and based on the disparity value, perform target segmentation on the disparity map received at any time; based on a preset screening size, divide the target segmentation result of the disparity map to obtain the segmentation target, wherein the segmentation target includes at least one of a first segmentation target and a second segmentation target, and the size of the second segmentation target is smaller than that of the first segmentation target; based on the disparity value, respectively calculate the spatial position of the segmentation target in the world coordinate system; according to the time sequence and based on the spatial position, respectively fit the motion trajectory of each segmentation target, and record the motion trajectory as road condition information.

[0014] The technical solution of the third aspect of the present application is: a computer storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, an autonomous driving vehicle takeover method as described in any one of the technical solutions of the first aspect of the present application is implemented.

[0015] The beneficial effects of this application are:

[0016] The technical solution in this application combines the road condition information perceived by the smart city with the planned driving path of the vehicle. When the road condition within the takeover detection interval on the path is determined to be simple / uncomplicated, it is considered that there is no collision risk and the automatic driving mode can be maintained; otherwise, it is considered that there is a collision risk, and a vehicle takeover prompt message is generated and sent to prompt the driver to take over the vehicle, so as to improve the reliability of the vehicle switching from automatic driving mode to manual driving mode in complex road environments and ensure vehicle driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The advantages of the above and / or additional aspects of the present application will become apparent and easily understood in the description of the embodiments in conjunction with the following drawings, in which:

[0018] Figure 1 is a schematic flow chart of an autonomous driving vehicle takeover method according to an embodiment of the present application;

[0019] Figure 2 is a schematic diagram of a field of view overlap region according to an embodiment of the present application;

[0020] Figure 3 It is a schematic block diagram of an autonomous driving vehicle takeover system according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0022] In the following description, many specific details are elaborated to facilitate a full understanding of the present application. However, the present application may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present application is not limited to the specific embodiments disclosed below.

[0023] With the continuous development of vehicle autonomous driving technology, the vehicle's autonomous driving function is gradually improving, but in some special scenarios, especially in the complex road environment of urban roads, it is necessary to switch between the autonomous driving mode and the manual driving model, and the driver will take over the autonomous driving vehicle. Therefore, the autonomous driving vehicle needs to be able to accurately perceive the surrounding scenes.

[0024] With the rise of smart cities, the process of urban scene reconstruction or real-time display also requires the perception of various scene environments in the city. Therefore, with the help of the Internet of Things and "edge cloud" technology, the computing power of cloud servers can be fully utilized to combine the environmental perception of smart cities with the mode switching of autonomous driving vehicles. When the cloud server in the smart city senses that the vehicle is entering a section with complex road conditions, it sends a prompt message to the autonomous driving vehicle, prompting the driver to take over the vehicle and switch the autonomous driving mode to manual driving mode to ensure the safety of the vehicle.

[0025] Embodiment 1:

[0026] like Figure 1 As shown, this embodiment provides a method for taking over an autonomous driving vehicle, which is applicable to a cloud server and includes:

[0027] Step 1: Obtain the location information and planned driving path of the autonomous driving vehicle, and determine the takeover detection interval in front of the autonomous driving vehicle based on the location information and the planned driving path;

[0028] Step 2: Based on the road condition information in front of the autonomous driving vehicle, determine whether there is a collision risk within the takeover detection interval.

[0029] If yes, generate a vehicle takeover prompt message and send the vehicle takeover prompt message to the autonomous driving vehicle.

[0030] If not, re-acquire the location information and proceed to step 1.

[0031] Specifically, the smart lamp poles on both sides of the city roads can be used as structures for installing environmental sensing devices, and binocular cameras can be selected as environmental sensing devices, that is, binocular cameras are installed on the smart lamp poles, and the binocular cameras provide images within the field of view (which can be either the left image or the right image) and the corresponding disparity map.

[0032] Through the above images, image recognition can be performed based on the network model deployed on the cloud server to obtain the category of each target. Through the above disparity map, the disparity algorithm can be used to calculate the distance between each target and the binocular camera, and then calculate its spatial coordinates and corresponding size in the binocular camera coordinate system, and then obtain the spatial position of each target in the world coordinate system through coordinate transformation, realize scene perception, and obtain the category, size, spatial position, etc. of each target.

[0033] For vehicles traveling on urban roads, they can upload information such as vehicle location and planned driving route (i.e., navigation route to the destination) to the cloud server through the network. The cloud server can use the results of urban scene perception as road condition information to determine whether there are complex road conditions in the takeover detection interval on the planned driving route ahead of the vehicle.

[0034] If the road conditions are complicated, it means there is a risk of collision. The cloud server will send the generated vehicle takeover prompt information to the autonomous driving vehicle, prompting the driver to take over the vehicle and switch the vehicle's autonomous driving mode to manual driving mode. Otherwise, there is no risk of collision and the vehicle can still maintain autonomous driving mode.

[0035] It should be noted that the above-mentioned environmental perception device can also be a combination of a camera and a lidar. In this case, the camera takes images for identification, and the lidar obtains the position of each target. The specific process will not be repeated here.

[0036] It should be noted that the takeover detection interval can be a road interval within a certain range (such as 100-300 meters) in front of the vehicle on the planned driving path, and the range can be set according to actual conditions and the vehicle's automatic driving function.

[0037] Through the technical solution in this embodiment, the road condition information perceived by the smart city is combined with the planned driving path of the vehicle. When the road condition within the takeover detection interval on the path is determined to be simple / uncomplicated, it is considered that there is no collision risk and the automatic driving mode can be maintained; otherwise, it is considered that there is a collision risk, and a vehicle takeover prompt message is generated and sent to prompt the driver to take over the vehicle, so as to improve the reliability of the vehicle switching from the automatic driving mode to the manual driving mode in complex road environments and ensure the safety of vehicle driving.

[0038] In any of the above embodiments, further, considering the time delay of mode switching and the fact that the autonomous driving vehicle itself will have a certain computing power, and the road conditions in the city are not static, therefore, a section of road within a certain interval in front of the vehicle can be selected as the takeover detection interval. In the sections within this interval, the cloud server will make the judgment on vehicle takeover; in the sections between this interval and the vehicle, the vehicle's own autonomous driving function will control the vehicle, such as deceleration, braking, changing lanes, etc., to fully ensure the safety of the vehicle. In the above step 1, it specifically includes:

[0039] According to the position information of the autonomous driving vehicle, determine the nth environmental sensing device in front of the autonomous driving vehicle; wherein n = 2, 3, 4, and its value is determined by time and weather;

[0040] Starting from the nth environmental sensing device, a preset number of environmental sensing devices are selected, and the sensing range of the selected environmental sensing devices is recorded as the takeover detection interval.

[0041] Specifically, considering that the distance between lamp poles in the city is usually about 30 meters, and the human eye can usually see objects about 50 meters away, the nth environmental sensing device in front of the vehicle is used as the dividing line, and the sensing range of multiple devices after the nth environmental sensing device is used as the takeover detection interval.

[0042] For example, assuming that the preset number is 4, in time periods with good lighting, the value of n can be set to 4, that is, the perception range of the binocular cameras in the 4th to 7th smart light poles is used as the takeover detection interval; in time periods with poor lighting, such as at night or in foggy weather, the value of n can be set to 2, that is, the perception range of the binocular cameras in the 2nd to 5th smart light poles is used as the takeover detection interval.

[0043] In any of the above embodiments, further, step 2 further includes:

[0044] Step 201: Obtain a disparity map within the takeover detection interval, and perform target segmentation on the disparity map received at any time based on the disparity value;

[0045] Step 202: based on a preset screening size, dividing the target segmentation result of the disparity map to obtain a segmentation target, wherein the segmentation target includes at least one of a first segmentation target and a second segmentation target, and the size of the second segmentation target is smaller than the first segmentation target;

[0046] Step 203: Based on the disparity value, respectively calculate the spatial position of the segmented target in the world coordinate system;

[0047] Step 204: Fit the motion trajectory of each segmented target in time sequence and based on the spatial position, and record the motion trajectory as road condition information.

[0048] Specifically, a binocular camera on any smart lamp pole is used as an example for explanation. A person skilled in the art can understand that for a dense disparity map, since the image (left image) and the disparity Figure 1 One-to-one correspondence means that any target in the image has a corresponding area in the disparity map, and the disparity values ​​of the disparity points in the area are the same and the disparity points are together. For sparse disparity maps, even if there may be disparity holes, the corresponding disparity clustering area can still be obtained by combining the corresponding disparity clustering algorithm, and the area also corresponds to one target in the image (left image).

[0049] Therefore, in order to improve the efficiency of data processing, this embodiment directly processes the disparity map after obtaining it, and performs target segmentation based on the disparity value. The categories of the obtained targets may be: trees, buildings, vehicles, pedestrians, signs, triangular cones, etc.

[0050] Considering that the factors that affect the normal driving of the vehicle are usually other vehicles (mainly), pedestrians and other targets, the target segmentation results of the disparity map can be filtered based on the preset filtering size to filter out vehicles and pedestrians, while trees, buildings, etc. can be filtered out.

[0051] It should be noted that the screened targets can be further divided according to categories so that different coefficients can be assigned when fitting the motion trajectory. For example, for vehicles and pedestrians, the vehicle can be used as the first segmentation target and the pedestrian can be used as the second segmentation target, and different coefficients can be assigned.

[0052] For any segmented target, there is a region that can be used as the target frame to obtain the image coordinates of the segmented target. Combined with the disparity value of the region and the internal and external parameters of the corresponding binocular camera, the spatial position and size of the region in the world coordinate system can be calculated through binocular camera coordinate calculation and coordinate transformation.

[0053] Afterwards, the target can be tracked in the multi-frame disparity map according to the size of the segmented target, so as to obtain the spatial position (x w ,y w ,z w ), forming an array including spatial position and time. Then, based on this array, the motion trajectory of the segmented target can be obtained by fitting.

[0054] It should be noted that in the fitting process, the polynomial fitting method can be selected and the polynomial is set as:

[0055] y=a n x n +a n-1 x n-1 +...+a1x+a0

[0056] Where y is the spatial position of any segmentation target in each frame disparity map, x is the time corresponding to the disparity map of the frame, and a n to a0 is the fitting coefficient, wherein, in the process of solving the fitting coefficient, a method such as least squares iterative optimization can be adopted.

[0057] Therefore, after obtaining the above fitting coefficients, the motion trajectory of the corresponding segmented target can be obtained, and the position prediction can be performed and used as the road condition information.

[0058] If the motion trajectory overlaps, intersects, or is close to the vehicle's planned driving path within the takeover detection interval, it indicates that there is a risk of collision and the driver needs to take over the vehicle.

[0059] In any of the above embodiments, further, the environment perception device is a binocular camera, the disparity map is generated by the binocular camera, the binocular camera is installed on a smart light pole, and a plurality of smart light poles are arranged in sequence on both sides of the road, and step 2 also includes:

[0060] Step 211: labeling the disparity maps generated by multiple binocular cameras received at the same time, wherein the label is determined by the number of the smart light pole on which the binocular camera is installed, and the number increases in sequence along the driving direction of the vehicle, and the smart light pole is located in the takeover detection interval;

[0061] Step 212: based on the disparity value, perform target segmentation on the labeled disparity map to generate third segmentation targets, and calculate the spatial position and target frame size of each segmented third segmentation target in the world coordinate system;

[0062] Step 213: When it is determined that the difference between the spatial positions of any two third segmentation targets in the disparity maps of two frames with adjacent labels is less than or equal to the distance threshold, and the difference between the target frame sizes is less than or equal to the size threshold, the third segmentation targets in the disparity map with a larger label are filtered out, and the target segmentation result of the disparity map is updated.

[0063] Specifically, considering that there are usually multiple smart lamp poles on both sides of the road, and the fields of view of the binocular cameras in each smart lamp pole may overlap, the same target may appear in the fields of view of two adjacent binocular cameras. Therefore, the disparity maps generated by multiple binocular cameras received at the same time can be numbered, where the number is determined by the number of the smart lamp pole on which the binocular camera is installed, and the number can be set to increase sequentially along the direction of vehicle travel.

[0064] like Figure 2 As shown, in simple terms, the disparity maps collected by each binocular camera are numbered in sequence according to the number of the smart light pole. For example, the disparity map obtained by the i-th smart light pole (binocular camera) at time t is numbered i t , and the parallax icon number obtained by the i+1th smart light pole is (i+1) t ; Correspondingly, at time t+1, the labels of the two disparity images are i and t+1 、(i+1) t+1 .

[0065] The binocular cameras of two adjacent smart lamp poles will have an overlapping field of view. When there is a target in this area, it will be presented in two disparity maps respectively, such as "Target 2".

[0066] At this time, if the motion trajectory fitting is performed directly, the trajectory fitting result will be affected, so one of the "targets 2" needs to be filtered out. The calculation process of the spatial position of "target 2" will not be repeated.

[0067] Therefore, when the difference (such as the Euclidean distance) between the two corresponding spatial positions of the target 2 and the target 2' is less than or equal to the distance threshold, it can be considered that the target 2 and the target 2' should be the same target.

[0068] Considering that in the parallax calculation process, “the accuracy of near parallax is higher than that of far parallax”, when the field of view of the binocular camera is tilted forward, the i-th t The disparity value of "target 2" in the frame disparity map is greater than that of the (i+1)th tTherefore, filter out the (i+1)th t The third segmentation target (target 2') and the corresponding spatial position in the frame disparity map, retaining the i-th t The third segmentation target (target 2) and the corresponding spatial position in the frame disparity map are used to update the target segmentation result of the disparity map.

[0069] In any of the above embodiments, further, the segmentation target includes a first segmentation target, the collision risk includes at least an abnormal behavior collision risk, and step 2 specifically includes:

[0070] Based on the motion trajectory, calculating the motion state information of the first segmented target in the takeover detection interval, wherein the motion state information includes the trajectory deviation angle and the predicted acceleration;

[0071] When the trajectory deviation angle is greater than the angle threshold, or

[0072] When the absolute value of the predicted acceleration is greater than the acceleration threshold,

[0073] The first segmented target is marked as an abnormal behavior target, and it is determined that there is an abnormal behavior collision risk.

[0074] Specifically, on urban roads, there will be abnormal vehicle driving, such as sudden braking, sudden stopping, illegal lane changing, and other sudden behaviors, and these sudden behaviors will affect the driving safety of normal vehicles.

[0075] The following is an example of a vehicle (the main factor affecting the normal driving of a vehicle).

[0076] When a vehicle changes lanes illegally, its driving trajectory will deviate significantly; similarly, when a vehicle brakes suddenly or accelerates suddenly, the spacing between its spatial positions collected at the same time interval will change significantly, and this information will be reflected in its motion trajectory.

[0077] Therefore, after obtaining the motion trajectory, its motion state information, such as trajectory deviation angle and predicted acceleration, can be calculated. The obtained motion state is compared with the set corresponding threshold to determine whether it is an abnormal behavior target.

[0078] For example, when the vehicle's trajectory deviation angle is greater than an angle threshold, it can be considered that the vehicle has violated the lane change rules; when the absolute value of the predicted acceleration calculated from the motion trajectory is greater than an acceleration threshold, it can be considered that the vehicle has suddenly braked or accelerated.

[0079] Therefore, when there is an abnormal behavior target within the takeover detection interval, it can be determined that there is an abnormal behavior collision risk.

[0080] In any of the above embodiments, further, considering that there may be objects spilled by freight vehicles on urban roads, and such spilled objects are usually non-standard obstacles after falling on the road, such as cartons, tarpaulins, etc., at this time, the network model may not be able to accurately identify them, and forcibly divide them into road surfaces or identify them as vehicles, pedestrians, cones, etc., which will affect the control of the vehicle by the vehicle's automatic driving mode, and a mode switch is required for the driver to take over the vehicle.

[0081] At the same time, considering that the size and movement of the spilled objects are different from those of the vehicle targets and the frequency of occurrence is low, they can be classified as the second segmentation target.

[0082] In this embodiment, the collision risk at least includes: the collision risk of spilled objects, and the above step 2 specifically includes:

[0083] Based on the disparity map, fit the ground plane within the detection interval;

[0084] When the motion trajectory of the second segmented target intersects with the fitted ground plane, or, based on a preset time interval, when the distance between the second segmented target and the ground plane decreases, and from a certain moment on, the distance is less than a preset distance threshold, the second segmented target is marked as a spilled object, and it is determined that there is a risk of spilled object collision.

[0085] Specifically, by analyzing the ground area in multiple disparity maps within the takeover detection interval, a preset plane fitting algorithm is used to establish a ground model, such as the RANSAC (RANdom SAmple Consensus) algorithm, the data of the ground area in the disparity map is processed, and the ground plane equation Ax+By+Cz+D=0 is established.

[0086] The RANSAC algorithm randomly selects points in the ground area and substitutes them into the plane equations x, y, and z. When the number of random sampling is large enough, the coefficients A, B, C, and D in the plane equation can be analyzed to obtain the fitted ground plane.

[0087] Considering that the spilled object will eventually fall to the ground, that is, there is an intersection between its motion trajectory and the ground plane, and some rigid objects may bounce after falling to the ground, so its trajectory may have multiple intersections with the ground. Therefore, the motion trajectory equation of the target can be substituted into the ground plane equation to find out whether there is an intersection. If there is an intersection, it can be considered that the target is in contact with the ground, and the second segmented target is determined to be a spilled object.

[0088] However, some objects (such as the tarpaulin on the truck) do not fall directly when they fall. Therefore, the distance between them and the ground can also be used as a reference for judging whether they are spilled objects. That is, when the distance between the two decreases and, starting from a certain moment, the distance is less than a preset distance threshold, it is considered that the target is in the process of falling and is about to come into contact with the ground, and the corresponding second segmented target is determined to be a spilled object. The three-dimensional spatial position (x w ,y w ,z w ) to the ground plane equation to calculate the distance from a point to a plane.

[0089] In any of the above embodiments, further, the collision risk at least includes: road congestion collision risk, and step 2 specifically includes:

[0090] When the motion trajectory is a stationary trajectory and the number of segmented targets is greater than the safe number, the segmented targets are marked as congested targets and it is determined that there is a risk of road congestion collision.

[0091] Embodiment 2:

[0092] like Figure 3 As shown, this embodiment provides an autonomous driving vehicle takeover system 100, which is deployed on a cloud server and includes:

[0093] A determination unit 10, the determination unit 10 is configured to obtain the position information and the planned driving path of the autonomous driving vehicle, and determine the takeover detection interval in front of the autonomous driving vehicle based on the position information and the planned driving path;

[0094] The judgment unit 20 is configured to judge whether there is a collision risk in the takeover detection interval based on the road condition information in front of the autonomous driving vehicle,

[0095] If yes, generate a vehicle takeover prompt message and send the vehicle takeover prompt message to the autonomous driving vehicle.

[0096] If not, re-acquire the location information.

[0097] In any of the above embodiments, further, the judging unit 20 is further configured to:

[0098] Obtain the disparity map within the takeover detection interval, and perform target segmentation on the disparity map received at any time based on the disparity value;

[0099] Based on the preset screening size, the target segmentation result of the disparity map is divided to obtain the segmentation target.

[0100] The segmentation target includes at least one of a first segmentation target and a second segmentation target, and the size of the second segmentation target is smaller than the first segmentation target;

[0101] Based on the disparity value, the spatial position of the segmented target in the world coordinate system is calculated respectively;

[0102] The motion trajectory of each segmented target is fitted in time sequence and based on the spatial position, and the motion trajectory is recorded as the road condition information.

[0103] In any of the above embodiments, further, the environment perception device is a binocular camera, the disparity map is generated by the binocular camera, the binocular camera is installed on a smart light pole, and a plurality of smart light poles are arranged in sequence on both sides of the road, and the judgment unit 20 is further configured as follows:

[0104] Label the disparity maps generated by multiple binocular cameras received at the same time.

[0105] Among them, the number is determined by the number of the smart light pole equipped with a binocular camera, the number increases in sequence along the direction of vehicle travel, and the smart light pole is located in the takeover detection interval;

[0106] Based on the disparity value, the labeled disparity map is segmented to generate a third segmentation target, and the spatial position and target frame size of each segmented third segmentation target in the world coordinate system are calculated;

[0107] When the difference between the spatial positions of any two third segmentation targets in two disparity maps with adjacent labels is determined to be less than or equal to the distance threshold, and the difference between the target frame sizes is less than or equal to the size threshold, the third segmentation target in the disparity map with a larger label is filtered out, and the target segmentation result of the disparity map is updated.

[0108] In any of the above embodiments, further, the segmentation target includes a first segmentation target, the collision risk includes at least an abnormal behavior collision risk, and the judgment unit 20 is further configured to:

[0109] Based on the motion trajectory, calculating the motion state information of the first segmented target in the takeover detection interval, wherein the motion state information includes the trajectory deviation angle and the predicted acceleration;

[0110] When the trajectory deviation angle is greater than the angle threshold, or

[0111] When the absolute value of the predicted acceleration is greater than the acceleration threshold,

[0112] The first segmented target is marked as an abnormal behavior target, and it is determined that there is an abnormal behavior collision risk.

[0113] In any of the above embodiments, further, the segmentation target includes a second segmentation target, the collision risk includes at least a spilled object collision risk, and the judgment unit 20 is further configured to:

[0114] Based on the disparity map, fit the ground plane within the detection interval;

[0115] When the motion trajectory of the second segmented target intersects with the fitted ground plane, or

[0116] According to the preset time interval, when the distance between the second segmentation target and the ground plane decreases and is less than the preset distance threshold,

[0117] The second segmented target is marked as a spilled object, and it is determined that there is a spilled object collision risk.

[0118] In any of the above embodiments, further, the collision risk includes at least: road congestion collision risk, and the judgment unit 20 is also configured to: when the motion trajectory is a stationary trajectory and the number of segmented targets is greater than the safe number, mark the segmented targets as congested targets and determine that there is a road congestion collision risk.

[0119] In any of the above embodiments, further, the determination unit 10 is further configured to: determine an nth environment perception device in front of the autonomous driving vehicle according to the position information of the autonomous driving vehicle;

[0120] Starting from the nth environmental sensing device, a preset number of environmental sensing devices are selected, and the sensing range of the selected environmental sensing devices is recorded as the takeover detection interval.

[0121] Embodiment 3:

[0122] This embodiment further provides a computer storage medium, in which a computer program is stored. When the stored computer program is executed by a processor, the autonomous driving vehicle takeover method of any one of the above-mentioned embodiments 1 is implemented.

[0123] Specifically, the computer program includes computer instructions, which are stored in a computer storage medium; a processor of a computer device reads the computer instructions from the computer storage medium, and the processor executes the computer instructions, so that the computer device executes the method steps in any one of the above-mentioned embodiments 1.

[0124] So far, the various embodiments of the present application have been described in detail. In order to avoid obscuring the concept of the present application, some details known in the art are not described. Based on the above description, those skilled in the art can fully understand how to implement the technical solution disclosed herein.

[0125] Although some specific embodiments of the present application have been described in detail through examples, those skilled in the art should understand that the above examples are only for illustration rather than for limiting the scope of the present application.

[0126] The steps in this application can be adjusted in order, combined, and deleted according to actual needs.

[0127] Although the present application is disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and are not intended to limit the application of the present application. The scope of protection of the present application is defined by the appended claims and may include various modifications, alterations and equivalents made to the invention without departing from the scope and spirit of the present application.

Claims

1. A method for taking over an autonomous driving vehicle, characterized in that: The method is applicable to a cloud server, and the method includes: Step 1: Obtaining the location information and the planned driving path of the autonomous driving vehicle, and determining the takeover detection interval in front of the autonomous driving vehicle based on the location information and the planned driving path; Step 2: Based on the road condition information in front of the autonomous driving vehicle, determine whether there is a collision risk in the takeover detection interval. If yes, generate vehicle takeover prompt information, and send the vehicle takeover prompt information to the autonomous driving vehicle, If not, reacquire the location information and execute step 1; The road condition information is obtained by the cloud server through environmental sensing devices deployed on both sides of the road.

2. The method for taking over an autonomous driving vehicle according to claim 1, characterized in that: The step 2 further includes: Step 201: obtaining a disparity map within the takeover detection interval, and performing target segmentation on the disparity map received at any moment based on the disparity value; Step 202: based on a preset screening size, the target segmentation result of the disparity map is divided to obtain a segmentation target. The segmentation target includes at least one of a first segmentation target and a second segmentation target, and the size of the second segmentation target is smaller than the first segmentation target; Step 203: Based on the disparity values, respectively calculating the spatial positions of the segmented targets in the world coordinate system; Step 204: fitting the motion trajectory of each of the segmented targets in time sequence based on the spatial position, and recording the motion trajectory as the road condition information.

3. The method for taking over an automatic driving vehicle according to claim 2, characterized in that: The environment perception device is a binocular camera, the disparity map is generated by the binocular camera, the binocular camera is installed on a smart light pole, and a plurality of the smart light poles are sequentially arranged on both sides of the road. The step 2 also includes: Step 211: labeling the disparity maps generated by the multiple binocular cameras received at the same time. The number is determined by the number of the smart lamp pole on which the binocular camera is installed, the number increases in sequence along the direction of vehicle travel, and the smart lamp pole is located within the takeover detection interval; Step 212: based on the disparity value, perform target segmentation on the labeled disparity map to generate third segmentation targets, and calculate the spatial position and target frame size of each segmented third segmentation target in the world coordinate system; Step 213: When it is determined that the difference between the spatial positions of any two third segmentation targets in the disparity maps of two frames with adjacent labels is less than or equal to the distance threshold, and the difference between the target frame sizes is less than or equal to the size threshold, filter out the third segmentation target in the disparity map with a larger label, and update the target segmentation result of the disparity map.

4. The method for taking over an automatic driving vehicle according to claim 2 or 3, characterized in that: The segmentation target includes the first segmentation target, the collision risk includes at least an abnormal behavior collision risk, and the step 2 specifically includes: Based on the motion trajectory, calculating the motion state information of the first segmented target in the takeover detection interval, wherein the motion state information includes a trajectory deviation angle and a predicted acceleration; When the trajectory deviation angle is greater than the angle threshold, or When the absolute value of the predicted acceleration is greater than the acceleration threshold, The first segmented target is marked as an abnormal behavior target, and it is determined that there is an abnormal behavior collision risk.

5. The method for taking over an automatic driving vehicle according to claim 2 or 3, characterized in that: The segmentation target includes the second segmentation target, the collision risk includes at least the spilled object collision risk, and the step 2 specifically includes: Fitting a ground plane within the takeover detection interval based on the disparity map; When the motion trajectory of the second segmented target intersects the fitting ground plane, or According to a preset time interval, when the distance between the second segmentation target and the ground plane decreases and is less than a preset distance threshold, The second segmented target is marked as a spilled object, and it is determined that there is a spilled object collision risk.

6. The method for taking over an automatic driving vehicle according to claim 2 or 3, characterized in that: The collision risk at least includes: road congestion collision risk, and the step 2 specifically includes: When the motion trajectory is a stationary trajectory and the number of the segmented targets is greater than the safe number, the segmented targets are marked as congested targets, and it is determined that there is a risk of road congestion collision.

7. The method for taking over an automatic driving vehicle according to any one of claims 1 to 3, characterized in that: The step 1 specifically includes: Determining an nth environmental perception device in front of the autonomous driving vehicle according to the position information of the autonomous driving vehicle; Starting from the nth environmental sensing device, a preset number of environmental sensing devices are selected, and the sensing range of the selected environmental sensing devices is recorded as the takeover detection interval.

8. An automatic driving vehicle takeover system, characterized in that: The system is deployed on a cloud server and includes: a determination unit, the determination unit being configured to obtain position information and a planned driving path of the autonomous driving vehicle, and determine a takeover detection interval ahead of the autonomous driving vehicle based on the position information and the planned driving path; a judgment unit, the judgment unit being configured to judge whether there is a collision risk in the takeover detection interval based on road condition information in front of the autonomous driving vehicle, If yes, generate vehicle takeover prompt information, and send the vehicle takeover prompt information to the autonomous driving vehicle, If not, re-acquire the location information; The road condition information is obtained by the cloud server through environmental sensing devices deployed on both sides of the road.

9. The automatic driving vehicle takeover system according to claim 8, characterized in that: The judging unit is further configured to: Obtaining a disparity map within the takeover detection interval, and performing target segmentation on the disparity map received at any time based on the disparity value; Based on a preset screening size, the target segmentation result of the disparity map is divided to obtain a segmentation target, The segmentation target includes at least one of a first segmentation target and a second segmentation target, and the size of the second segmentation target is smaller than the first segmentation target; Based on the disparity values, respectively calculating the spatial positions of the segmented targets in the world coordinate system; The motion trajectories of the segmented targets are respectively fitted in time sequence based on the spatial position, and the motion trajectories are recorded as the road condition information.

10. A computer storage medium, characterized in that The computer storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images