Anti-collision method, device, storage medium and electronic equipment

Abstract

The present application relates to the technical field of safety warning, and specifically to an anti-collision method, device and electronic equipment, the method comprising: acquiring image data of a target area; performing target detection on the image data, so as to accurately acquire information of a first target to be detected and information of a second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected; judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected; if it is determined that there is a collision risk between the first target to be detected and the second target to be detected, performing an anti-collision warning according to the target type of the second target to be detected, thereby ensuring that an effective anti-collision warning can be achieved.

Description

Technical Field

[0001] The present application relates to the field of safety warning technology, and in particular to an anti-collision method, device, storage medium and electronic equipment. Background Art

[0002] In today's industrial production scenarios and even daily life scenarios, more and more unmanned vehicles using unmanned driving technology are used to transport items. Unmanned vehicles generally move according to routes planned in advance by the operator.

[0003] In the current common unmanned vehicle, in order to avoid collisions, the unmanned vehicle will keep issuing warning sounds to ensure that pedestrians around are aware that an unmanned vehicle is approaching. However, this method of avoiding collisions by continuously issuing sounds will cause problems such as noise pollution in this scenario. Noise pollution greatly affects the working status of production personnel, thereby affecting production efficiency. Summary of the invention

[0004] In response to the above problems, the present application provides an anti-collision method, device, storage medium and electronic device.

[0005] In a first aspect, the present application provides a collision avoidance method, the method comprising:

[0006] Acquire image data of the target area;

[0007] Performing target detection on the image data to obtain information of a first target to be detected and information of at least one second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected;

[0008] Determining whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected;

[0009] If it is determined that there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected.

[0010] In the above implementation, target detection is performed on the image data of the target area, and the information of the first target to be detected and the information of the second target to be detected can be accurately obtained, and according to the target position in the information, it can be judged whether there is a collision risk between the first target to be detected and the second target to be detected. When there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected, thereby achieving a more effective anti-collision warning.

[0011] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision method, acquiring image data of the target area includes:

[0012] Different images of the target area taken by at least two camera devices are acquired; wherein the at least two camera devices are at different positions.

[0013] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision method, performing target detection on the image data to obtain information of a first target to be detected and information of at least one second target to be detected includes:

[0014] Acquire two images captured by any two different camera devices in the image data;

[0015] Target detection is performed on the two images respectively, and information of a first target to be detected and information of at least one second target to be detected in the two images are obtained respectively.

[0016] In the above implementation, any two images taken by different camera devices are taken from different shooting points, so the description angles of the target area of ​​the images obtained are different. Performing target detection on the two different images separately can more accurately determine the target to be detected.

[0017] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision method, performing target detection on the two images respectively, and obtaining information of a first target to be detected and information of at least one second target to be detected in the two images respectively, includes:

[0018] Obtaining a transformation matrix between the two images, the two images comprising a first image and a second image;

[0019] Acquire the target position and target type of the first target to be detected, and the target position and target type of the second target to be detected in the first image; acquire the target position and target type of the first target to be detected, and the target position and target type of the second target to be detected in the second image;

[0020] Transforming the target position of the first target to be detected and the target position of the second target to be detected in the first image according to the transformation matrix to obtain a first expected target position and a second expected target position of the first target to be detected and the second target to be detected in the second image;

[0021] Compare the first expected target position and the target type of the first target to be detected in the first image with the target position and target type of the first target to be detected in the second image to determine whether the first targets to be detected in the two images are the same target to be detected

[0022] The second expected target position and the target type of the second target to be detected in the first image are compared with the target position and the target type of the second target to be detected in the second image to determine whether the second targets to be detected in the two images are the same target to be detected.

[0023] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision method, the step of obtaining the target position of the first target to be detected and the target position of the second target to be detected in the first image; and the step of obtaining the target position of the first target to be detected and the target position of the second target to be detected in the second image include:

[0024] Marking each target to be detected identified in the two images with a directed bounding box to obtain a bounding box corresponding to each target to be detected;

[0025] The coordinate information of the bounding box corresponding to each target to be detected is obtained as the target position of each target to be detected.

[0026] According to an embodiment of the present application, optionally, in the above-mentioned collision avoidance method, judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected includes:

[0027] determining a first distance according to a target position of the first target to be detected and a target position of the second target to be detected in the first image;

[0028] Determine whether the first distance is less than a first preset threshold value, and if so, determine the second distance according to the target position of the first target to be detected and the target position of the second target to be detected in the second image, wherein the first target to be detected in the second image and the first target to be detected in the first image are the same target to be detected, and the second target to be detected in the second image and the second target to be detected in the first image are the same target to be detected;

[0029] It is determined whether the second distance is less than a second preset threshold, and if so, it is determined that there is a collision risk between the first target to be detected and the second target to be detected.

[0030] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision method, the target type of the first target to be detected includes inanimate moving targets, and the target type of the second target to be detected includes inanimate moving targets, living targets and inanimate fixed targets.

[0031] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision method, performing anti-collision warning according to the target type of the second target to be detected includes:

[0032] If the target type of the second target to be detected is a living target, controlling the first target to be detected to decelerate and issuing a collision warning prompt;

[0033] If the target type of the second target to be detected is a non-living moving target or a non-living fixed target, the first target to be detected is controlled to stop running, and an anti-collision processing reminder is sent to the management end.

[0034] According to an embodiment of the present application, optionally, in the above anti-collision method, after performing an anti-collision warning according to the target type of the second target to be detected, the method further includes:

[0035] Store the recorded data corresponding to each anti-collision warning;

[0036] The recorded data is analyzed to generate a management strategy.

[0037] In a second aspect, the present application further provides an anti-collision device, the device comprising:

[0038] An image data acquisition module, used to acquire image data of a target area;

[0039] A target detection module, used to perform target detection on the image data, and obtain information of a first target to be detected and information of a second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected;

[0040] a risk judgment module, used for judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected;

[0041] The anti-collision warning module is used to perform an anti-collision warning according to the target type of the second target to be detected if it is determined that there is a collision risk between the first target to be detected and the second target to be detected.

[0042] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the image data acquisition module includes:

[0043] The image acquisition unit is used to acquire different images of the target area taken by at least two camera devices; wherein the positions of the at least two camera devices are different.

[0044] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the target detection module includes:

[0045] An image acquisition unit, used to acquire two images captured by any two different camera devices in the image data;

[0046] The to-be-detected target determination unit is used to perform target detection on the two images respectively, and obtain information of a first to-be-detected target and information of at least one second to-be-detected target in the two images respectively.

[0047] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the target determination unit to be detected includes:

[0048] A transformation matrix acquisition subunit, used to acquire a transformation matrix between the two images, the two images comprising a first image and a second image;

[0049] an information acquisition subunit, configured to acquire a target position and a target type of the first target to be detected, and a target position and a target type of the second target to be detected in the first image; and to acquire a target position and a target type of the first target to be detected, and a target position and a target type of the second target to be detected in the second image;

[0050] a position change subunit, configured to transform the target position of the first target to be detected and the target position of the second target to be detected in the first image according to the transformation matrix, so as to obtain a first expected target position and a second expected target position of the first target to be detected and the second target to be detected in the second image;

[0051] a first target to be detected information acquisition subunit, configured to compare the first expected target position and the target type of the first target to be detected in the first image with the target position and the target type of the first target to be detected in the second image, so as to determine whether the first targets to be detected in the two images are the same target to be detected;

[0052] The second target to be detected information acquisition subunit is used to compare the second expected target position and the target type of the second target to be detected in the first image with the target position and the target type of the second target to be detected in the second image to determine whether the second targets to be detected in the two images are the same target to be detected.

[0053] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the information acquisition subunit includes:

[0054] The bounding box marking subunit is used to mark each target to be detected identified in the two images with a directional bounding box to obtain a bounding box corresponding to each target to be detected;

[0055] The target position determination subunit is used to obtain the coordinate information of the bounding box corresponding to each target to be detected as the target position of each target to be detected.

[0056] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the risk judgment module includes:

[0057] A first distance determining unit, configured to determine a first distance according to a target position of the first target to be detected and a target position of the second target to be detected in the first image;

[0058] A second distance determination unit, used to determine whether the first distance is less than a first preset threshold, and if so, determine the second distance according to the target position of the first target to be detected and the target position of the second target to be detected in the second image, wherein the first target to be detected in the second image and the first target to be detected in the first image are the same target to be detected, and the second target to be detected in the second image and the second target to be detected in the first image are the same target to be detected;

[0059] The collision risk determination unit is used to determine whether the second distance is less than a second preset threshold, and if so, determine that there is a collision risk between the first target to be detected and the second target to be detected.

[0060] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the target type of the first target to be detected includes inanimate moving targets, and the target type of the second target to be detected includes inanimate moving targets, living targets and inanimate fixed targets.

[0061] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the anti-collision warning module includes:

[0062] a first warning unit, configured to control the first target to be detected to decelerate and issue a collision warning prompt if the target type of the second target to be detected is a living target;

[0063] The second early warning unit is used to control the first target to be detected to stop running and send an anti-collision processing reminder to the management end if the target type of the second target to be detected is a non-living moving target or a non-living fixed target.

[0064] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the device further includes:

[0065] A recording module, used to store the recorded data corresponding to each anti-collision warning;

[0066] The analysis module is used to analyze the recorded data and generate a management strategy.

[0067] In a third aspect, the present application provides a storage medium, which stores a computer program that can be executed by one or more processors and can be used to implement the anti-collision method as described above.

[0068] In a fourth aspect, the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the above-mentioned anti-collision method is performed.

[0069] Compared with the prior art, one or more embodiments of the above scheme may have the following advantages or beneficial effects:

[0070] The present application provides an anti-collision method, device, storage medium and electronic device, the method comprising acquiring image data of a target area; performing target detection on the image data to acquire information of a first target to be detected and information of at least a second target to be detected, wherein the information of the first target to be detected and the information of the at least second target to be detected both include at least the target type and target position of the target to be detected; judging whether there is a collision risk between the first target to be detected and the at least one second target to be detected according to the target position of the first target to be detected and the target position of the at least one second target to be detected; if it is determined that there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected. Performing target detection on the image data of the target area can accurately acquire information of the first target to be detected and information of the second target to be detected, and judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position in the information. When there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected, thereby achieving a more effective anti-collision warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Hereinafter, the present application will be described in more detail based on embodiments and with reference to the accompanying drawings.

[0072] Figure 1 A flowchart of an anti-collision method provided in Example 1 of the present application.

[0073] Figure 2 This is a structural block diagram of an anti-collision device provided in Example 5 of the present application.

[0074] Figure 3 A connection block diagram of an electronic device provided in Example 7 of the present application.

[0075] In the drawings, the same reference numerals are used for the same components, and the drawings are not drawn to scale. DETAILED DESCRIPTION

[0076] The following will describe the implementation methods of the present application in detail with reference to the accompanying drawings and embodiments, so that the implementation process of how the present application applies technical means to solve technical problems and achieve corresponding technical effects can be fully understood and implemented accordingly. The embodiments of the present application and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present application.

[0077] Embodiment 1

[0078] The present invention provides an anti-collision method, please refer to Figure 1 , the method comprises the following steps:

[0079] Step S110: Acquire image data of the target area.

[0080] When acquiring image data of a target area, different images of the target area taken by at least two camera devices may be first acquired; wherein the at least two camera devices are at different positions.

[0081] Among them, one of the at least two camera devices can shoot in one direction of the target area, and the other camera device can shoot in another direction of the target area, and the two shooting directions are different. The specific shooting direction of the camera device can be determined according to the actual on-site conditions of the target area. For example, one camera device can be set to the east side of the target area and shoot in the west direction, while the other camera device can be set to the south side of the target area and shoot in the north direction.

[0082] Step S120: performing target detection on the image data to obtain information of a first target to be detected and information of at least one second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected.

[0083] When performing target detection on the image data, target recognition can be performed on each image in the image data based on deep learning to identify all the targets to be detected in each image, and determine the target position and target type corresponding to each target to be detected. For example, by performing YOLOv3 deep learning on the acquired image data to perform target detection, unmanned vehicles, pedestrians, temporary materials, etc. in the target area can be distinguished, and fixed materials in the target area can be processed as background. It can be understood that when performing target recognition on each image in the image data, other methods can also be used, which are not limited here.

[0084] Among them, the target type of the first target to be detected includes inanimate moving targets, and the target type of the second target to be detected includes inanimate moving targets, living targets and inanimate fixed targets. Inanimate moving targets can be inanimate and can move according to a certain rule or command. Inanimate moving targets can be different in different scenes. For example, in a warehouse where unmanned vehicles are used to transport goods, the inanimate moving target is an unmanned vehicle. In an area where an automatic cleaning robot is used to clean the floor, the inanimate moving target is the automatic cleaning robot. In addition, it can be understood that multiple different inanimate moving targets can appear in the same scene. For example, in a warehouse where unmanned vehicles are used to transport goods, an automatic cleaning robot can be used to clean the floor of the warehouse. At this time, the inanimate moving target refers to the unmanned vehicle and the automatic cleaning robot. Inanimate fixed targets are inanimate and cannot move according to a certain rule or command. For example, in a warehouse, inanimate fixed targets can be goods that need to be transported, goods that fall during transportation, and items temporarily placed in the warehouse. A living target can be an individual target that has a living form and can respond to external stimuli.

[0085] Step S130: determining whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected.

[0086] Step S140: If it is determined that there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected.

[0087] After performing target detection on the image data, all targets to be detected and related information in the image data can be obtained, all targets to be detected include a first target to be detected and a second target to be detected, and related information of the targets to be detected includes a target type and a target position. After identifying the target type and target position of the first target to be detected and the second target to be detected, it can be judged whether there is a risk of collision between the first target to be detected and the second target to be detected based on the target position of the first target to be detected and the target position of the second target to be detected. Specifically, the distance between the first target to be detected and the second target to be detected can be calculated based on the target position of the first target to be detected and the target position of the second target to be detected, and then based on the distance, it can be judged whether there is a risk of collision between the first target to be detected and the second target to be detected.

[0088] When it is determined that there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning can be performed according to the target type of the first target to be detected. If the target type of the second target to be detected is a living body target, the first target to be detected is controlled to slow down and issue a collision warning prompt. When the living body target hears the collision warning prompt issued by the first detection target, it can avoid it, thereby achieving anti-collision. For example, in a certain application scenario, the first target to be detected is an unmanned car, and the second target to be detected is a pedestrian. When it is determined that the unmanned car and the pedestrian are in danger of collision, since the target type of the second target to be detected is a living body target, a deceleration signal can be sent to the unmanned car, and the unmanned car can be made to issue a prompt sound such as "please pay attention" to remind the pedestrian to pay attention to safety. Among them, the first detection target slows down and travels, which can effectively avoid the collision caused by the second target to be detected not having time to avoid. If the target type of the second target to be detected is a non-living moving target or a fixed obstacle target, the first target to be detected is controlled to stop running, and an anti-collision processing reminder is sent to the management end. Non-living moving targets or fixed obstacle targets cannot respond to external stimuli. Therefore, the first target to be detected can be directly controlled to stop running to avoid collision, and an anti-collision processing reminder can be sent to the management end to remind the corresponding staff to move the second target to be detected as soon as possible, or remind the corresponding staff to move the first target to be detected again. For example, in a certain application scenario, the first target to be detected is an unmanned car, and the second target to be detected is another unmanned car. When it is determined that the unmanned car and another unmanned car are in danger of collision, a stop command can be issued to the unmanned car, and the system administrator can be notified to handle it. The administrator can re-plan the route of the unmanned car or re-plan the route of another unmanned car according to the actual situation. If the first target to be detected in the application scenario is an unmanned car, and the second target to be detected is a material temporarily placed by the staff, when it is determined that the unmanned car and the temporarily placed materials are in danger of collision, a stop command can be issued to the unmanned car, and the system administrator can be notified to handle it. The administrator can re-plan the route of the unmanned car or move the temporarily placed materials away according to the actual situation.

[0089] In summary, the present application provides an anti-collision method, which includes: acquiring image data of a target area; performing target detection on the image data to obtain information of a first target to be detected and information of at least one second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least the target type and target position of the target to be detected; judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected; if it is determined that there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected. Performing target detection on the image data of the target area can accurately obtain the information of the first target to be detected and the information of the second target to be detected, and judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position in the information. When there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected, thereby achieving a more effective anti-collision warning.

[0090] Embodiment 2

[0091] Based on the first embodiment, this embodiment illustrates the method in the first embodiment through a specific implementation case.

[0092] When performing target detection on the image data to obtain information about a first target to be detected and information about at least one second target to be detected, two images taken by any two different camera devices in the image data can be first obtained; then target detection is performed on the two images respectively, and information about the first target to be detected and information about at least one second target to be detected in the two images are respectively obtained.

[0093] In the above implementation, any two images taken by different camera devices are taken from different shooting points, so the description angles of the target area of ​​the images obtained are different. Performing target detection on the two different images separately can more accurately determine the target to be detected.

[0094] Specifically, when target detection is performed on the two images respectively, and information of the first target to be detected and information of at least one second target to be detected in the two images are obtained respectively, the following process is included. First, the transformation matrix between the two images is obtained, and the two images include the first image and the second image; then the target position and target type of the first target to be detected, and the target position and target type of the second target to be detected in the first image are obtained; and the target position and target type of the first target to be detected, and the target position and target type of the second target to be detected in the second image are obtained. Then, the target position of the first target to be detected in the first image and the target position of the second target to be detected are transformed according to the transformation matrix to obtain the first expected target position and the second expected target position of the first target to be detected and the second target to be detected in the second image. The first expected target position and the target type of the first target to be detected in the first image are compared with the target position and target type of the first target to be detected in the second image to determine whether the first target to be detected in the two images is the same target to be detected. The second expected target position and the target type of the second target to be detected in the first image are compared with the target position and the target type of the second target to be detected in the second image to determine whether the second targets to be detected in the two images are the same target to be detected.

[0095] In the above implementation, the position of the target to be detected and the type of the target to be detected in one image of the target area can be identified, and then the position of the target to be detected is converted into the expected target position according to the transformation matrix between the two different images. The expected target position is then compared with the actual target position of the target to be detected in another image to determine whether the targets to be detected in the images taken by two different camera devices are the same target to be detected, thereby ensuring accurate identification of the target to be detected.

[0096] The transformation matrix can be obtained by analyzing the change rules between the positions of known targets in two images. For example, if there is an unmanned vehicle, a pedestrian, and an obstacle in the actual scene, one camera is C1, and the other camera is C2. Camera C1 detects the position of the unmanned vehicle as A1, the position of the pedestrian as B1, and the position of the obstacle as D1. Camera C2 detects the position of the unmanned vehicle as A2, the position of the pedestrian as B2, and the position of the obstacle as D2. According to the transformation matrix between the two images, the correspondence between the two images taken by camera C1 and camera C2 can be determined. If the coordinates of A1 are {A11, A12, A13, A14}, then according to the known transformation matrix H (the transformation matrix H can be obtained in advance through the position coordinates of a marker in the image taken by C1 and the position coordinates in the image taken by C2), the position coordinates of A1 in C2 {A15, A16, A17, A18} can be obtained. Because they both represent the same unmanned vehicle in C1 and C2, the transformed {A15, A16, A17, A18} and the coordinates of A2 {A21, A22, A23, A24} are mostly overlapped. Therefore, the position of the unmanned vehicle in the image taken by C1 can be converted into the expected position of the unmanned vehicle in the image taken by C2, and then the overlap between the expected position and the actual target position of the unmanned vehicle in the image taken by C2 can be used to determine whether the targets to be detected in the two images are the same target to be detected. When the position of the unmanned vehicle in the image captured by C1 is converted to the expected position of the unmanned vehicle in the image captured by C2 and the overlap degree of the actual target position of the unmanned vehicle in the image captured by C2 is greater than the preset threshold, it can be further determined whether the target type of the unmanned vehicle in the image captured by C1 is consistent with the target type of the unmanned vehicle in the image captured by C2. When the expected target position of the target to be detected in the image captured by C2 and the target type in the image captured by C1 are the same as the actual target position and target type of the target to be detected in the image captured by C2, it can be determined that the targets to be detected in the images captured by two different camera devices are the same target to be detected.

[0097] Embodiment 3

[0098] Based on the second embodiment, this embodiment illustrates the method in the first embodiment through a specific implementation case.

[0099] When obtaining the target position of the first target to be detected and the target position of the second target to be detected in the first image, or when obtaining the target position of the first target to be detected and the target position of the second target to be detected in the second image, each target to be detected identified in the two images can be first marked with a directed bounding box to obtain the bounding box corresponding to each target to be detected, and then the coordinate information of the bounding box corresponding to each target to be detected is obtained as the target position of each target to be detected.

[0100] Suppose there is an unmanned car, a pedestrian, and an obstacle in the actual scene. The targets detected by deep learning in the two cameras are marked with directed bounding boxes, that is, the bounding box is always kept as the minimum circumscribed rectangle of the target area. Then the unmanned car detected by camera C1 is marked with a directed bounding box, and the coordinates of the bounding box corresponding to the unmanned car are the target position A1 of the unmanned car. The pedestrian detected by camera C1 is marked with a directed bounding box, and the coordinates of the bounding box corresponding to the pedestrian are the target position B1 of the pedestrian. The obstacle detected by camera C1 is marked with a directed bounding box, and the coordinates of the bounding box corresponding to the obstacle are the target position D1 of the obstacle. Then the unmanned car detected by camera C2 is marked with a directed bounding box, and the coordinates of the bounding box corresponding to the unmanned car are the target position A2 of the unmanned car. The pedestrian detected by camera C2 is marked with a directed bounding box, and the coordinates of the bounding box corresponding to the pedestrian are the target position B2 of the pedestrian. The obstacle detected by the camera C2 is marked with a directed bounding box, and the coordinates of the bounding box corresponding to the obstacle are the target position D2 of the obstacle.

[0101] The step of judging whether there is a risk of collision between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected may include the following process. First, determine a first distance according to the target position of the first target to be detected and the target position of the second target to be detected in the first image; judge whether the first distance is less than a first preset threshold value, and if so, determine a second distance according to the target position of the first target to be detected and the target position of the second target to be detected in the second image, wherein the first target to be detected in the second image and the first target to be detected in the first image are the same target to be detected, and the second target to be detected in the second image and the second target to be detected in the first image are the same target to be detected. Judge whether the second distance is less than the second preset threshold value, and if so, determine that there is a risk of collision between the first target to be detected and the second target to be detected.

[0102] When the first distance between the first target to be detected and the second target to be detected in the first image is less than the first preset threshold, it means that the first image shows that there is a risk of collision between the first target to be detected and the second target to be detected. At this time, continue to determine the second distance based on the target position of the first target to be detected and the target position of the second target to be detected in the second image. If the second distance is still less than the second preset threshold, it means that there is a risk of collision between the first target to be detected and the second target to be detected. If the second distance is greater than the second preset threshold, it means that the first target to be detected and the second target to be detected in the first image are only caused by visual misalignment, and there is actually no risk of collision. It can be understood that the first preset threshold and the second preset threshold can be the same or different, and the setting of the specific value needs to be set according to the actual situation.

[0103] Specifically, if two objects collide, their directed bounding boxes will overlap. For example, if the bounding box of the unmanned car position A1 and the bounding box of the pedestrian position B1 overlap in the image acquired by camera C1, then one possibility is that the unmanned car and the pedestrian collide, and another possibility is that there is an occlusion between the unmanned car and the pedestrian, and there is no actual contact. In order to further determine whether the two have contacted, it is possible to check whether the bounding boxes corresponding to the unmanned car position A2 and the pedestrian position B2 in the image captured by camera C2 overlap. If the bounding boxes of A2 and B2 also overlap, it can be determined that the unmanned car and the pedestrian have contacted and collided. If the bounding boxes of A2 and B2 do not overlap, it can be determined that the unmanned car and the pedestrian are just blocked and not actually in contact. Therefore, it is possible to judge whether there is a risk of collision between the two targets to be detected based on the distance between the targets to be detected in the image captured by camera C1 and the image captured by camera C2. When the distance between two targets to be detected in the image taken by camera C1 is particularly close, and the distance between the same two targets to be detected in the image taken by camera C2 is also particularly close, it can be determined that the two targets to be detected are in danger of contact, and an early warning can be issued.

[0104] Embodiment 4

[0105] Based on the first embodiment, this embodiment illustrates the method in the first embodiment through a specific implementation case.

[0106] After the anti-collision warning is performed according to the target type of the second target to be detected, the recorded data corresponding to each anti-collision warning can also be stored; the recorded data can be analyzed to generate a management strategy.

[0107] The data corresponding to each anti-collision warning is recorded and stored, and then analyzed. For example, the nature, time and location of the event each time the anti-collision warning is issued can be analyzed. It can be analyzed that there are more pedestrians in which time period and which locations are often occupied by obstacles. Production management strategies based on the above analysis results can effectively help system administrators manage the target area. For example, the route of the unmanned vehicle can be replanned, and the speed of the unmanned vehicle can be reduced during the time period with more pedestrians, so as to increase the safety of unmanned vehicle transportation. In addition, it is convenient for managers to customize corresponding management strategies according to accident situations. It can be understood that the scenarios in which this anti-collision method is applied can include AGV production systems, conventional logistics and transportation, and transportation management.

[0108] Embodiment 5

[0109] Please see Figure 2 , the present application provides an anti-collision device 200, the device comprising:

[0110] An image data acquisition module 210 is used to acquire image data of a target area;

[0111] The target detection module 220 is used to perform target detection on the image data to obtain information about a first target to be detected and information about a second target to be detected, wherein the information about the first target to be detected and the information about the second target to be detected both include at least a target type and a target position of the target to be detected;

[0112] A risk judgment module 230, configured to judge whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected;

[0113] The anti-collision warning module 240 is configured to, if it is determined that there is a collision risk between the first target to be detected and the second target to be detected, issue an anti-collision warning according to the target type of the second target to be detected.

[0114] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the image data acquisition module 210 includes:

[0115] The image acquisition unit is used to acquire different images of the target area taken by at least two camera devices; wherein the positions of the at least two camera devices are different.

[0116] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the target detection module 220 includes:

[0117] An image acquisition unit, used to acquire two images captured by any two different camera devices in the image data;

[0118] The to-be-detected target determination unit is used to perform target detection on the two images respectively, and obtain information of a first to-be-detected target and information of at least one second to-be-detected target in the two images respectively.

[0119] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the target determination unit to be detected includes:

[0120] A transformation matrix acquisition subunit, used to acquire a transformation matrix between the two images, the two images comprising a first image and a second image;

[0121] an information acquisition subunit, configured to acquire a target position and a target type of the first target to be detected, and a target position and a target type of the second target to be detected in the first image; and to acquire a target position and a target type of the first target to be detected, and a target position and a target type of the second target to be detected in the second image;

[0122] a position change subunit, configured to transform the target position of the first target to be detected and the target position of the second target to be detected in the first image according to the transformation matrix, so as to obtain a first expected target position and a second expected target position of the first target to be detected and the second target to be detected in the second image;

[0123] a first target to be detected information acquisition subunit, configured to compare the first expected target position and the target type of the first target to be detected in the first image with the target position and the target type of the first target to be detected in the second image, so as to determine whether the first targets to be detected in the two images are the same target to be detected;

[0124] The second target to be detected information acquisition subunit is used to compare the second expected target position and the target type of the second target to be detected in the first image with the target position and the target type of the second target to be detected in the second image to determine whether the second targets to be detected in the two images are the same target to be detected.

[0125] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the information acquisition subunit includes:

[0126] The bounding box marking subunit is used to mark each target to be detected identified in the two images with a directional bounding box to obtain a bounding box corresponding to each target to be detected;

[0127] The target position determination subunit is used to obtain the coordinate information of the bounding box corresponding to each target to be detected as the target position of each target to be detected.

[0128] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the risk judgment module 230 includes:

[0129] A first distance determining unit, configured to determine a first distance according to a target position of the first target to be detected and a target position of the second target to be detected in the first image;

[0130] A second distance determination unit, used to determine whether the first distance is less than a first preset threshold, and if so, determine the second distance according to the target position of the first target to be detected and the target position of the second target to be detected in the second image, wherein the first target to be detected in the second image and the first target to be detected in the first image are the same target to be detected, and the second target to be detected in the second image and the second target to be detected in the first image are the same target to be detected;

[0131] The collision risk determination unit is used to determine whether the second distance is less than a second preset threshold, and if so, determine that there is a collision risk between the first target to be detected and the second target to be detected.

[0132] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the target type of the first target to be detected includes inanimate moving targets, and the target type of the second target to be detected includes inanimate moving targets, living targets and inanimate fixed targets.

[0133] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the anti-collision warning module 240 includes:

[0134] a first warning unit, configured to control the first target to be detected to decelerate and issue a collision warning prompt if the target type of the second target to be detected is a living target;

[0135] The second early warning unit is used to control the first target to be detected to stop running and send an anti-collision processing reminder to the management end if the target type of the second target to be detected is a non-living moving target or a non-living fixed target.

[0136] According to an embodiment of the present application, optionally, in the above-mentioned anti-collision device, the device further includes:

[0137] A recording module, used to store the recorded data corresponding to each anti-collision warning;

[0138] The analysis module is used to analyze the recorded data and generate a management strategy.

[0139] In summary, the present application provides an anti-collision device, including: an image data acquisition module 210, used to acquire image data of a target area; a target detection module 220, used to perform target detection on the image data, and acquire information of a first target to be detected and information of a second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected; a risk judgment module 230, used to judge whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected; an anti-collision warning module 240, used to perform an anti-collision warning according to the target type of the second target to be detected if it is determined that there is a collision risk between the first target to be detected and the second target to be detected. By performing target detection on the image data of the target area, the information of the first target to be detected and the information of the second target to be detected can be accurately acquired, and according to the target position in the information, it can be judged whether there is a collision risk between the first target to be detected and the second target to be detected. When there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected, thereby achieving a more effective anti-collision warning.

[0140] Embodiment 6

[0141] This embodiment also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., on which a computer program is stored. When the computer program is executed by a processor, the method steps in the above embodiment can be implemented, and this embodiment will not be repeated here.

[0142] Embodiment 7

[0143] The present application embodiment provides an electronic device, which may be a mobile phone, a computer or a tablet computer, etc., including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the anti-collision method described in the first embodiment is implemented. It can be understood that Figure 3 As shown, the electronic device 300 may further include: a processor 301 , a memory 302 , a multimedia component 303 , an input / output (I / O) interface 304 , and a communication component 305 .

[0144] The processor 301 is used to execute all or part of the steps in the anti-collision method in Embodiment 1. The memory 302 is used to store various types of data, which may include instructions of any application or method in the electronic device, and data related to the application.

[0145] The processor 301 can be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components, and is used to execute the anti-collision method in the above-mentioned embodiment 1.

[0146] The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.

[0147] The multimedia component 303 may include a screen and an audio component, the screen may be a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone, and the microphone is used to receive external audio signals. The received audio signal may be further stored in a memory or sent via a communication component. The audio component also includes at least one speaker for outputting audio signals.

[0148] The I / O interface 304 provides an interface between the processor 301 and other interface modules, which may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons.

[0149] The communication component 305 is used for wired or wireless communication between the electronic device 300 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 305 can include: Wi-Fi module, Bluetooth module, NFC module.

[0150] In summary, the present application provides an anti-collision method, device, storage medium and electronic device, the method comprising acquiring image data of a target area; performing target detection on the image data to acquire information of a first target to be detected and information of a second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected; judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected; if it is determined that there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected. Performing target detection on the image data of the target area can accurately acquire information of the first target to be detected and information of the second target to be detected, and judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position in the information. When there is a collision risk between the first target to be detected and the second target to be detected, an anti-collision warning is performed according to the target type of the second target to be detected, thereby achieving a more effective anti-collision warning.

[0151] In the several embodiments provided in the present application, it should be understood that the disclosed system and method can also be implemented in other ways. The above described system and method embodiments are merely illustrative.

[0152] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0153] Although the implementation methods disclosed in this application are as above, the contents described are only implementation methods adopted for facilitating the understanding of this application, and are not intended to limit this application. Any technician in the technical field to which this application belongs can make any modifications and changes in the form and details of implementation without departing from the spirit and scope disclosed in this application, but the scope of patent protection of this application shall still be based on the scope defined in the attached claims.

Claims

1. A collision prevention method, characterized in that: The method comprises: Acquire image data of the target area; Performing target detection on the image data to obtain information of a first target to be detected and information of at least one second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected; Determining whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected; If it is determined that there is a collision risk between the first target to be detected and the second target to be detected, performing an anti-collision warning according to the target type of the second target to be detected; The step of performing target detection on the image data to obtain information of a first target to be detected and information of at least one second target to be detected includes: Acquire two images captured by any two different camera devices in the image data; Performing target detection on the two images respectively, and obtaining information of a first target to be detected and information of at least one second target to be detected in the two images respectively; Performing target detection on the two images respectively, and obtaining information of a first target to be detected and information of at least one second target to be detected in the two images respectively, comprises: Obtaining a transformation matrix between the two images, the two images comprising a first image and a second image; Acquire the target position and target type of the first target to be detected, and the target position and target type of the second target to be detected in the first image; acquire the target position and target type of the first target to be detected, and the target position and target type of the second target to be detected in the second image; Transforming the target position of the first target to be detected and the target position of the second target to be detected in the first image according to the transformation matrix to obtain a first expected target position and a second expected target position of the first target to be detected and the second target to be detected in the second image; Comparing the first expected target position and the target type of the first target to be detected in the first image with the target position and the target type of the first target to be detected in the second image to determine whether the first targets to be detected in the two images are the same target to be detected; Comparing the second expected target position and the target type of the second target to be detected in the first image with the target position and the target type of the second target to be detected in the second image to determine whether the second targets to be detected in the two images are the same target to be detected; Judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected, includes: determining a first distance according to a target position of the first target to be detected and a target position of the second target to be detected in the first image; Determine whether the first distance is less than a first preset threshold value, and if so, determine the second distance according to the target position of the first target to be detected and the target position of the second target to be detected in the second image, wherein the first target to be detected in the second image and the first target to be detected in the first image are the same target to be detected, and the second target to be detected in the second image and the second target to be detected in the first image are the same target to be detected; It is determined whether the second distance is less than a second preset threshold, and if so, it is determined that there is a collision risk between the first target to be detected and the second target to be detected.

2. The method according to claim 1, characterized in that Obtain image data of the target area, including: Different images of the target area taken by at least two camera devices are acquired; wherein the at least two camera devices are at different positions.

3. The method according to claim 1, characterized in that acquiring a target position of the first target to be detected and a target position of the second target to be detected in the first image; Acquiring the target position of the first target to be detected and the target position of the second target to be detected in the second image includes: Marking each target to be detected identified in the two images with a directed bounding box to obtain a bounding box corresponding to each target to be detected; The coordinate information of the bounding box corresponding to each target to be detected is obtained as the target position of each target to be detected.

4. The method according to claim 1, characterized in that The target type of the first target to be detected includes a non-living moving target, and the target type of the second target to be detected includes a non-living moving target, a living target, and a non-living fixed target.

5. The method according to claim 1, characterized in that Performing an anti-collision warning according to the target type of the second target to be detected includes: If the target type of the second target to be detected is a living target, controlling the first target to be detected to decelerate and issuing a collision warning prompt; If the target type of the second target to be detected is a non-living moving target or a non-living fixed target, the first target to be detected is controlled to stop running, and an anti-collision processing reminder is sent to the management end.

6. The method according to claim 1, characterized in that After performing anti-collision warning according to the target type of the second target to be detected, the method further includes: Store the recorded data corresponding to each anti-collision warning; The recorded data is analyzed to generate a management strategy.

7. An anti-collision device, using the anti-collision method according to any one of claims 1 to 6, characterized in that: The device comprises: An image data acquisition module, used to acquire image data of a target area; A target detection module, used to perform target detection on the image data, and obtain information of a first target to be detected and information of at least one second target to be detected, wherein the information of the first target to be detected and the information of the second target to be detected both include at least a target type and a target position of the target to be detected; a risk judgment module, used for judging whether there is a collision risk between the first target to be detected and the second target to be detected according to the target position of the first target to be detected and the target position of the second target to be detected; The anti-collision warning module is used to perform an anti-collision warning according to the target type of the second target to be detected if it is determined that there is a collision risk between the first target to be detected and the second target to be detected.

8. A storage medium, characterized in that: The computer program stored in the storage medium is used to implement the method according to any one of claims 1 to 6 when executed by one or more processors.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the method according to any one of claims 1 to 6 is executed.

Images