Embedded drowning emergency identification and rescue system

Abstract

The invention designs a drowning emergency rescue system which integrates drowning monitoring and rescue and is based on an embedded technology and an internet of things technology. The system is of a fixable integrated structure, images are collected through infrared and thermal imaging binocular cameras, an overall hardware circuit of the system serves as a medium, data processing is conducted on the collected images through an autonomously designed drowning monitoring algorithm, and a target is tracked and locked. And analyzing the overall strength of the thermal imaging signal and the rough posture of the human body to judge whether drowning occurs. When a drowning situation occurs, angle and distance parameters provided by a front-end camera are used for calculation, a friction wheel is used for delivering lifesaving equipment at a fixed point, an air bag is inflated in a delayed mode, meanwhile, the drowning situation is uploaded on a cloud platform, the specific position is sent, and rescue workers are guided to conduct rescue. According to the invention, drowning detection and recognition and an autonomous rescue scheme are systematized, the drowning recognition rate of a complex water area is improved, and meanwhile, the rescue probability of a drowning person is increased. The self-designed drowning recognition algorithm takes embedded equipment as a medium, so that efficient and stable operation is ensured, and the installation cost is greatly reduced.

Description

Technical Field

[0001] The invention belongs to the field of electronic information and embedded systems, and particularly relates to an embedded drowning emergency recognition and rescue system. Background Art

[0002] When it comes to drowning safety, we cannot just strengthen our awareness of prevention, but we must also monitor the waters and quickly rescue people after drowning incidents occur to improve the survival rate after drowning.

[0003] Currently, drowning detection in advanced waters uses visual cameras and high-computing computers to perform posture calculations to implement monitoring. This is greatly affected by environmental conditions and the complexity of the detection site, and is costly and cannot be widely deployed.

[0004] Most of the mainstream civilian anti-drowning equipment on the market are wearable devices, but most drowning people do not prepare self-rescue measures in advance and cannot effectively rescue themselves after drowning. Therefore, in order to provide accurate judgment and rapid rescue response for drowning incidents, we designed an embedded drowning emergency identification and rescue system. Summary of the invention

[0005] This design aims to use embedded devices to balance cost and drowning recognition accuracy, solve the problem of drowning judgment in complex waters and drowning rescue within the golden rescue time.

[0006] In order to solve the above technical problems, the functions realized by the present invention are: F1. Accurate detection: The system can accurately detect the occurrence of drowning through the front-end camera and send a drowning signal to the main control chip; F2, Autonomous rescue: After drowning is detected, the system receives the detected drowning location information, and the chip issues a rescue command. The single gimbal on the side of the launch airbag autonomously adjusts the horizontal angle, controls the rotation speed of the friction wheel to launch the airbag, and activates the airbag by delaying the signal criticality to perform rescue; F3, Real-time alarm: After the system detects drowning data, it will quickly send a text message alarm and upload the location information; F4. Remote synchronization: Upload the detected images to the cloud platform synchronously, and guide rescuers to rescue through the drowning location and real-time situation on the scene.

[0007] The YOLOV5 algorithm for human body tracking and locking based on the image uploaded by the visual camera takes the average of the first few frames as the background to obtain the background image, differentiates the subsequent images from the background image, converts them into grayscale values, performs binarization processing, calculates the center coordinates of the contour, constrains the coordinates to [0,100], frames the target with a rectangle on the new image, and outputs the target coordinates through the serial port.

[0008] The infrared image stitching algorithm, the data transmitted by thermal imaging is the temperature data of each pixel and a 32*24 temperature matrix. Each temperature data is accurate to two decimal places and consists of two 8-bit data. Therefore, after receiving the data, the data must be spliced ​​and CRC checked first, and the temperature data must be converted into grayscale values, and then converted from grayscale values ​​into 16-bit arrays in RGB565 format. Finally, the spliced ​​thermal map is displayed on the LCD screen.

[0009] The dual gimbal design for placing the infrared camera uses a stepper motor directly attached to the bottom plate as a force point. The two layers of rings in the middle wrap around the deep groove ball bearing. The upper plane passes through the bearing and connects to the lower motor to perform 360-degree rotation in the horizontal direction. A semi-enclosed shell is made for the infrared camera on top. The motor is fixed on the side plate, and the front circular hole is embedded in the infrared camera to rotate the motor in the vertical direction. A reset switch is installed on the other side of the shell to calibrate the angle of the infrared camera.

[0010] The autonomous life-saving airbag launching device uses an electronic controller to control a brushless motor, sends a 20ms signal through a timer, adjusts the duty cycle to simulate the unlocking of the electronic controller and the setting of the throttle range, and accurately adjusts the rotation speed of the friction wheel according to the established airbag launching model to achieve accurate delivery of the autonomous life-saving airbag.

[0011] The critical point activation of the autonomous life-saving airbag calculates the flight time in the air according to the coordinate parameters of the drowning position given by the visual camera, and at the critical time point when the person is about to fall into the water, the Bluetooth board is remotely controlled to activate the life-saving airbag.

[0012] We designed a double-layer cylindrical structure for the airbag excitation device. With a DC reduction motor as the main body, a spring is inserted into the torsion bar, the excitation block above compresses the spring to the specified position, the round rod on the side wall of the excitation block is embedded in the card slot, and a carbon dioxide cylinder is installed in the direction of the excitation block popping out. When the Bluetooth board on the side of the motor receives the delay signal, the motor is started, the excitation block is popped out of the card slot, the inert gas cylinder is pierced, and the airbag is inflated.

[0013] In the part of uploading to the cloud platform, the transparent transmission mode of ESP01S is used to realize the JPEG image transmission function, and the OV2640 is used to realize image compression through the DVP function of the development board. The image is transparently transmitted to the computer side and OPENCV is used to lock the target and transmit the coordinate information back; the development board sends AT commands through the serial port and uses the MQTT protocol to connect to the cloud platform to report the drowning information.

[0014] In order to ensure the stability of the entire system, we cut some boards, punched round holes of a certain size, and accurately connected the upper and lower parts using bolt structures.

[0015] The waterproof treatment of the system adopts a closed structure at the bottom layer, and an embedded development board for normal operation of the control system is placed, and waterproof treatment is performed around the camera. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a design flow chart of the present invention; Figure 2 It is a schematic diagram of the main structure of the present invention; Figure 3 It is a schematic diagram of assembling the airbag excitation device of the present invention; Figure 4 It is a schematic diagram of the internal structure of the airbag excitation device of the present invention; Figure 5 It is the principle diagram of autonomous life-saving airbag launch; Figure 6 A schematic diagram of a Bluetooth board for activating an autonomous rescue device in the present invention; In the figure: camera group 1, dual gimbal structure 2, brushless motor 3, single gimbal structure 4, autonomous life-saving airbag 5, push rod 6, friction wheel 7, life-saving airbag shell 8, life-saving airbag main structure 9, stepping motor 10, spring 11, side rotation rod 12, excitation block 13, firing pin 14, gas cylinder 15, Bluetooth board 16, lithium battery 17, card slot 18. DETAILED DESCRIPTION

[0017] The technical route of this product is as follows Figure 1 shown.

[0018] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the invention is further described below with reference to specific illustrations.

[0019] The embedded drowning emergency identification and rescue system of the present invention proposes the following technical solutions: F1. Accurate detection: The system can accurately detect the occurrence of drowning through the front-end camera and send a drowning signal to the main control chip; F2, Autonomous rescue: After drowning is detected, the system receives the detected drowning location information, and the chip issues a rescue command. The single gimbal on the side of the launch airbag autonomously adjusts the horizontal angle, controls the rotation speed of the friction wheel to launch the airbag, and activates the airbag by delaying the signal criticality to perform rescue; F3, Real-time alarm: After the system detects drowning data, it will quickly send a text message alarm and upload the location information; F4. Remote synchronization: Upload the detected images to the cloud platform synchronously, and guide rescuers to rescue through the drowning location and real-time situation on the scene.

[0020] Figure 2For the main part of the embedded drowning emergency identification and rescue system, a suitable position is selected to fix the base, and the entire system is powered on. The main control uses the flip IO solution to convert the driving voltage curve into a smoother sine curve. The number of steps of the motor movement is recorded through the AT24C04 EEPROM chip on the homemade expansion board. When powered on again, the photoelectric sensor on the side of the pan / tilt uses the EXTI external interrupt to complete the power-on reset of the motor, so that the camera can rotate at a fixed angle each time to complete the splicing of the thermal map.

[0021] After the system is powered on, the visual camera starts working, takes the average of the first few frames collected as the background, obtains the background image, differentiates the subsequent images from the background image, converts them into grayscale values, performs binarization processing, calculates the center coordinates of the contour, constrains the coordinates to [0,100], and frames the target with a rectangle on the new image. The target coordinates are output through the serial port to track and lock the drowning target.

[0022] When drowning is detected, the launching device starts, and the position sensor detects the real-time position of the servo system. The received signal is compared as the controller input. The output control acts on the crank slider mechanism after the controller calculation, causing the flight angle and pressure of the airbag delay device to change in the force direction, and finally acts on the controlled object to form a closed-loop control of the launching speed, completing the launching of the autonomous life-saving airbag.

[0023] The clock signal is controlled by the main control development board. According to the calculated time, a signal is sent out one second before falling into the water to control the mechanical device to pierce the placed gas cylinder for inflating. At the same time, the chip is connected to the air pressure sensor module. If it is affected by external force and the gas is not inflated normally, the rapid change of air pressure after falling into the water will also cause the air pressure sensor module to send a signal to control the mechanical device to pierce the gas cylinder.

[0024] When drowning occurs, the transparent transmission mode of ESP01S is used to realize the JPEG image transmission function, and the OV2640 is used to compress the image through the DVP function of the development board. The image is transparently transmitted to the computer and OPENCV is used to lock the target and transmit the coordinate information back. The development board sends AT commands through the serial port and uses the MQTT protocol to connect to the cloud platform to report the drowning information.

[0025] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The drowning emergency identification and rescue system is characterized by: The functions implemented are: F1. Accurate detection: The system can accurately detect the occurrence of drowning through the front-end camera and send a drowning signal to the main control chip F2. Autonomous rescue: After drowning is detected, the system receives the detected drowning location information, and the chip issues a rescue command. The single gimbal on the side of the launch airbag autonomously adjusts the horizontal angle, controls the rotation speed of the friction wheel to launch the airbag, and activates the airbag by delaying the signal critical point to perform rescue. F3, Real-time alarm: After the system detects drowning data, it will quickly send a text message alarm and upload the location information F4. Remote synchronization: Upload the detected images to the cloud platform synchronously, and guide rescuers to rescue through the drowning location and real-time situation on the scene.

2. The embedded drowning emergency identification and rescue system according to claims 1, 2 and 3, characterized in that: The described solution for monitoring the water area within the range and realizing efficient and accurate drowning detection is based on the stitching algorithm of infrared images and the capture of disappearing thermal imaging signals. The data transmitted by thermal imaging is defined as the temperature data of each pixel and a 32*24 temperature matrix. Each temperature data is accurate to two decimal places and consists of two 8-bit data. Therefore, after receiving the data, the data must be spliced ​​and then converted into grayscale values ​​after CRC verification. The grayscale values ​​are then converted into a 16-bit array in RGB565 format. Finally, the image stitching of the infrared part is realized through the independently designed dual gimbal structure, and the spliced ​​thermal map is displayed on the LCD screen. When the system captures the disappearing thermal imaging signal, the system also performs posture detection on the locked human target. When it matches the large-scale struggling model in the database, a drowning signal is issued.

3. The embedded drowning emergency identification and rescue system according to claims 1 and 2, characterized in that: The human body tracking and locking algorithm, which is independently designed based on the Yolo algorithm, takes the average of the first few frames as the background to obtain a background image, differs the subsequent images from the background image, converts them into grayscale values, performs binarization processing, calculates the center coordinates of the contour, constrains the coordinates to [0,100], and frames the target with a rectangle on the new image, and outputs the target coordinates through the serial port.

4. The embedded drowning emergency identification and rescue system according to claims 1 and 2, characterized in that: The independently designed drowning detection algorithm based on rough judgment of human body posture obtains the coordinates of 18 key points of the human body in the water in advance, constructs the key point distance vector of the human body, and determines whether the swimmer is drowning by calculating the similarity between the key point distance vector of the human body and the key point distance vector of the drowning state.

5. The embedded drowning emergency identification and rescue system according to claim 1, 2 or 3, characterized in that: The accurate coordinates of the drowning are determined by the front-end camera, and a three-dimensional coordinate system is established with the center of the camera lens as the origin, so that the drowning position is reduced to a range of 50cm×50cm×50cm.

6. The embedded drowning emergency identification and rescue system according to claim 1, 2 or 3, characterized in that: After the drowning incident occurs, the autonomous rescue device is launched, the brushless motor is controlled by the electronic regulator, a 20ms signal is sent through the timer, the duty cycle is adjusted to unlock the electronic regulator and set the throttle range, the rotation speed of the friction wheel is adjusted according to the inertial motion model established in advance, and the accurate launch of the autonomous rescue device is controlled.

7. The embedded drowning emergency identification and rescue system according to claim 1 or 2, characterized in that: The life-saving device is activated at a critical moment, and the rotation of the motor is controlled by a self-designed Bluetooth board to start the mechanical structure to penetrate the inert gas cylinder, thereby achieving the inflation of the life-saving airbag.

8. The embedded drowning emergency identification and rescue system according to claim 4 is characterized in that: The dual gimbal structure uses TMC2209 to drive 42 stepper motors, and uses the flip IO port solution to convert the driving voltage curve into a smoother sine curve, so that the stepper motor controls the camera module to swing a fixed distance in the horizontal and vertical directions to complete image stitching. At the same time, the AT24C04 EEPROM chip on the homemade expansion board is used to record the number of steps of the motor movement. Each time the power is turned on, the photoelectric sensor on one side of the gimbal uses the EXIT external interrupt to complete the power-on reset of the motor.

9. The embedded drowning emergency identification and rescue system according to claim 1 or 7, characterized in that: The breakdown mechanical structure can be divided into the following two steps: S1. Insert the motor's rotary rod into the bottom of the firing device, put the spring on the rotary rod, insert the iron needle into the trigger block, put the trigger block in from one side of the spring, compress the spring to accurately insert the rotary rod into the trigger block, insert the side rotary rod into the hole reserved in the trigger block, rotate the side rotary rod to make it fit into the side groove, put the gas cylinder in on the same side and fix it, put on the shell and life-saving airbag S2. When a drowning signal is received, the main control chip sends a clock signal to the Bluetooth board. The Bluetooth board controls the motor to rotate, driving the excitation block to rotate together. When the side rotating rod is separated from the side groove, the compressed spring pops out the excitation block, and the firing pin on the top of the excitation block pierces the gas cylinder, and the released gas is filled into the life-saving airbag.

10. The embedded drowning emergency identification and rescue system according to claim 7, characterized in that: There are two ways to activate the system at critical time. One is to actively calculate the time of falling into the water and send a signal, and the other is to passively detect the water pressure. When the water pressure threshold given in advance is reached, a signal is sent to ensure the successful activation of the autonomous rescue device.

11. The embedded drowning emergency identification and rescue system according to claim 1, 2 or 3, characterized in that: The image information collected by the camera is used to determine drowning and transmit the image. The drowning results obtained after analysis are uploaded to the supervision system through the image transmission unit to ensure that the supervision system can quickly receive and process the results. At the same time, a drowning database is established in the supervision system to record and store the obtained drowning images and analysis results, so as to improve the data model for determining drowning.

12. The embedded drowning emergency identification and rescue system according to claim 1, 2 or 3, characterized in that: The YOLOv5-based target tracking algorithm, human posture estimation drowning algorithm, and life-saving device delayed activation algorithm can all run normally on new embedded devices, reducing installation costs and operating costs.

Images