Information acquisition method, device, computer equipment and storage medium

Abstract

The present invention relates to an information acquisition method, apparatus, computer equipment, and storage medium. The method comprises: obtaining a segmentation result of a target environment image captured by a camera of a target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling; determining whether a target pitch angle associated with the target environment image exists based on the current position of the camera, the segmentation result, and the detected three-dimensional lane line point cloud; and when the target pitch angle exists, transforming the position of a cone barrel in the target environment image in the image coordinate system of the target environment image into a corrected position of the cone barrel in the vehicle body coordinate system using the target corrected position.

Description

Technical Field

[0001] The present invention relates to the field of vehicle technology, and in particular to an information acquisition method, device, computer equipment, and storage medium. Background Art

[0002] Obtaining the location of cones is a key factor that the vehicle's autonomous driving system must consider when making driving decisions in scenarios such as road construction diversions and temporary road closures.

[0003] Related technologies rely on vehicle-mounted LiDAR to obtain information such as laser reflection intensity and laser point cloud density for determining the location of cones. This information is then used to predict the location of the cones. When the vehicle is far from the cones, the laser reflection intensity and laser point cloud density detected by the vehicle are low, resulting in a low amount of information captured by the vehicle for determining the location of the cones, and a low accuracy in the determined location of the cones. Improving the accuracy of the detected location of the cones has become a pressing issue. Summary of the Invention

[0004] One of the purposes of the present invention is to provide an information acquisition method, apparatus, computer equipment and storage medium to solve the problem of how to improve the accuracy of the acquired position of the cone barrel.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] An information acquisition method, comprising:

[0007] Obtaining a segmentation result of a target environment image captured by a camera of the target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling, wherein the segmentation result indicates a plurality of lane line pixels of the target lane line, and the detected three-dimensional lane line point cloud is detected based on the target environment image;

[0008] Determining whether there is a target pitch angle associated with a target environment image based on the current pose of the camera, the segmentation result, and the detected three-dimensional lane line point cloud, wherein an error between a target projected three-dimensional lane line point cloud obtained by inversely projecting the plurality of lane line pixel points into a body coordinate system of a target vehicle using a target corrected pose corresponding to the target pitch angle and the detected three-dimensional lane line point cloud is less than an error threshold, and the target corrected pose is obtained by replacing a current pitch angle in the current pose of the camera with the target pitch angle;

[0009] When it is determined that the target pitch angle exists, the target corrected posture is used to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the vehicle body coordinate system.

[0010] According to the above technical means, the segmentation results of the target environment image captured by the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road on which the target vehicle is traveling are used as information involved in obtaining the position of the cone barrel. The segmentation results of the target environment image captured by the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road on which the target vehicle is traveling can indicate the pixel features such as the position and type of the pixels of the cone barrel in the environment image. When the distance between the vehicle and the cone barrel is far, the full amount of pixel features of the cone barrel can still be obtained, that is, the segmentation results of the target environment image captured by the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road on which the target vehicle is traveling can indicate the full amount of pixel features of the cone barrel. The amount of information involved in determining the position of the cone barrel is large, which improves the accuracy of the obtained position of the cone barrel and avoids the situation where the accuracy of the position of the cone barrel is not affected by the distance between the vehicle and the cone barrel.

[0011] The information acquisition method provided by the embodiment of the present invention takes into account that the actual position of the camera of the target vehicle and the current position of the camera of the target vehicle may be different when the position of the cone barrel in the target environment image in the image coordinate system of the target environment image is transformed into the position of the cone barrel in the body coordinate system of the target vehicle. Among them, the intrinsic parameters in the position of the camera are fixed, and the change of the extrinsic parameters in the position of the camera is usually caused by the change of the pitch angle. The difference in the position of the same camera at different times is usually caused by the difference in the pitch angle of the same camera at different times. Directly using the current position of the camera of the target vehicle to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the position of the cone barrel in the body coordinate system of the target vehicle will lead to a decrease in the accuracy of the position of the cone barrel.

[0012] The information acquisition method provided by the embodiment of the present invention determines whether there is a target pitch angle related to the target environment image based on the current posture of the target vehicle's camera, the segmentation result, and the detected three-dimensional lane line point cloud. This is equivalent to determining whether there is a more accurate pitch angle, namely the target pitch angle, relative to the current posture of the target vehicle's camera. When it is determined that there is a more accurate pitch angle, a more accurate target correction posture relative to the current posture of the target vehicle's camera is obtained by replacing the current pitch angle in the current posture of the target vehicle's camera with the target pitch angle. The more accurate target correction posture is used to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the body coordinate system of the target vehicle. A more accurate corrected position of the cone barrel in the body coordinate system of the target vehicle is obtained, thereby improving the accuracy of the acquired position of the cone barrel.

[0013] Furthermore, the method further includes: determining the outline of a target cone barrel according to the segmentation result of the target environment image, wherein the target cone barrel is any cone barrel in the target environment image;

[0014] The position of the intersection of the outline of the target cone barrel and the perpendicular bisector of the detection frame surrounding the target cone barrel is determined as the position of the target cone barrel in the image coordinate system of the target environment image.

[0015] Furthermore, the method further comprises:

[0016] Perform cone-bucket tracking corresponding to the target environment image, the cone-bucket tracking comprising:

[0017] Performing a tracking operation corresponding to a target environment image, and after performing the tracking operation, searching for a target particle from a target particle set, using the position of the target particle as the tracking position of the corresponding cone barrel in the target environment image, and outputting the tracked position, wherein the target particle set is initialized to an initial particle set before performing the first cone barrel tracking, and the initial particle set is obtained by generating particles according to the corresponding position of the cone barrel in the environment image used for particle initialization in the vehicle body coordinate system; the tracking operation includes:

[0018] When a first particle from the target particle set exists in the target area associated with the target cone, the weight of the first particle is updated according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle. When a second particle exists in the target particle set, the weight of the second particle is reduced, wherein the second particle is a particle that is not in the target area associated with any cone in the target environment image.

[0019] When there are no particles from the target particle set in the target area, new particles generated according to the corresponding position of the target cone barrel in the vehicle coordinate system are added to the target particle set.

[0020] Furthermore, the method further comprises:

[0021] When the target cone has a corrected position in the vehicle coordinate system, using the first weight as the weight of the target cone;

[0022] When the target cone barrel does not have a corrected position in the vehicle body coordinate system, the second weight is used as the weight of the target cone barrel, wherein the second weight is smaller than the first weight.

[0023] Furthermore, the method further comprises:

[0024] The position of the center point is the corresponding position of the target cone barrel in the vehicle body coordinate system, and the circular area with a preset first radius as the radius is determined as the target area related to the target cone barrel.

[0025] Furthermore, the target particles are found from the target particle set including:

[0026] When there are particles with weights less than 0 in the target particle set, the particles with weights less than 0 are deleted from the target particle set;

[0027] Find multiple candidate particles from the target particle set, and the weight of the candidate particles is greater than the weight threshold;

[0028] Find the target particle from multiple candidate particles.

[0029] Furthermore, finding the target particle from multiple candidate particles includes:

[0030] Sort multiple candidate particles according to their weights from large to small to obtain the order of the candidate particles;

[0031] The search operation is iteratively performed until a target number of target particles is found, where the target number is the number of cone barrels in the target environment image. The search operation includes:

[0032] determining whether there are other particles in a target area associated with the candidate particle targeted by the search operation that have not been associated with the candidate particle, wherein the candidate particle targeted by the first search operation is a particle with the largest weight among the multiple candidate particles, and the candidate particle targeted by the non-first search operation is a next candidate particle of the candidate particle targeted by the last search operation of the non-first search operation, wherein the next candidate particle of the candidate particle targeted by the last search operation is indicated by the order;

[0033] If yes, determining the candidate particle targeted by the search operation as the target particle, and associating other particles in the target area associated with the candidate particle targeted by the search operation that have not been associated with the candidate particle with the candidate particle targeted by the search operation, and determining the next candidate particle of the candidate particle targeted by the search operation as the candidate particle targeted by the next search operation of the search operation;

[0034] If not, the next candidate particle of the candidate particle targeted by the search operation is determined as the candidate particle targeted by the next search operation of the search operation, wherein the next candidate particle of the candidate particle targeted by the search operation is indicated by the order.

[0035] Furthermore, the method further comprises:

[0036] A circular area having a center point as the position of the candidate particle targeted by the search operation in the vehicle body coordinate system and having a preset second radius as a radius is determined as a target area related to the candidate particle targeted by the search operation.

[0037] An information acquisition device, comprising:

[0038] an acquisition unit, configured to acquire a segmentation result of a target environment image captured by a camera of a target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling, wherein the segmentation result indicates a plurality of lane line pixels of the target lane line, and the detected three-dimensional lane line point cloud is detected based on the target environment image;

[0039] a determining unit, configured to determine whether a target pitch angle associated with a target environment image exists based on a current pose of the camera, the segmentation result, and the detected three-dimensional lane line point cloud, wherein an error between a target projected three-dimensional lane line point cloud obtained by inversely projecting the plurality of lane line pixel points into a body coordinate system of a target vehicle using a target corrected pose corresponding to the target pitch angle and the detected three-dimensional lane line point cloud is less than an error threshold, and the target corrected pose is obtained by replacing a current pitch angle in the current pose of the camera with the target pitch angle;

[0040] A conversion unit is used to, when it is determined that the target pitch angle exists, use the target corrected posture to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the vehicle body coordinate system.

[0041] Furthermore, the information acquisition device further includes:

[0042] The cone position acquisition unit is used to determine the outline of the target cone based on the segmentation result of the target environment image, wherein the target cone is any cone in the target environment image; and the position of the intersection of the outline of the target cone and the perpendicular bisector of the detection frame surrounding the target cone is determined as the position of the target cone in the image coordinate system of the target environment image.

[0043] Furthermore, the information acquisition device further includes:

[0044] A tracking unit is configured to perform cone-bucket tracking corresponding to a target environment image, wherein the cone-bucket tracking includes:

[0045] Performing a tracking operation corresponding to a target environment image, and after performing the tracking operation, searching for a target particle from a target particle set, using the position of the target particle as the tracking position of the corresponding cone barrel in the target environment image, and outputting the tracked position, wherein the target particle set is initialized to an initial particle set before performing the first cone barrel tracking, and the initial particle set is obtained by generating particles according to the corresponding position of the cone barrel in the environment image used for particle initialization in the vehicle body coordinate system; the tracking operation includes:

[0046] When a first particle from the target particle set exists in the target area associated with the target cone, the weight of the first particle is updated according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle. When a second particle exists in the target particle set, the weight of the second particle is reduced, wherein the second particle is a particle that is not in the target area associated with any cone in the target environment image.

[0047] When there are no particles from the target particle set in the target area, new particles generated according to the corresponding position of the target cone barrel in the vehicle coordinate system are added to the target particle set.

[0048] Furthermore, the information acquisition device further includes:

[0049] A weight allocation unit is used to use a first weight as the weight of the target cone barrel when the target cone barrel has a corrected position in the vehicle body coordinate system; when the target cone barrel does not have a corrected position in the vehicle body coordinate system, use a second weight as the weight of the target cone barrel, wherein the second weight is smaller than the first weight.

[0050] Furthermore, the information acquisition device further includes:

[0051] The first target area determination unit is used to determine a circular area with a center point as the corresponding position of the target cone barrel in the vehicle body coordinate system and a preset first radius as the target area related to the target cone barrel.

[0052] Furthermore, the tracking unit is also used to delete particles with weights less than 0 from the target particle set when there are particles with weights less than 0 in the target particle set; find multiple candidate particles from the target particle set, where the weights of the candidate particles are greater than a weight threshold; and find the target particle from the multiple candidate particles.

[0053] Furthermore, the tracking unit is further used to sort multiple candidate particles from large to small according to the weights of the candidate particles to obtain the order of the candidate particles; iteratively perform the search operation until a target number of target particles are found, wherein the target number is the number of cone barrels in the target environment image, and the search operation includes: determining whether there are other particles that have not been associated with the candidate particle in the target area related to the candidate particle targeted by the search operation, wherein the candidate particle targeted by the first search operation is the particle with the largest weight among the multiple candidate particles, and the candidate particle targeted by the non-first search operation is the next candidate particle of the candidate particle targeted by the previous search operation of the non-first search operation, wherein the previous search operation is performed. The next candidate particle of the candidate particle targeted by the search operation is indicated by the sequence; if so, the candidate particle targeted by the search operation is determined as the target particle, and other particles in the target area related to the candidate particle targeted by the search operation that have not been associated with the candidate particle are associated with the candidate particle targeted by the search operation, and the next candidate particle of the candidate particle targeted by the search operation is determined as the candidate particle targeted by the next search operation of the search operation; if not, the next candidate particle of the candidate particle targeted by the search operation is determined as the candidate particle targeted by the next search operation of the search operation, wherein the next candidate particle of the candidate particle targeted by the search operation is indicated by the sequence.

[0054] Furthermore, the information acquisition device further includes:

[0055] The second target area determination unit is used to determine a circular area having a center point as the position of the candidate particle targeted by the search operation in the vehicle body coordinate system and a preset second radius as a target area related to the candidate particle targeted by the search operation.

[0056] Beneficial effects of the present invention:

[0057] The segmentation results of the target environment image captured by the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road where the target vehicle is traveling are used as information involved in obtaining the position of the cone barrel. The segmentation results of the target environment image captured by the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road where the target vehicle is traveling can indicate the pixel features such as the position and type of the pixel of the cone barrel in the environment image. When the distance between the vehicle and the cone barrel is far, the full amount of pixel features of the cone barrel can still be obtained, that is, the segmentation results of the target environment image captured by the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road where the target vehicle is traveling can indicate the full amount of pixel features of the cone barrel. The amount of information involved in determining the position of the cone barrel is large, which improves the accuracy of the obtained position of the cone barrel and avoids the situation where the accuracy of the position of the cone barrel is not affected by the distance between the vehicle and the cone barrel.

[0058] The information acquisition method provided by the embodiment of the present invention takes into account that the actual position of the camera of the target vehicle and the current position of the camera of the target vehicle may be different when the position of the cone barrel in the target environment image in the image coordinate system of the target environment image is transformed into the position of the cone barrel in the body coordinate system of the target vehicle. Among them, the intrinsic parameters in the position of the camera are fixed, and the change of the extrinsic parameters in the position of the camera is usually caused by the change of the pitch angle. The difference in the position of the same camera at different times is usually caused by the difference in the pitch angle of the same camera at different times. Directly using the current position of the camera of the target vehicle to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the position of the cone barrel in the body coordinate system of the target vehicle will lead to a decrease in the accuracy of the position of the cone barrel.

[0059] The information acquisition method provided by the embodiment of the present invention determines whether there is a target pitch angle related to the target environment image based on the current posture of the target vehicle's camera, the segmentation result, and the detected three-dimensional lane line point cloud. This is equivalent to determining whether there is a more accurate pitch angle, namely the target pitch angle, relative to the current posture of the target vehicle's camera. When it is determined that there is a more accurate pitch angle, a more accurate target correction posture relative to the current posture of the target vehicle's camera is obtained by replacing the current pitch angle in the current posture of the target vehicle's camera with the target pitch angle. The more accurate target correction posture is used to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the body coordinate system of the target vehicle. A more accurate corrected position of the cone barrel in the body coordinate system of the target vehicle is obtained, thereby improving the accuracy of the acquired position of the cone barrel. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A flowchart of an information acquisition method provided in an embodiment of the present disclosure;

[0061] Figure 2 A flowchart of another information acquisition method provided by an embodiment of the present disclosure;

[0062] Figure 3 A schematic diagram of an example of generating particles based on the corrected position of the cone in the target vehicle's body coordinate system;

[0063] Figure 4 A schematic diagram showing the effect of performing a tracking operation corresponding to a target environment image;

[0064] Figure 5 A schematic diagram of the effect of performing an example of a search operation;

[0065] Figure 6 A schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The following describes the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art will readily appreciate the other advantages and benefits of the present invention from the disclosure herein. The present invention may also be implemented or applied through various other specific embodiments, and the various details in this specification may be modified or altered based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are intended only to illustrate the present invention and are not intended to limit the scope of protection of the present invention.

[0067] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0068] refer to Figure 1 , which shows a flow chart of the information acquisition method provided by an embodiment of the present disclosure.

[0069] In step S101 , a segmentation result of a target environment image captured by a camera of a target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling are obtained.

[0070] The target vehicle can be any vehicle to which the information acquisition method provided in the embodiments of the present disclosure can be applied. The target environment image is captured by a camera of the target vehicle. The target environment image describes the environment in which the target vehicle is located.

[0071] It should be noted that the method provided by the embodiment of the present disclosure can be performed within a target time period. Each environmental image collected within the target time period includes the cone barrels on the road on which the target vehicle is traveling. The target environmental image can be any environmental image collected within the target time period. For each environmental image collected within the target time period, at least steps S101-S102 can be performed. The starting time of the target time period can be: the first environmental image collected within the target time period, that is, the collection time of the environmental image with the earliest collection time, and the ending time of the target time period can be: the last environmental image collected within the target time period, that is, the collection time of the environmental image with the latest collection time.

[0072] In the disclosed embodiment, the cone barrel in the target environment image can be detected by a target detection network.

[0073] As an example, the target detection network is a YOLO series network, Faster RCNN, etc.

[0074] In the disclosed embodiment, the target lane line may be a lane line randomly selected from lane lines on both sides of the lane where the target vehicle is traveling. The target lane line may also be a lane line randomly selected from lane lines detected on the target road where the target vehicle is traveling.

[0075] The segmentation result of the target environment image indicates: a plurality of lane line pixel points of a target lane line on the target road.

[0076] That is, for a target lane line on a target road, the segmentation result of the target environment image indicates which pixel points in the target environment image are lane line pixel points of the target lane line.

[0077] Each lane line pixel point among the multiple lane line pixel points of the target lane line has coordinates in the image coordinate system of the target environment image.

[0078] In step S101, the target environment image may be input into an instance segmentation network, and the instance segmentation network outputs a segmentation result of the target environment image.

[0079] In an embodiment of the present disclosure, the detected three-dimensional lane line point cloud of the target lane line is detected based on the target environment image by a neural network for detecting lane lines.

[0080] In step S101, the target environment image may be input into a three-dimensional lane line detection network, and the lane line detection network outputs a detected three-dimensional lane line point cloud of a target lane line on a target road on which a target vehicle is traveling.

[0081] The detected 3D lane line point cloud includes: a plurality of 3D points of a target lane line on a target road on which the target vehicle is traveling, and the 3D points of the target lane line are in the body coordinate system of the target vehicle.

[0082] In one possible implementation, the three-dimensional lane detection network is a Bev lane detection network.

[0083] In step S102, based on the current posture of the camera of the target vehicle, the segmentation result of the target environment image, and the detected three-dimensional lane line point cloud of the target lane line, it is determined whether there is a target pitch angle related to the target environment image.

[0084] The current position and posture of the camera of the target vehicle may refer to the position and posture of the camera of the target vehicle that was last determined before the start of S102 .

[0085] In the embodiment of the present disclosure, the target corrected posture corresponding to the target pitch angle associated with the target environment image is obtained by replacing the current pitch angle in the current posture of the camera of the target vehicle with the target pitch angle associated with the target environment image.

[0086] The error between the target projected three-dimensional lane line point cloud obtained by inversely projecting multiple lane line pixel points of the target lane line into the body coordinate system of the target vehicle with a target corrected posture corresponding to the target pitch angle related to the target environment image and the detected three-dimensional lane line point cloud is less than an error threshold.

[0087] In other words, for a pitch angle, if the target corrected posture corresponding to the pitch angle reversely projects multiple lane line pixel points of the target lane line into the body coordinate system of the target vehicle, and the error between the projected three-dimensional lane line point cloud of the target lane line and the detected three-dimensional lane line point cloud is less than the error threshold, then the pitch angle is determined as the target pitch angle associated with the target environment image.

[0088] As an example, to determine whether a target pitch angle associated with the target environment image exists, a pitch angle search operation can be iteratively performed until the target pitch angle is found or the number of pitch angle search operations performed reaches a threshold. Each pitch angle search operation targets a reference pitch angle. The reference pitch angle targeted by the first pitch angle search operation is the current pitch angle of the target vehicle's camera. The reference pitch angle targeted by the i-th pitch angle search operation after the first pitch angle search operation is the subinterval endpoint pitch angle corresponding to the 3D lane line point cloud with the minimum error between the detected 3D lane line point cloud and the i-1th pitch angle search operation.

[0089] In this example, the k-th pitch angle search operation includes: subtracting a preset angle from the reference pitch angle for the k-th pitch angle search operation to obtain a left endpoint pitch angle corresponding to the k-th pitch angle search operation; and adding the preset angle to the reference pitch angle for the k-th pitch angle search operation to obtain a right endpoint pitch angle corresponding to the k-th pitch angle search operation. The k-th time is the time other than the last time.

[0090] In this example, a first pitch angle interval is divided into a plurality of first sub-intervals, and each endpoint of each of the plurality of first sub-intervals serves as a sub-interval endpoint pitch angle corresponding to the k-th pitch angle search operation. The two endpoints of the first pitch angle interval are the left endpoint pitch angle corresponding to the k-th pitch angle search operation and the reference pitch angle targeted by the k-th pitch angle search operation. A second pitch angle interval is divided into a plurality of second sub-intervals, and the two endpoints of the second pitch angle interval are the right endpoint pitch angle corresponding to the k-th pitch angle search operation and the reference pitch angle targeted by the k-th pitch angle search operation.

[0091] In this example, each endpoint of each second sub-interval in the plurality of second sub-intervals is used as the sub-interval endpoint pitch angle corresponding to the k-th execution of the pitch angle search operation.

[0092] In this example, the pose corresponding to the pitch angle of each subinterval endpoint corresponding to the k-th execution of the pitch angle search operation is determined. The pose corresponding to the pitch angle of the subinterval endpoint is obtained by replacing the current pitch angle in the current pose of the target vehicle's camera with the pose corresponding to the pitch angle of the subinterval endpoint.

[0093] In this example, for each sub-interval endpoint pitch angle corresponding to the pitch angle search operation executed for the kth time, multiple lane line pixel points of the target lane line are reversely projected to the body coordinate system of the target vehicle with the posture corresponding to the sub-interval endpoint pitch angle to obtain a projected three-dimensional lane line point cloud corresponding to the sub-interval endpoint pitch angle, and the error between the projected three-dimensional lane line point cloud corresponding to the sub-interval endpoint pitch angle and the detected three-dimensional lane line point cloud is determined.

[0094] In this example, the projected 3D lane line point cloud corresponding to the subinterval endpoint pitch angle with the smallest error relative to the detected 3D lane line point cloud is recorded as 3D lane line point cloud j. If the error between 3D lane line point cloud j and the detected 3D lane line point cloud is less than an error threshold, the subinterval endpoint pitch angle corresponding to 3D lane line point cloud j is determined as the target pitch angle associated with the target environment image. If the error between 3D lane line point cloud j and the detected 3D lane line point cloud is not less than the error threshold, the subinterval endpoint pitch angle corresponding to 3D lane line point cloud j is determined as the reference pitch angle for the k+1th pitch angle search operation.

[0095] In an embodiment of the present disclosure, for a pitch angle, to determine the error between a projected 3D lane line point cloud obtained by reverse-projecting multiple lane line pixel points of the target lane line into the target vehicle's body coordinate system at the pose corresponding to the pitch angle and a 3D point in the detected 3D lane line point cloud, the projected 3D lane line point cloud and the detected 3D lane line point cloud can be sampled using the same sampling method, such as uniform sampling, to obtain multiple sampling points of the projected 3D lane line point cloud and multiple sampling points of the detected 3D lane line point cloud. A sampling point pair corresponding to each sampling point of the projected 3D lane line point cloud is determined to obtain multiple sampling point pairs. The sampling point pair corresponding to a sampling point of the projected 3D lane line point cloud includes: a sampling point of the projected 3D lane line point cloud and a sampling point in the detected 3D lane line point cloud that is closest to the sampling point of the projected 3D lane line point cloud. For each sampling point pair, the distance between the two sampling points in the pair is calculated to obtain the distance of the sampling point pair. The distances of each sampling point pair are summed to obtain the error.

[0096] In step S103, when it is determined that there is a target pitch angle associated with the target environment image, the target correction posture corresponding to the target pitch angle associated with the target environment image is used to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the body coordinate system of the target vehicle.

[0097] For a cone barrel in the target environment image, the position of the cone barrel in the image coordinate system of the target environment image is: the position of the center point of the detection box surrounding the cone barrel output by the target detection network in the image coordinate system of the target environment image.

[0098] In the disclosed embodiment, the current pitch angle in the current pose of the target vehicle's camera can be replaced with the target pitch angle associated with the target environment image to obtain a target corrected pose corresponding to the target pitch angle associated with the target environment image. The target corrected pose corresponding to the target pitch angle associated with the target environment image can also be obtained using the following formula:

[0099]

[0100] Candidate corrected pose corresponding to the candidate pitch angle = ΔT v'v T vc

[0101] Among them, T vc is the current pose of the target vehicle, and θ is the target pitch angle associated with the target environment image.

[0102] refer to Figure 2 , which shows a flow chart of another information acquisition method provided by an embodiment of the present disclosure.

[0103] In step S201 , a segmentation result of a target environment image and a detected three-dimensional lane line point cloud of a target lane line on a target road on which a target vehicle is traveling are obtained.

[0104] It should be noted that another information acquisition method provided by an embodiment of the present disclosure can be performed within a target time period. Each environmental image captured by the camera of the target vehicle within the target time period includes the cone barrels on the road on which the target vehicle is traveling. The target environmental image can be any environmental image captured within the target time period. For each environmental image captured within the target time period, at least step S201, step S202, and step S204 can be executed. The starting time of the target time period can be: the first environmental image captured within the target time period, that is, the capture time of the environmental image with the earliest capture time, and the ending time of the target time period can be: the last environmental image captured within the target time period, that is, the capture time of the environmental image with the latest capture time.

[0105] During the target time period, the first environmental image can be used as the environmental image for particle initialization. For each other environmental image except the first environmental image collected during the target time period, cone bucket tracking corresponding to the other image is performed, that is, step S204 is performed for the other environmental image.

[0106] In step S202, it is determined whether there is a target pitch angle associated with the target environment image based on the current posture of the camera of the target vehicle and the detected three-dimensional lane line point cloud of the target lane line on the target road on which the target vehicle is traveling.

[0107] In step S203, when it is determined that there is a target pitch angle associated with the target environment image, the target correction posture corresponding to the target pitch angle associated with the target environment image is used to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the body coordinate system of the target vehicle.

[0108] In one possible implementation, the outline of the target cone is determined based on the segmentation result of the target environment image, where the target cone is any cone in the target environment image; the position of the intersection of the outline of the target cone and the perpendicular bisector of the detection frame surrounding the target cone is determined as the position of the target cone in the image coordinate system of the target environment image.

[0109] In step S204, cone-bucket tracking corresponding to the target environment image is performed.

[0110] Considering the possibility of a longitudinal deviation between the corrected position of the cone in the target environment image in the body coordinate system of the target vehicle and the actual position of the cone in the target environment image in the body coordinate system of the target vehicle, to ensure that the position of the cone in the body coordinate system of the target vehicle can be stably output, S204 is executed to track the position of the cone in the body coordinate system of the target vehicle, i.e., to perform cone tracking corresponding to the target environment image, and obtain and output the stable position of the cone, i.e., the tracked position of the target cone.

[0111] Step S204 includes: step S2041-step S2042.

[0112] In step S2041, a tracking operation corresponding to the target environment image is performed.

[0113] In step S2042, the target particle is found from the target particle set, the position of the target particle is used as the tracking position of the corresponding cone barrel in the target environment image, and the tracking position of the corresponding cone barrel in the target environment image is output.

[0114] In one possible implementation, for a target particle, the cone barrel whose corresponding position in the target vehicle's body coordinate system in the target environment image is closest to the target particle's position can be determined, and the position of the target particle can be used as the tracking position of the cone barrel whose corresponding position in the target vehicle's body coordinate system in the target environment image is closest to the target particle.

[0115] In the embodiment of the present disclosure, particles in the target particle set have positions in the body coordinate system of the target vehicle.

[0116] It should be noted that the target particle set does not specifically refer to a single particle set. The target particle set should be understood as a variable. As cone tracking is performed sequentially for each additional environment image, the particles in the target particle set will change. In the disclosed embodiments, when performing operations involving the target particle set, the particles in the target particle set should be determined in context.

[0117] In the embodiment of the present disclosure, the target particle set is initialized as an initial particle set before the first cone-bucket tracking is performed.

[0118] It should be noted that the first cone tracking specifically refers to: cone tracking corresponding to the second environmental image collected within the target time period. In other words, cone tracking starts from the second environmental image collected within the target time period. All environmental images collected within the target time period are sorted from earliest to latest according to the collection time, and the order of all environmental images collected within the target time period is obtained. The first environmental image collected within the target time period is the earliest environmental image collected, and the last environmental image collected within the target time period is the latest environmental image collected.

[0119] In the disclosed embodiment, the initial particle set is obtained by generating particles according to the corresponding positions of the cone barrels in the body coordinate system of the target vehicle in the environment image used for particle initialization.

[0120] It should be noted that the environment image used for particle initialization may be: the first environment image collected within the target time period.

[0121] In the embodiment of the present disclosure, for each cone barrel in the first environmental image collected within the target time period, if the corrected position of the cone barrel in the body coordinate system of the target vehicle is obtained, then the corresponding position of the cone barrel in the body coordinate system of the target vehicle is: the corrected position of the cone barrel in the body coordinate system of the target vehicle; if the corrected position of the cone barrel in the body coordinate system of the target vehicle is not obtained, then the corresponding position of the cone barrel in the body coordinate system of the target vehicle is: the original position of the cone barrel in the body coordinate system of the target vehicle. The original position of the cone barrel in the body coordinate system of the target vehicle refers to: the position of the cone barrel in the image coordinate system of the target environmental image is transformed into the position in the body coordinate system of the target vehicle based on the current posture of the target vehicle.

[0122] In the disclosed embodiment, to obtain an initial particle set, for each cone in the first environmental image captured by the target vehicle's camera during a target time period, a particle is generated based on the corresponding position of the cone in the target vehicle's body coordinate system. The particles generated based on the corresponding position of each cone in the first environmental image captured during the target time period in the target vehicle's body coordinate system constitute the initial particle set.

[0123] In the disclosed embodiment, for a cone barrel i in the target environment image, when generating particles based on the corresponding position of the cone barrel in the body coordinate system of the target vehicle, a square having a center point as the corresponding position of the cone barrel i in the body coordinate system of the target vehicle and a side length of a preset side length can be determined. The corresponding position of the cone barrel i in the body coordinate system of the target vehicle, the corner points of the square, and the center of each side of the square are respectively used as particle positions, for a total of 9 particle positions. For each particle position, a particle is generated at that particle position. A total of 9 particles are generated.

[0124] Among them, cone bucket i is any cone bucket in any environment image collected within the target time period.

[0125] In the embodiment of the present disclosure, step S2041 includes: step S20411 and step S20412.

[0126] In step S20411, when a first particle from the target particle set exists in the target area associated with the target cone, the weight of the first particle is updated according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle, and when a second particle exists in the target particle set, the weight of the second particle is reduced, wherein the second particle is a particle that is not in the target area associated with any cone in the target environment image.

[0127] In step S20412, if no particles from the target particle set exist within the target area associated with the target cone, a new particle generated based on the corresponding position of the target cone in the target vehicle's body coordinate system is added to the target particle set. When generating a new particle based on the corresponding position of the target cone in the target vehicle's body coordinate system, a square having a center point corresponding to the corresponding position of the target cone in the target vehicle's body coordinate system and a side length of a preset length can be determined. The corresponding position of the target cone in the target vehicle's body coordinate system, the corner points of the square, and the center of each side of the square are used as particle positions. For each particle position, a particle at that particle position is generated, and the particle at that particle position is considered a new particle, which is then added to the target particle set. Thus, the target particle set is updated. Upon completion of the update, the target particle set consists of all particles in the target particle set at the time step S2041 is initiated, plus each new particle.

[0128] It should be noted that the target cone is any cone in the target environment image. The particles in the target area associated with the target cone and from the target particle set can be called first particles, and the particles in the target area not associated with any cone in the target environment image can be called second particles.

[0129] The target area associated with the target cone is an area whose center point is the corresponding position of the target cone in the body coordinate system of the target vehicle and whose shape is a preset shape.

[0130] For each first particle, the weight of the first particle is updated according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle.

[0131] The current weight of the first particle may refer to: the weight of the first particle at the moment when the weight of the first particle begins to be updated according to the distance between the first particle and the target cone barrel, the weight of the target cone barrel, and the current weight of the first particle.

[0132] For a first particle, the distance between the first particle and the target cone barrel may specifically refer to: the distance between the position of the first particle in the body coordinate system of the target vehicle and the corresponding position of the target cone barrel in the body coordinate system of the target vehicle.

[0133] In the embodiment of the present disclosure, each cone bucket in the target environment image has a weight.

[0134] In a possible implementation, each cone bucket in the target environment image has the same preset weight.

[0135] In another possible implementation, for a cone barrel in the target environment image, when the cone barrel has a corrected position in the body coordinate system of the target vehicle, the first weight is used as the weight of the cone barrel; when the cone barrel does not have a corrected position in the body coordinate system of the target vehicle, the second weight is used as the weight of the target cone barrel, wherein the second weight is less than the first weight.

[0136] The second weight is smaller than the first weight because: when the cone barrel has a corrected position in the body coordinate system of the target vehicle, the higher the credibility of the cone barrel's corrected position in the body coordinate system or the cone barrel's corrected position in the body coordinate system as the tracking position of the cone barrel, the more important the cone barrel's corrected position in the body coordinate system of the target vehicle is in the cone barrel tracking process, and a higher weight is given to the cone barrel, thereby improving the accuracy of the corresponding result obtained in the cone barrel tracking according to the corrected position of the cone barrel in the body coordinate system of the target vehicle, and improving the accuracy of the cone barrel's tracking position obtained by cone barrel tracking.

[0137] As an example, the weight of a cone barrel with a corrected position in the body coordinate system of the target vehicle is 1, and the weight of a cone barrel without a corrected position in the body coordinate system of the target vehicle is 0.5.

[0138] It should be noted that for a cone in the target environment image, each particle generated based on the cone's position has an initial weight. In other words, for a particle, after it is generated and before its weight is updated for the first time, its weight is the initial weight of the particle.

[0139] In the disclosed embodiment, for a cone barrel in the target environment image, the weight of the particle generated according to the corresponding position of the cone barrel in the body coordinate system of the target vehicle is less than the weight of the cone barrel. The weight of the particle generated according to the corresponding position of the cone barrel in the body coordinate system of the target vehicle may have a multiple relationship with the weight of the cone barrel. The weight of the particle with the position corresponding to the cone barrel in the body coordinate system of the target vehicle among all the particles generated according to the corresponding position of the cone barrel in the body coordinate system of the target vehicle is greater than the weights of other particles among all the particles. The weight of the particle with the position corresponding to the cone barrel in the body coordinate system of the target vehicle among all the particles may have a multiple relationship with the weights of other particles among all the particles.

[0140] As an example, for a cone barrel in the target environment image, the weight of the cone barrel is denoted as w, the weight of the particle at the corresponding position of the cone barrel in the body coordinate system of the target vehicle is 4*w / 9, and the weight of other particles is w / 9.

[0141] refer to Figure 3 , which shows a schematic diagram of the effect of an example of generating particles based on the corrected position of the cone barrel in the body coordinate system of the target vehicle.

[0142] In this example, a total of nine particles are generated based on the corrected position of a cone in the target environment image within the target vehicle's body coordinate system. A square with a center point determined as the cone's corresponding position within the target vehicle's body coordinate system and a side length of a preset length is used as the particle position. For each particle position, a particle is generated at that position, for a total of nine particles. The cone's weight is denoted as w. The weight of particle 301 at the cone's corresponding position within the target vehicle's body coordinate system is 4*w / 9, and the weights of the other particles 302 are w / 9.

[0143] In one possible implementation, in step S20411, for a first particle, according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle, updating the weight of the first particle includes: determining the coefficient corresponding to the distance between the first particle and the target cone according to the set distance between the particle and the target cone and the correlation relationship between the coefficients, multiplying the coefficient corresponding to the distance between the first particle and the target cone, the weight of the target cone, and a preset constant to obtain a product, adding the product to the current weight of the first particle to obtain the updated weight of the first particle, and updating the weight of the first particle to the updated weight of the first particle.

[0144] As an example, for a first particle in the target area relative to the target cone, the weight of the first particle can be updated using the following formula:

[0145] w2=w1+w new *0.4*e (-3.078*distance)

[0146] Among them, w new is the weight of the target cone barrel, distance is the distance between the first particle and the target cone barrel, w1 is the current weight of the first particle, and w2 is the updated weight of the first particle.

[0147] In step S20412, for a second particle, the weight of the second particle is reduced by subtracting a preset weight reduction amount from the current weight of the second particle. The current weight of the second particle is the weight of the second particle at the time the weight reduction is started.

[0148] As an example, the reduction amount is, for example, 0.2.

[0149] In step S20412, when there are no particles from the target particle set in the target area, new particles generated according to the corresponding position of the target cone barrel in the vehicle body coordinate system are added to the target particle set.

[0150] In one possible implementation of step S2042, multiple third particles are retrieved from the target particle set, where the weight of the third particles is greater than a weight threshold; multiple fourth particles are retrieved from the multiple third particles; the multiple fourth particles are sorted from largest to smallest according to their weights; and after sorting, each of the first target number of fourth particles is determined as a target particle, where the target number is the number of cones in the target environment image. For a third particle, if the third particle meets a first condition, the third particle is determined as a fourth particle, where the first condition includes: there are other particles within the target region associated with the third particle, and the other particles within the target region associated with the third particle do not appear in other target regions, where the other particles are particles other than the third particle, and the other target regions are regions associated with particles other than the third particle. For a third particle, if the third particle meets the second condition, the third particle is determined as the fourth particle, and the second condition includes: there are other particles in the target area related to the third particle, the corresponding other particles in the target area related to the third particle appear in the corresponding other target area, and the weight of the third particle is greater than the weight of the other particles related to the corresponding other target area.

[0151] It should be noted that, in the embodiment of the present disclosure, for a particle, the region associated with the particle is: a region having a center point that is the position of the particle in the body coordinate system of the target vehicle and having a shape that is a preset shape.

[0152] As an example, the preset shape is a circle.

[0153] refer to Figure 4 , which shows a schematic diagram of the effect of performing a tracking operation corresponding to a target environment image.

[0154] Figure 4 Particle 401 is shown, having a position corresponding to cone 1 in the body coordinate system of the target vehicle. The target area associated with cone 401 is circular area 402. Particle 403 is the first particle from the target particle set within the target area associated with the target cone, and particle 404 is the second particle. Figure 4 The particle 405 having the position corresponding to the cone barrel 2 in the body coordinate system of the target vehicle is shown. There are no particles from the target particle set in the target area associated with the cone barrel 2. A new particle is generated based on the position corresponding to the cone barrel 2 in the body coordinate system of the target vehicle. Figure 4 A new particle 406 is shown. The new particle generated according to the corresponding position of the cone barrel 2 in the body coordinate system of the target vehicle is added to the target particle set. Thus, the particles in the target particle set are updated.

[0155] In another possible implementation of step S2042, step S2042 includes: step S20421.

[0156] In step S20421, when there are particles with weights less than 0 in the target particle set, the particles with weights less than 0 are deleted from the target particle set; candidate particles with weights greater than a weight threshold are found from the target particle set; and the target particle is found from multiple candidate particles.

[0157] As an example, the weight threshold is 1.

[0158] In one possible implementation, finding a target particle from a plurality of candidate particles includes: sorting the plurality of candidate particles from large to small according to the weights of the candidate particles to obtain an order of the candidate particles; iteratively performing a search operation until a target number of target particles is found, wherein the target number is the number of cone barrels in the target environment image. The search operation performed for the i-th time includes: determining whether there are other particles that have not been associated with the candidate particle in the target area related to the candidate particle targeted by the search operation performed for the i-th time, wherein the candidate particle targeted by the first search operation is the particle with the largest weight among the plurality of candidate particles, the candidate particle targeted by the non-first search operation is the next candidate particle of the candidate particle targeted by the previous search operation of the non-first search operation, the non-first search operation is a search operation other than the first search operation, and the next candidate particle of the candidate particle targeted by the previous search operation of the non-first search operation is indicated by the order; if so, the candidate particle targeted by the search operation performed for the i-th time is determined as the target particle, And other particles in the target area related to the candidate particle targeted by the search operation executed for the i-th time that have not been associated with the candidate particle are associated with the candidate particle targeted by the search operation executed for the i-th time, and the next candidate particle of the candidate particle targeted by the search operation executed for the i-th time is determined as the candidate particle targeted by the next search operation of the search operation executed for the i-th time, that is, the i+1-th search operation; if not, the next candidate particle of the candidate particle targeted by the search operation executed for the i-th time is determined as the candidate particle targeted by the next search operation of the search operation executed for the i-th time, that is, the i+1-th search operation, wherein the next candidate particle of the candidate particle targeted by the search operation executed for the i-th time is indicated by this order.

[0159] The target area associated with a candidate particle is an area having a center point that is the position of the candidate particle in the body coordinate system of the target vehicle and a shape that is a preset shape.

[0160] In one possible implementation, a circular area having a center point as the position of the candidate particle targeted by the search operation in the body coordinate system of the target vehicle and a preset second radius as the radius is determined as the target area related to the candidate particle targeted by the search operation.

[0161] As an example, the second radius is preset to be 0.9 m.

[0162] refer to Figure 5 , which shows a schematic diagram of the effect of performing an example of a search operation.

[0163] Schematic diagram of the effect. Figure 5 Candidate particle 501, candidate particle 503, and candidate particle 505 are shown. Candidate particle 501 has a weight of 1.34, candidate particle 503 has a weight of 1.31, and candidate particle 505 has a weight of 1.21.

[0164] In this example, the first search operation is for candidate particle 501 , the second search operation is for candidate particle 502 , and the third search operation is for candidate particle 503 .

[0165] During the first search operation, it is determined that other particles that have not been associated with the candidate particle 501 exist in the target region 502 associated with the candidate particle 501 . Therefore, the candidate particle 501 is determined as the target particle.

[0166] During the second search operation, it is determined that there are no other particles in the target region 504 associated with the candidate particle 503 that have not been associated with the candidate particle. Therefore, during the second search operation, the candidate particle 503 is not determined as the target particle.

[0167] During the third search operation, it is determined that other particles that have not been associated with the candidate particle 505 exist in the target region 506 associated with the candidate particle 505 . Therefore, the candidate particle 505 is determined as the target particle.

[0168] An information acquisition device is also provided in an embodiment of the present invention, which is used to implement the above-mentioned method embodiments and preferred implementation methods, and will not be repeated here. As used below, the term "unit" can implement a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceived. The devices in the embodiments of the present invention are presented in the form of functional units, where the functional units refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories that execute one or more software or fixed programs, and / or other devices that can provide the above-mentioned functions.

[0169] The information acquisition device includes:

[0170] an acquisition unit, configured to acquire a segmentation result of a target environment image captured by a camera of a target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling, wherein the segmentation result indicates a plurality of lane line pixels of the target lane line, and the detected three-dimensional lane line point cloud is detected based on the target environment image;

[0171] a determining unit, configured to determine whether a target pitch angle associated with a target environment image exists based on a current pose of the camera, the segmentation result, and the detected three-dimensional lane line point cloud, wherein an error between a target projected three-dimensional lane line point cloud obtained by inversely projecting the plurality of lane line pixel points into a body coordinate system of a target vehicle using a target corrected pose corresponding to the target pitch angle and the detected three-dimensional lane line point cloud is less than an error threshold, and the target corrected pose is obtained by replacing a current pitch angle in the current pose of the camera with the target pitch angle;

[0172] A conversion unit is used to, when it is determined that the target pitch angle exists, use the target corrected posture to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the vehicle body coordinate system.

[0173] Furthermore, the information acquisition device further includes:

[0174] The cone position acquisition unit is used to determine the outline of the target cone based on the segmentation result of the target environment image, wherein the target cone is any cone in the target environment image; and the position of the intersection of the outline of the target cone and the perpendicular bisector of the detection frame surrounding the target cone is determined as the position of the target cone in the image coordinate system of the target environment image.

[0175] Furthermore, the information acquisition device further includes:

[0176] A tracking unit is configured to perform cone-bucket tracking corresponding to a target environment image, wherein the cone-bucket tracking includes:

[0177] Performing a tracking operation corresponding to a target environment image, and after performing the tracking operation, searching for a target particle from a target particle set, using the position of the target particle as the tracking position of the corresponding cone barrel in the target environment image, and outputting the tracked position, wherein the target particle set is initialized to an initial particle set before performing the first cone barrel tracking, and the initial particle set is obtained by generating particles according to the corresponding position of the cone barrel in the environment image used for particle initialization in the vehicle body coordinate system; the tracking operation includes:

[0178] When a first particle from the target particle set exists in the target area associated with the target cone, the weight of the first particle is updated according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle. When a second particle exists in the target particle set, the weight of the second particle is reduced, wherein the second particle is a particle that is not in the target area associated with any cone in the target environment image.

[0179] When there are no particles from the target particle set in the target area, new particles generated according to the corresponding position of the target cone barrel in the vehicle coordinate system are added to the target particle set.

[0180] Furthermore, the information acquisition device further includes:

[0181] A weight allocation unit is used to use a first weight as the weight of the target cone barrel when the target cone barrel has a corrected position in the vehicle body coordinate system; when the target cone barrel does not have a corrected position in the vehicle body coordinate system, use a second weight as the weight of the target cone barrel, wherein the second weight is smaller than the first weight.

[0182] Furthermore, the information acquisition device further includes:

[0183] The first target area determination unit is used to determine a circular area with a center point as the corresponding position of the target cone barrel in the vehicle body coordinate system and a preset first radius as the target area related to the target cone barrel.

[0184] Furthermore, the tracking unit is also used to delete particles with weights less than 0 from the target particle set when there are particles with weights less than 0 in the target particle set; find multiple candidate particles from the target particle set, where the weights of the candidate particles are greater than a weight threshold; and find the target particle from the multiple candidate particles.

[0185] Furthermore, the tracking unit is further used to sort multiple candidate particles from large to small according to the weights of the candidate particles to obtain the order of the candidate particles; iteratively perform the search operation until a target number of target particles are found, wherein the target number is the number of cone barrels in the target environment image, and the search operation includes: determining whether there are other particles that have not been associated with the candidate particle in the target area related to the candidate particle targeted by the search operation, wherein the candidate particle targeted by the first search operation is the particle with the largest weight among the multiple candidate particles, and the candidate particle targeted by the non-first search operation is the next candidate particle of the candidate particle targeted by the previous search operation of the non-first search operation, wherein the previous search operation is performed. The next candidate particle of the candidate particle targeted by the search operation is indicated by the sequence; if so, the candidate particle targeted by the search operation is determined as the target particle, and other particles in the target area related to the candidate particle targeted by the search operation that have not been associated with the candidate particle are associated with the candidate particle targeted by the search operation, and the next candidate particle of the candidate particle targeted by the search operation is determined as the candidate particle targeted by the next search operation of the search operation; if not, the next candidate particle of the candidate particle targeted by the search operation is determined as the candidate particle targeted by the next search operation of the search operation, wherein the next candidate particle of the candidate particle targeted by the search operation is indicated by the sequence.

[0186] Furthermore, the information acquisition device further includes:

[0187] The second target area determination unit is used to determine a circular area having a center point as the position of the candidate particle targeted by the search operation in the vehicle body coordinate system and a preset second radius as a target area related to the candidate particle targeted by the search operation.

[0188] refer to Figure 6 , Figure 61 is a schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. The computer device includes: one or more processors 10, a memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components communicate with each other using different buses and can be installed on a common motherboard or in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in or on the memory for displaying graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used with multiple memories and multiple storage devices. Similarly, multiple vehicles can be connected, with each device providing some of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). The processor 10 can be a central processing unit, a network processor, or a combination thereof. The processor 10 can further include a hardware chip. The hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device can be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof. The memory 20 stores instructions executable by at least one processor 10, causing the at least one processor 10 to perform the methods described in the above embodiments. The memory 20 may include a program storage area and a data storage area. The program storage area may store an operating system and applications required for at least one function; the data storage area may store data generated based on vehicle usage. Furthermore, the memory 20 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state memory device. In some optional embodiments, the memory 20 may optionally include memory remote from the processor 10, which may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. The memory 20 may include volatile memory, such as random access memory; non-volatile memory, such as flash memory, a hard disk, or a solid-state drive; or a combination of these types of memory. The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30, and the output device 40 may be connected via a bus or other means. The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touchpad, an indicator stick, one or more mouse buttons, a trackball, a joystick, etc.The output device 40 may include a display device, an auxiliary lighting device (e.g., LED), and a tactile feedback device (e.g., a vibration motor). The display device includes, but is not limited to, a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device may be a touch screen.

[0189] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0190] A portion of the embodiments of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0191] The above embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Any equivalent substitution or modification made by those skilled in the art based on the present invention is within the protection scope of the present invention.

Claims

1. An information acquisition method, characterized in that: The method comprises: Obtaining a segmentation result of a target environment image captured by a camera of the target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling, wherein the segmentation result indicates a plurality of lane line pixels of the target lane line, and the detected three-dimensional lane line point cloud is detected based on the target environment image; Determining whether there is a target pitch angle associated with a target environment image based on the current pose of the camera, the segmentation result, and the detected three-dimensional lane line point cloud, wherein an error between a target projected three-dimensional lane line point cloud obtained by inversely projecting the plurality of lane line pixel points into a body coordinate system of a target vehicle using a target corrected pose corresponding to the target pitch angle and the detected three-dimensional lane line point cloud is less than an error threshold, and the target corrected pose is obtained by replacing a current pitch angle in the current pose of the camera with the target pitch angle; When it is determined that the target pitch angle exists, the target corrected posture is used to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the vehicle body coordinate system.

2. The method according to claim 1, wherein: The method further comprises: Determine the outline of the target cone according to the segmentation result of the target environment image, wherein the target cone is any cone in the target environment image; The position of the intersection of the outline of the target cone barrel and the perpendicular bisector of the detection frame surrounding the target cone barrel is determined as the position of the target cone barrel in the image coordinate system of the target environment image.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: Perform cone-bucket tracking corresponding to the target environment image, the cone-bucket tracking comprising: Performing a tracking operation corresponding to a target environment image, and after performing the tracking operation, searching for a target particle from a target particle set, using the position of the target particle as the tracking position of the corresponding cone barrel in the target environment image, and outputting the tracked position, wherein the target particle set is initialized to an initial particle set before performing the first cone barrel tracking, and the initial particle set is obtained by generating particles according to the corresponding position of the cone barrel in the environment image used for particle initialization in the vehicle body coordinate system; the tracking operation includes: When a first particle from the target particle set exists in the target area associated with the target cone, the weight of the first particle is updated according to the distance between the first particle and the target cone, the weight of the target cone, and the current weight of the first particle. When a second particle exists in the target particle set, the weight of the second particle is reduced, wherein the second particle is a particle that is not in the target area associated with any cone in the target environment image. When there are no particles from the target particle set in the target area associated with the target cone, new particles generated according to the corresponding position of the target cone in the vehicle coordinate system are added to the target particle set.

4. The method according to claim 3, wherein: The method further comprises: When the target cone has a corrected position in the vehicle coordinate system, using the first weight as the weight of the target cone; When the target cone barrel does not have a corrected position in the vehicle body coordinate system, the second weight is used as the weight of the target cone barrel, wherein the second weight is smaller than the first weight.

5. The method according to claim 3, wherein: The method further comprises: The position of the center point is the corresponding position of the target cone barrel in the vehicle body coordinate system, and the circular area with a preset first radius as the radius is determined as the target area related to the target cone barrel.

6. The method according to claim 3, wherein: Finding target particles from the target particle set includes: When there are particles with weights less than 0 in the target particle set, the particles with weights less than 0 are deleted from the target particle set; Find multiple candidate particles from the target particle set, where the weight of the candidate particles is greater than the weight threshold; Find the target particle from multiple candidate particles.

7. The method according to claim 6, characterized in that: Finding the target particle from multiple candidate particles includes: Sort multiple candidate particles according to their weights from large to small to obtain the order of the candidate particles; The search operation is iteratively performed until a target number of target particles is found, where the target number is the number of cone barrels in the target environment image. The search operation includes: determining whether there are other particles in a target area associated with the candidate particle targeted by the search operation that have not been associated with the candidate particle, wherein the candidate particle targeted by the first search operation is a particle with the largest weight among the multiple candidate particles, and the candidate particle targeted by the non-first search operation is a next candidate particle of the candidate particle targeted by the last search operation of the non-first search operation, wherein the next candidate particle of the candidate particle targeted by the last search operation is indicated by the order; If yes, determining the candidate particle targeted by the search operation as the target particle, and associating other particles in the target area associated with the candidate particle targeted by the search operation that have not been associated with the candidate particle with the candidate particle targeted by the search operation, and determining the next candidate particle of the candidate particle targeted by the search operation as the candidate particle targeted by the next search operation of the search operation; If not, the next candidate particle of the candidate particle targeted by the search operation is determined as the candidate particle targeted by the next search operation of the search operation, wherein the next candidate particle of the candidate particle targeted by the search operation is indicated by the order.

8. The method according to claim 7, wherein: The method further comprises: A circular area having a center point as the position of the candidate particle targeted by the search operation in the vehicle body coordinate system and having a preset second radius as a radius is determined as a target area related to the candidate particle targeted by the search operation.

9. An information acquisition device, characterized in that: The device comprises: an acquisition unit, configured to acquire a segmentation result of a target environment image captured by a camera of a target vehicle and a detected three-dimensional lane line point cloud of a target lane line on a target road on which the target vehicle is traveling, wherein the segmentation result indicates a plurality of lane line pixels of the target lane line, and the detected three-dimensional lane line point cloud is detected based on the target environment image; a determining unit, configured to determine whether a target pitch angle associated with a target environment image exists based on a current pose of the camera, the segmentation result, and the detected three-dimensional lane line point cloud, wherein an error between a target projected three-dimensional lane line point cloud obtained by inversely projecting the plurality of lane line pixel points into a body coordinate system of a target vehicle using a target corrected pose corresponding to the target pitch angle and the detected three-dimensional lane line point cloud is less than an error threshold, and the target corrected pose is obtained by replacing a current pitch angle in the current pose of the camera with the target pitch angle; A conversion unit is used to, when it is determined that the target pitch angle exists, use the target corrected posture to transform the position of the cone barrel in the target environment image in the image coordinate system of the target environment image into the corrected position of the cone barrel in the vehicle body coordinate system.

10. A computer device installed on a vehicle, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 8 by executing the computer instructions.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 8.

12. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the method according to any one of claims 1 to 8.

Images