Automatic water sprinkling starting and stopping method and system based on visual identification

The automatic start-stop watering method using visual recognition in sanitation vehicles addresses driver distraction and water wastage by accurately identifying traffic participants and controlling watering operations, enhancing safety and efficiency.

CN119068460APending Publication Date: 2024-12-03CHENGDU YIWEI NEW ENERGY VEHICLE CO LTD

Patent Information

Application Number
CN202411148733.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Existing water-based sanitation vehicles face challenges in automatically adjusting watering operations to avoid distracting drivers, leading to potential accidents and water wastage, especially when pedestrians or vehicles are present, and require manual intervention which is inefficient.

Method used

An automatic start-stop watering method using visual recognition, employing monocular cameras to identify traffic participants and control watering based on target types and vehicle states, including pedestrians, non-motor vehicles, traffic lights, and motor vehicles, with control logic determining watering ranges and states.

Benefits of technology

The method accurately identifies and classifies traffic participants, enabling intelligent watering control, reducing driver distraction, preventing accidents, and conserving water by avoiding unnecessary spraying.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068460A_ABST
    Figure CN119068460A_ABST
Patent Text Reader

Abstract

The invention provides an automatic sprinkling starting and stopping method and system based on visual identification, and relates to the technical field of sprinkling equipment, and the automatic sprinkling starting and stopping method comprises the steps: obtaining images of the front and / or side of a sprinkling truck in the driving process; obtaining a target type of a traffic participation object according to the image; starting a corresponding control logic according to the target type; and performing watering control according to the judgment condition of the control logic. According to the automatic water sprinkling starting and stopping method and system based on visual identification, complex traffic participation objects can be identified, judged and classified, different coping strategies can be made according to different participation categories of the traffic objects, pedestrians and non-motor vehicles can be commissioned, intelligent starting and stopping can be achieved in the traffic light and traffic jam states, and the water sprinkling efficiency is improved. And waste of water resources is avoided. Meanwhile, distance and azimuth angle positioning is adopted for the target object, and the fan-shaped watering early warning range is matched, so that whether the target object falls into the watering early warning range or not can be judged more accurately.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of sprinkler equipment, and in particular to an automatic start-stop sprinkler method and system based on visual recognition. Background Art

[0002] When existing water-based sanitation vehicles such as sprinkler trucks or dust suppression trucks are performing watering operations, when pedestrians are walking on the street or bicycles / motorcycles are driving on the road, the sanitation vehicle driver needs to manually close the water valve to achieve the purpose of courtesy and avoid spraying water on pedestrians. However, if the driver operates the center console to close the water valve while the vehicle is driving, it is easy for the driver to be distracted, thereby creating the risk of causing a traffic accident.

[0003] In response to the problem of courtesy, the following solution is disclosed in the invention patent with publication number CN116433956A and titled "Automatic watering method, system, terminal and medium for green belts based on sprinkler trucks": the original image of the target area on the right side of the target sprinkler truck is obtained through a visual sensor, and the radar perception result is obtained through a millimeter-wave radar at the same time. The radar perception result and the visual perception result are fused to obtain the current distance between the pedestrian target and the target sprinkler truck, and then the sprinkler system is automatically shut down by comparing the current distance between the pedestrian target and the target sprinkler truck and the preset watering range of the target sprinkler truck. This solution uses visual sensors and millimeter-wave radars to detect visual perception results and radar perception results. Not only is the cost high, but the calculation process of perception result fusion is more complicated.

[0004] On the other hand, when water sanitation vehicles are stuck in traffic or waiting at traffic lights, the drivers are also required to manually close the water valves to save water. This operation often takes up the driver's energy and distracts him, or some drivers may forget this operation, resulting in a waste of water resources. Summary of the invention

[0005] The object of the present invention is to provide a method and system for automatically starting and stopping watering based on visual recognition, so as to at least partially solve the problems mentioned in the background technology.

[0006] On one hand, the present invention provides an automatic start-stop watering method based on visual recognition, which is applied to a watering system of a watering truck, and the method comprises:

[0007] Acquire images in front of and / or to the sides of the sprinkler truck during its driving;

[0008] Acquire the target type of the traffic participant object according to the image;

[0009] Enable corresponding control logic according to the target type;

[0010] The watering control is performed according to the judgment of the control logic.

[0011] In one embodiment, the target categories include pedestrians and non-motor vehicles, and traffic lights and motor vehicles.

[0012] In one embodiment, the target type is pedestrians and non-motor vehicles, and the control logic includes:

[0013] Get the position parameters of the target object;

[0014] It is determined whether the target object is within the watering warning range of the sprinkler truck according to the position parameters.

[0015] In one embodiment, the position parameters include the distance D from the target object to the visual acquisition module, the azimuth angle of the target object from the visual acquisition module

[0016] In one embodiment, the distance D is obtained by:

[0017]

[0018] Among them, F is the focal length value of the visual acquisition module, H is the actual height of the target object, and P is the pixel height of the target object in the picture.

[0019] In one embodiment, the azimuth Obtained through:

[0020]

[0021] Among them, P R and P L is the pixel position of the target object in the image, θ is the horizontal field of view of the visual acquisition module, and f w The pixel width of the image.

[0022] In one embodiment, the watering warning range is a sector-shaped area defined according to the water pump speed of the watering system.

[0023] In one embodiment, the target categories are traffic lights and motor vehicles, and the control logic includes:

[0024] Get the gear position and speed information of the sprinkler truck;

[0025] Determine whether the gear is in the P gear, or whether the vehicle speed is lower than a preset value.

[0026] Another aspect of the present invention provides a watering system for a watering truck, comprising:

[0027] A deployment unit, including a visual acquisition module for image acquisition and a position determination module for target object position determination;

[0028] A training unit, used to construct a target detection model, wherein the target detection model is used to recognize images collected by the visual acquisition module;

[0029] The control unit is used for controlling the watering according to the recognition result.

[0030] Another aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program executable by the processor, and when the processor executes the computer program, the automatic start and stop watering method based on visual recognition described in the above scheme is implemented.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The automatic start-stop watering method and system based on visual recognition provided by the present invention can identify, judge and classify complex traffic participants, and can make different response strategies according to different traffic object participation categories. It can not only give way to pedestrians and non-motor vehicles, but also can start and stop intelligently when waiting for traffic lights and in traffic jams, so as to avoid wasting water resources.

[0033] On the other hand, the method provided by the present invention can accurately locate the position of the target object by simultaneously acquiring the distance and azimuth of the target object when giving way to pedestrians and non-motor vehicles. The method provided by the present invention has a simple calculation process and does not require the use of radar and other judgment equipment, which is more cost-effective. At the same time, the method defines the watering warning range in a fan-shaped area according to the actual speed of the water pump, which is more scientific and reasonable. The target object is located by distance and azimuth, and the fan-shaped watering warning range is used to more accurately determine whether the target object falls into the watering warning range. Compared with the traditional method of only relying on distance comparison to confirm whether the target object is within the watering spray range, the method provided by the present invention is more reliable and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A schematic diagram of a flow chart of an automatic start-stop watering method based on visual recognition provided by an embodiment of the present invention;

[0036] Figure 2A flowchart of an automatic start-stop watering method based on visual recognition provided by an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of a flow chart of a method for acquiring target types provided by an embodiment of the present invention;

[0038] Figure 4 A schematic diagram of a process for constructing a target detection model provided by an embodiment of the present invention;

[0039] Figure 5 A schematic flow chart of a first control logic determination method provided by an embodiment of the present invention;

[0040] Figure 6 is a schematic diagram of an electronic device according to an embodiment of the present invention;

[0041] Figure 7 Schematic diagram of the installation position of the side monocular camera in an embodiment of the present invention;

[0042] Figure 8 Schematic diagram of the installation position of the front monocular camera in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] The present invention discloses an automatic start-stop watering method based on visual recognition, which is applied to a watering system of a watering truck. The method comprises the following steps:

[0045] S1: Acquire images in front and / or on the sides of the sprinkler truck during its driving process.

[0046] Specifically, a visual acquisition module is provided in the sprinkler system of the sprinkler truck. When the sprinkler truck is driving, the visual acquisition module collects images in front of or on the left and right sides of the sprinkler truck, or collects images in front of and on the left and right sides at the same time.

[0047] The visual acquisition module is a monocular camera or a binocular camera installed in front and / or on the side of the sprinkler truck. For cost considerations, a monocular camera is preferably used, and an AHD analog camera among monocular cameras is more preferred. This camera uses analog signal transmission. Compared with monocular camera devices such as webcams and digital cameras, it is not easily affected by complex electromagnetic environments, has higher stability, and can reduce abnormal shooting situations such as jamming.

[0048] The side monocular cameras are respectively installed at the two side ends of the rear of the watering can, and correspond to the sprinkler heads on the corresponding sides in the vertical direction. Preferably, as Figure 7 As shown in the figure, the side monocular cameras are installed at the upper positions of both sides of the tail of the watering can. The installation position of the monocular camera is set at a high point, which can not only reduce the risk of being scratched by obstacles, but also the high installation position will give it a higher shooting field of view to ensure the complete shooting of the target object. The vertical field of view angle α of the side monocular camera lens is 128°, and the angle β between the center line of the lens and the vertical line of the ground is set between 70-77°, so that the bottom of its shooting field of view corresponds to the convex point position on the corresponding side of the watering can to reduce the shooting blind area.

[0049] The front monocular camera is installed below the center axis of the front windshield of the cab. Figure 8 As shown, the distance between the bottom edge and the bottom edge of the front windshield is set at 3-5cm, which is convenient for installation and wiring arrangement, and can also prevent the driver's driving vision from being blocked. The angle α1 between the center line of the front monocular camera lens and the vertical line of the ground is set between 100-110°, so that the top of its shooting field of view can correspond to the traffic light at least 15 meters away.

[0050] S2: Obtain the target type of the traffic participant according to the image collected by the visual acquisition module.

[0051] Traffic participants include at least pedestrians, non-motor vehicles such as bicycles, motorcycles, and tricycles, traffic lights, and motor vehicles such as cars and trucks. Based on the above participants, they are divided into two target categories: pedestrians and non-motor vehicles, and traffic lights and motor vehicles.

[0052] Exemplarily, the visual acquisition module installed in front of the sprinkler truck mainly collects information on traffic lights and motor vehicles, and the visual acquisition module installed on the side of the sprinkler truck can simultaneously collect information on pedestrians, non-motor vehicles, and motor vehicles.

[0053] Specifically, the method for obtaining the target type includes the following steps:

[0054] S201: pre-processing the image collected by the visual acquisition module and inputting it into the pre-trained target detection model;

[0055] S202: After model inference, the target type, confidence level, and pixel location information are obtained.

[0056] More specifically, on the one hand, the method of building the target detection model is as follows:

[0057] S2011: Label the data of traffic participants including pedestrians, bicycles, motorcycles, traffic lights, cars, trucks, etc. to form an image dataset for identifying road traffic participants;

[0058] The specific steps include:

[0059] Data collection: Obtain a large amount of driving video data of different sprinkler trucks in different working environments. The video data covers different weather, lighting conditions and environmental scenes. Video data of different scenes can ensure the generalization ability of the model;

[0060] Data cleaning: Screen and capture videos where the identified target appears, and convert the screened videos into .jpg format images at a frequency of 1FPS;

[0061] Data labeling: Use the LabelImg labeling tool to label the converted images. The labeling targets include pedestrians, bicycles, motorcycles, traffic lights, cars, and trucks. The labeling results are saved in txt format.

[0062] Data augmentation: augment the image dataset by performing rotation, cropping, flipping, etc. to increase the robustness of the model;

[0063] Resizing: Resize the image dataset images to the size (640*640) required by the object detection network.

[0064] S2012: Use the anchor-based object detection network, the YOLOv5 recognition network, to train the image dataset for at least 500 rounds;

[0065] S2013: After the training is completed, based on the optimal training result, a pt model file storing the optimal model parameters is generated, and then the pt model file is converted into a rknn model file that can be used by the deployment unit, that is, a target detection model;

[0066] The optimal training structure is based on the mean average precision (mAP). After each training, the mean average precision of the current training round is calculated. When the calculated mean average precision is higher than that of the previous round, it is considered to be the optimal model under the current training round.

[0067] Taking the detection target as a pedestrian as an example, the calculation process of the average precision mean is:

[0068] 1. Sort detection results: Sort the detection results from high to low according to the confidence level of identifying pedestrian targets;

[0069] 2. Calculate the point-by-point recall and precision: Calculate the precision and recall corresponding to the pedestrian detection results in each image. The recall rate of pedestrian detection is:

[0070]

[0071] Among them: TP is the number of correctly detected pedestrian targets, that is, the number of pedestrians who are actually pedestrians and also recognized; FN is the number of undetected pedestrian targets, that is, the number of pedestrians who are actually pedestrians but not recognized; The recall rate represents the ratio of truly detected pedestrians to the number of pedestrians that should be detected.

[0072] The accuracy of pedestrian detection is:

[0073]

[0074] Among them: TP is the number of correctly detected pedestrian targets, that is, the number of pedestrians who are actually pedestrians and also identified; FP is the number of incorrectly detected pedestrian targets, that is, the number of pedestrians who are not actually pedestrians but are incorrectly identified as pedestrians; the accuracy rate represents the ratio of truly detected pedestrians to the total number of detected pedestrians.

[0075] 3. Calculate the maximum pedestrian precision corresponding to different recall rates under different pedestrian confidence levels: For each recall rate point, find the maximum precision P from that point to the maximum recall rate max (R i );

[0076] 4. Integrate and calculate AP: AP is the accumulation of precision and recall of all these points, that is, add up the maximum precisions and calculate their average.

[0077]

[0078] Where: R i is the i-th recall point, P max (R i ) is the maximum precision corresponding to the i-th recall rate point.

[0079] 5. Calculate the average accuracy of multi-category target detection (when the detection target is more than just pedestrians, take the average accuracy of each detection target)

[0080]

[0081] Where: N is the number of categories, AP i is the average precision of the i-th category.

[0082] S2014: Deploy the object detection model to the edge computing device;

[0083] S2015: Input the preprocessed image into the object detection model;

[0084] S2016: After NPU model inference, the target type, confidence level, and pixel location information are obtained.

[0085] On the other hand, the image preprocessing method is as follows:

[0086] Use gst-launch-1.0 to build a video stream decoder, decode the read AHD analog camera video source into a 640x480 pixel NV12 format image; then preprocess the decoded image by letterbox scaling and converting color channels to convert the BGR channel into RGB channel.

[0087] S3: Activate corresponding control logic according to the target type.

[0088] Specifically, when the identified target type is pedestrians and non-motor vehicles, the first control logic is activated for judgment; when the identified target type is traffic lights and motor vehicles, the second control logic is activated for judgment;

[0089] The first control logic is: judging whether the target object enters the watering warning range according to the position parameters of the target object;

[0090] The second control logic is: judging whether the current state of the sprinkler truck is waiting for a traffic light or being stuck in a traffic jam according to the gear position and speed information of the sprinkler truck.

[0091] It should be noted that if the identified target types include both pedestrians and non-motor vehicles as well as traffic lights and motor vehicles, the second control logic is first enabled to determine whether the vehicle is waiting for a traffic light or is in a traffic jam. If so, watering control is performed according to the determination of the second control logic; if not, the first control logic is enabled.

[0092] S4: Perform watering control according to the judgment of control logic.

[0093] The present invention is further described in detail below in conjunction with the embodiments. It should also be understood that the following embodiments are only used to further illustrate the present invention and cannot be understood as limiting the scope of protection of the present invention. Some non-essential improvements and adjustments made by technicians in this field based on the above content of the present invention all belong to the scope of protection of the present invention.

[0094] Embodiment 1

[0095] In this embodiment, the automatic start and stop watering method based on visual recognition includes the following steps:

[0096] S1: The monocular camera in the sprinkler system of the sprinkler truck is used to obtain the images in front and on the sides of the sprinkler truck during its driving process;

[0097] S2: Preprocess the image captured by the monocular camera and input it into the target detection model to obtain the target types of traffic participants as pedestrians and non-motor vehicles;

[0098] S3: Activate the first control logic according to the target type of pedestrians and non-motor vehicles;

[0099] The first control logic judgment method is:

[0100] S301: Obtaining position parameters of a target object;

[0101] S302: Determine whether the target object is within the watering warning range of the sprinkler truck according to the position parameters.

[0102] Specifically, S301 further includes the following steps:

[0103] S3011: Get the distance D from the target object to the monocular camera:

[0104] The monocular camera distance measurement is completed using the similar triangle method, that is, through the following formula:

[0105]

[0106] Where: F is the focal length of the monocular camera, P is the pixel height of the target object in the image (preprocessed image), D is the distance from the target object to the monocular camera, and H is the actual height of the target object;

[0107] The camera focal length F is a fixed value for each monocular camera. The pixel height P of the target object in the image is obtained by the inference of the rknn model. The actual height H of the target object is obtained by the preset target type height. Therefore, the distance D from the target object to the monocular camera is:

[0108]

[0109] It should be noted that only by capturing the complete target object can the pixel height P be more accurate. If the target object is not captured completely, there will be a deviation between the pixel height P and the actual pixel height, which will inevitably lead to a deviation in the distance D from the target object to the monocular camera, thereby affecting the orientation judgment of the target object. Therefore, combined with the installation position of the aforementioned side monocular camera, installing the side monocular camera at a high point will give it a higher shooting field of view to ensure a complete capture of the target object.

[0110] S3012: Get the azimuth angle of the target object from the monocular camera The calculation is as follows:

[0111]

[0112] Through the reasoning of the rknn model, the pixel position P of the identified target object in the picture can be obtained R (right pixel horizontal coordinate), P L(left pixel horizontal coordinate), and then calculate the horizontal coordinate of the center pixel of the current target object, θ is the horizontal field of view of the monocular camera, f w The pixel width of the current image.

[0113] In S302, the watering warning range of the sprinkler truck is a fan-shaped area delineated according to the actual speed n of the sprinkler pump. The radius of the fan-shaped watering warning range is (0.3~0.5)n / 100 meters, and the angle of the fan-shaped watering warning range is (3~5)n / 100 degrees. The radius and angle of the fan-shaped watering warning range increase with the increase of the actual speed of the sprinkler pump.

[0114] S4: Perform watering control according to the judgment of control logic.

[0115] If the polar coordinates of the target object (distance D, azimuth If it falls within the fan-shaped sprinkler warning range, the hedge valve of the corresponding water pump will be closed.

[0116] In this embodiment, when pedestrians, bicycles, motorcycles and other target types that are easily splashed by water sprayed by sprinkler trucks are identified, the deployment unit calculates the distance and azimuth of each corresponding target object, that is, determines the polar coordinates of each target object, and sends the polar coordinates (distance and azimuth) of the target object to the control unit through the CAN bus. The control unit demarcates the current sprinkler warning range in a fan-shaped area according to the current actual speed of the water pump, and determines whether the polar coordinates of each target object are within the fan-shaped sprinkler warning range. If any target object falls into the sprinkler warning range, the hedge valve of the corresponding water pump is closed.

[0117] This embodiment can accurately locate the position of the target object by simultaneously acquiring the distance and azimuth of the target object. At the same time, because the sprinkler truck usually sprays water at an angle, this embodiment defines the sprinkler warning range in a fan-shaped area according to the actual speed of the water pump, which is more scientific and reasonable. The target object is located by distance and azimuth, and the fan-shaped sprinkler warning range will make it more accurate to determine whether the target object falls into the sprinkler warning range. Compared with the traditional method of only relying on distance comparison to confirm whether the target object is within the sprinkler spray range, the method provided by this solution is more reliable and accurate.

[0118] Embodiment 2

[0119] In this embodiment, the automatic start and stop watering method based on visual recognition includes the following steps:

[0120] S1: The monocular camera in the sprinkler system of the sprinkler truck is used to obtain the images in front and on the sides of the sprinkler truck during its driving process;

[0121] S2: Preprocess the image captured by the monocular camera and input it into the target detection model to obtain the target types of traffic participants as traffic lights and motor vehicles, specifically traffic lights;

[0122] S3: activating the second control logic according to the target types of traffic lights and motor vehicle categories;

[0123] The second control logic judgment method is:

[0124] The gear position and speed information of the sprinkler truck is obtained through the chassis of the sprinkler truck: if the gear position is in D gear or N gear and the speed is less than or equal to 5km / h, or the gear position is in P gear, it is determined that the sprinkler truck is currently in the state of waiting for the traffic light.

[0125] S4: Perform watering control according to the judgment of control logic.

[0126] If the sprinkler truck enters the state of waiting for traffic lights, the control unit of the sprinkler system will issue a shutdown operation command to control the sprinkler motor to stop and avoid wasting water resources.

[0127] Embodiment 3

[0128] In this embodiment, the automatic start and stop watering method based on visual recognition includes the following steps:

[0129] S1: The monocular camera in the sprinkler system of the sprinkler truck is used to obtain the images in front and on the sides of the sprinkler truck during its driving process;

[0130] S2: Preprocess the image captured by the monocular camera and input it into the target detection model to obtain the target types of traffic participants as traffic lights and motor vehicles, specifically motor vehicles, and at the same time, count the number of motor vehicles;

[0131] S3: activating the second control logic according to the target types of traffic lights and motor vehicle categories;

[0132] The second control logic judgment method is:

[0133] If the monocular camera detects that the number of motor vehicles within the surrounding 5 meters environment exceeds 10, the gear and speed information of the sprinkler truck will be obtained through the chassis of the sprinkler truck: if the gear is in D gear or N gear and the speed is less than or equal to 5km / h, or the gear is in P gear, it is judged that the sprinkler truck is in a traffic jam.

[0134] S4: Perform watering control according to the judgment of control logic.

[0135] If the sprinkler truck enters a traffic jam, the control unit of the sprinkler system will issue a shutdown operation command to control the sprinkler motor to shut down and avoid wasting water resources.

[0136] Embodiment 4

[0137] In this embodiment, the automatic start and stop watering method based on visual recognition includes the following steps:

[0138] S1: The monocular camera in the sprinkler system of the sprinkler truck is used to obtain the images in front and on the sides of the sprinkler truck during its driving process;

[0139] S2: Preprocess the image captured by the monocular camera and input it into the target detection model to obtain the target types of traffic participants as traffic lights and motor vehicles, specifically including traffic lights and motor vehicles, and at the same time, count the number of motor vehicles;

[0140] S3: activating the second control logic according to the target types of traffic lights and motor vehicle categories;

[0141] The second control logic judgment method is:

[0142] The gear position and speed information of the sprinkler truck is obtained through the chassis of the sprinkler truck: if the gear position is in the D gear position or the N gear position and the speed is less than or equal to 5km / h, or the gear position is in the P gear position, it is determined that the sprinkler truck is currently in the state of waiting for the traffic light;

[0143] If the monocular camera detects that the number of motor vehicles within the surrounding 5 meters exceeds 10, it is judged that the sprinkler truck is in a traffic jam at the same time; if the number of motor vehicles does not exceed 10, it is judged that the sprinkler truck is only waiting for the traffic light;

[0144] S4: Perform watering control according to the judgment of control logic.

[0145] When the sprinkler truck enters the state of waiting for the traffic light, no matter whether the sprinkler truck is in a traffic jam or not, the control unit of the sprinkler system will issue a shutdown operation command to control the sprinkler motor to stop, so as to avoid wasting water resources.

[0146] Embodiment 5

[0147] In this embodiment, the automatic start and stop watering method based on visual recognition includes the following steps:

[0148] S1: The monocular camera in the sprinkler system of the sprinkler truck is used to obtain the images in front and on the sides of the sprinkler truck during its driving process;

[0149] S2: Preprocess the image captured by the monocular camera and input it into the target detection model to obtain the target types of traffic participants, including traffic lights and motor vehicles, as well as pedestrians and non-motor vehicles;

[0150] S3: First, the second control logic is enabled, and whether to enable the first control logic is determined according to the judgment of the second control logic;

[0151] Specifically, according to the judgment method of the second control logic described in the above embodiment, it is judged whether the sprinkler truck is currently in a state of waiting for a traffic light or being stuck in a traffic jam. If yes, there is no need to activate the first control logic. If not, the first control logic is activated.

[0152] S4: Perform watering control according to the judgment of control logic.

[0153] If the sprinkler truck enters the state of waiting for traffic lights or traffic jam, the control unit of the sprinkler system issues a shutdown operation command to control the sprinkler motor to stop;

[0154] If the sprinkler truck is not in the state of waiting for traffic lights or traffic jam, the polar coordinates (distance D, azimuth angle) of the target object such as pedestrians or non-motor vehicles are determined. Whether it falls within the fan-shaped sprinkler warning range, sprinkler control is carried out based on the judgment situation.

[0155] Based on the automatic start-stop watering method based on visual recognition disclosed in the above embodiment, the present invention also discloses a watering system for a watering truck, which includes a training unit, a deployment unit and a control unit:

[0156] The deployment unit includes a visual acquisition module for image acquisition and a position determination module for target object position determination. The visual acquisition module is a monocular camera installed in front of and / or on the side of the sprinkler truck, and is used to acquire images in front of or on the left and right sides of the sprinkler truck, or can acquire images in front of and on the left and right sides at the same time; the position determination module is used to determine the distance or azimuth of the target object (pedestrian, non-motor vehicle, motor vehicle, etc.);

[0157] The training unit is used to construct a target detection model, and the target detection model is used to identify the images collected by the visual acquisition module and confirm the target type of the traffic participant object;

[0158] The control unit is used to perform watering control according to the recognition result. Specifically, the type of control logic to be enabled is determined according to the target type recognized by the target detection model, and the watering control is performed according to the corresponding control logic judgment situation.

[0159] Another aspect of the present invention provides an electronic device, Figure 6 This is a structural block diagram of the electronic device. The electronic device may include one or more of the following components: a memory, a processor, and one or more computer programs, wherein the one or more computer programs may be stored in the memory and configured to be executed by one or more processors, and the one or more computer programs are configured to execute the aforementioned automatic start and stop sprinkler method based on visual recognition.

[0160] The memory may include a random access memory (RAM) or a read-only memory (ROM). The memory may be used to store instructions, programs, codes, code sets or instruction sets. The memory may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function, instructions for implementing the aforementioned various method embodiments, etc. The data storage area may also store data created by the sprinkler system during use.

[0161] The processor may include one or more processing cores. The processor uses various interfaces and lines to connect the various parts of the entire sprinkler system, and executes various functions of the sprinkler system and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory. Optionally, the processor can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor can integrate one or a combination of a central processing unit (CPU) and a modem. Among them, the CPU mainly processes the operating system and application programs; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor, but may be implemented separately through a communication chip.

[0162] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic start-stop watering method based on visual recognition, applied to the watering system of a watering truck, characterized in that: The method comprises: Acquire images in front of and / or to the sides of the sprinkler truck during its driving; Acquire the target type of the traffic participant object according to the image; Enable corresponding control logic according to the target type; The watering control is performed according to the judgment of the control logic.

2. The automatic start-stop watering method based on visual recognition according to claim 1 is characterized in that: The target categories include pedestrians and non-motor vehicles, traffic lights and motor vehicles.

3. The automatic start-stop watering method based on visual recognition according to claim 2 is characterized in that: The target type is pedestrians and non-motor vehicles, and the control logic includes: Get the position parameters of the target object; It is determined whether the target object is within the watering warning range of the sprinkler truck according to the position parameters.

4. The automatic start-stop watering method based on visual recognition according to claim 3 is characterized in that: The position parameters include the distance D from the target object to the visual acquisition module, the azimuth angle of the target object from the visual acquisition module 5. The automatic start-stop watering method based on visual recognition according to claim 4 is characterized in that: The distance D is obtained by: Among them, F is the focal length value of the visual acquisition module, H is the actual height of the target object, and P is the pixel height of the target object in the picture.

6. The automatic start-stop watering method based on visual recognition according to claim 4 is characterized in that: The azimuth Obtained through: Among them, P R and P L is the pixel position of the target object in the image, θ is the horizontal field of view of the visual acquisition module, and f w The pixel width of the image.

7. The automatic start-stop watering method based on visual recognition according to claim 3 is characterized in that: The sprinkler warning range is a sector-shaped area defined according to the water pump speed of the sprinkler system.

8. The automatic start-stop watering method based on visual recognition according to claim 2 is characterized in that: The target categories are traffic lights and motor vehicles, and the control logic includes: Get the gear position and speed information of the sprinkler truck; Determine whether the gear is in the P gear, or whether the vehicle speed is lower than a preset value.

9. A watering system for a watering truck, characterized in that: include: A deployment unit, including a visual acquisition module for image acquisition and a position determination module for target object position determination; A training unit, used to construct a target detection model, wherein the target detection model is used to recognize images collected by the visual acquisition module; The control unit is used for controlling the watering according to the recognition result.

10. An electronic device, characterized in that: The invention comprises a memory and a processor, wherein the memory stores a computer program executable by the processor, and the processor implements the automatic start and stop watering method according to any one of claims 1 to 8 when executing the computer program.

Citation Information

Patent Citations

  • Automatic green belt watering method and system based on watering cart, terminal and medium

    CN116433956A

  • Unmanned multifunctional road sweeper

    CN109235332A

  • Unmanned sweeping vehicle and water sprinkler system control method thereof

    CN109610396A

  • Watering method of watering cart, automatic watering control system, main control box and watering cart

    CN111178223A

  • Energy-saving watering cart capable of automatically avoiding pedestrians

    CN116824538A

Cited By

  • Method and device for water sprinkler to avoid pedestrians based on visual identification area proportion detection

    CN119478896A