Under-ice acoustic-optical combined guidance and recovery method for underwater unmanned vehicle

Abstract

The application discloses an ice-underwater sound-light combined guidance recovery method of an unmanned underwater vehicle and relates to the ice-underwater sound-light combined guidance recovery method. The ice-underwater sound-light combined guidance recovery method is used to solve the problems of the ice-underwater environment interference and unpredictability, the large guidance docking operation difficulty of the existing underwater docking recovery platform, the poor stability, the poor safety, the easy environmental factor interference, and the poor acoustic guidance precision. The ice-underwater sound-light combined guidance recovery method can realize the ice-underwater autonomous docking of the unmanned underwater vehicle with high robustness, high success rate, high stability, high safety and high precision. In theory, the effective guidance depth meeting the actual engineering requirements can be obtained by adjusting the side length of the regular pentagon light source array. The ice-underwater sound-light combined guidance recovery method belongs to the technical field of unmanned underwater vehicles.

Description

Technical Field

[0001] This invention relates to a combined acoustic and optical guidance and recovery method under ice, belonging to the field of underwater unmanned vehicle technology. Background Technology

[0002] Unmanned underwater vehicles (UUVs) are playing an increasingly important role in marine engineering, marine scientific research, and military activities. [1] The ice sheets in most polar regions have been drastically altered by climate change. Global warming has reduced the thickness, age, and extent of sea ice, and thinning ice may open new shipping lanes and trade routes in previously inaccessible areas. Furthermore, the receding ice sheets create new opportunities for the exploitation of valuable natural resources. Underwater UUVs (Underwater Vehicles), as a novel polar survey and monitoring platform, have a very broad application prospect in polar underwater environments due to their cost and operational flexibility, and are becoming a major trend in the future development of polar environmental sensing. After a period of underwater operation, UUVs need to recover and replenish energy or extract experimental data due to insufficient energy or mission completion. This requires UUVs to be able to successfully complete recovery tasks, including energy replenishment or data extraction, which involves many key technologies. Energy replenishment and data exchange are two important factors affecting the duration and efficiency of UUV operations. Constrained by their own energy carrying capacity and the underwater environment, UUV energy replenishment and data exchange generally require recovery to a mother ship. The mother ship release and recovery process is highly intensive and dangerous, greatly limiting the engineering application of UUVs. Underwater docking technology has enabled UUVs to complete energy replenishment and data exchange underwater, avoiding problems such as low automation, high risk, and poor stealth in the deployment and recovery process. In recent years, it has become a hot topic in underwater robotics research. Meanwhile, to improve UUV recovery efficiency and increase automation, autonomous homing and docking technology has become a key research area for researchers worldwide, with some achievements made.

[0003] UUV underwater docking technology utilizes acoustic, optical, and visual guidance methods to determine the spatial relationship between the UUV and the docking device, and then employs path planning to achieve docking. The Odyssey IIB AUV underwater docking system, jointly developed by the Woods Hole Oceanographic Institution and MIT, is an example. [2] The Marine-bird underwater docking system from Kawasaki Heavy Industries of Japan, and the electromagnetic guidance docking system from the team at Hangzhou Dianzi University of China. [3]Static docking systems are typical examples. Because the docking device is fixed in a specific underwater area, the docking area is limited, and long-term use is significantly affected by seawater corrosion and marine organism attachment. Dynamic docking occurs below the ice layer, and the surrounding environment is uncontrollable, making operation more difficult and dangerous, and unsuitable for ice-covered areas. In recent years, static docking has primarily utilized acoustic beacons, such as ultra-short baselines (USBL), for long-range initial guidance, and optical beacons for short-range precise guidance. This combined acoustic-optical guidance method has gradually become the mainstream approach for AUV recovery. [4] Park, etc. [5] Using a monochrome camera to capture a five-light array, traditional segmentation and shape processing methods were employed to successfully identify the position and pose of the light source at a distance of 15m in a pool. However, due to the scattering of the light source, the center of the light was not well extracted, resulting in significant errors. Palomeras et al. [6] Combining active and passive beacons, augmented reality (AR) markers were used to supplement near-field visual blind spots, and a docking experiment was completed in lake water. (Liu et al.) [7] A recognition framework based on detection, segmentation, and pose estimation is proposed, and the YOLO (you only look once) network is applied to underwater base station identification. Optical guidance within a 10m range is successfully achieved in a lake environment; however, the algorithm fails when the neural network detection is lost. (Lin et al.) [8] Two-degree-of-freedom positioning was achieved using a single light source, and optical guidance over a distance of 20-30m on the lake was completed by combining it with a line-of-sight (LOS) scheme. Summary of the Invention

[0004] To address the problems of numerous and unpredictable interferences from the underwater environment and the difficulties, instability, safety, susceptibility to environmental interference, and poor acoustic guidance accuracy of existing underwater docking and recovery platforms, this invention proposes a combined acoustic and optical guidance and recovery method for underwater unmanned vehicles.

[0005] The technical solution adopted by the present invention to solve the above problems is as follows: The present invention specifically includes:

[0006] Step 1: The acoustic array near the shore-side recovery well sends an interrogation signal to the transponder inside the underwater unmanned vehicle. After receiving the interrogation signal, the transponder sends a response signal. The distance and angle are calculated by calculating the time difference and phase difference between sending the interrogation signal and receiving the response signal.

[0007] Step 2: When the underwater unmanned vehicle reaches the vicinity of the recovery platform, it encodes the depth information and sends it to the shore control box via underwater acoustic communication. The light source is then adjusted to the guidance brightness via the serial port server.

[0008] Step 3: The motion control unit of the underwater unmanned vehicle adjusts to a suitable search depth by collecting data from the altimeter. The vehicle performs a comb search through the motion control unit. The image processing unit acquires and detects images through the upward-looking optical camera. Proceed to Step 4.

[0009] Step 4: The image processing unit acquires the latest frame image for light source detection and calculates the position and orientation of the center of the guiding light source array relative to the vehicle.

[0010] Step 5: The motion control unit adjusts the motion direction of the underwater unmanned vehicle and moves it toward the target point based on the information provided by the image processing unit. The trajectory deviation is controlled by a two-layer tracking control system, which includes top-level heading control and bottom-level execution control. For heading control, a fuzzy PID controller is used to compensate for the heading angle. For execution control, an S-plane controller is used to operate the rudder and thrusters so that the vehicle reaches the target position with the desired heading angle and speed. After the adjustment is completed, proceed to step 6.

[0011] Step 6: The image processing unit acquires the latest frame image to detect light sources, calculates the center position and direction of the light source array, and determines the number of detected light sources. If the vehicle reaches the center of the light source array, proceed to step 7; otherwise, repeat step 4.

[0012] Step 7: The image processing unit acquires multiple consecutive frames of images for light source detection. If the number of light sources in the acquired 30 frames is 5, the motion control unit completes the vehicle's turning based on the S-plane controller using the information provided by the image processing unit. After the turning is completed, the bow of the vehicle is facing a vertex and perpendicular to the side opposite the vertex. Proceed to Step 8. If the number of light sources in the acquired 30 frames is less than 5, the motion control unit controls the vehicle to dive to a certain depth and encodes the depth information acquired by the altimeter and sends it to the shore control box via underwater acoustic communication. The light source is linearly increased to a suitable brightness. Proceed to Step 4.

[0013] Step 8: Move the center of the vehicle to the center of the light source array. After the adjustment is completed, the unmanned vehicle arrives at the predetermined docking and recovery position below the guidance and recovery platform. The motion control unit collects data from the altimeter and rises in stages. The underwater unmanned vehicle emerges from the ice well surface, and the operation ends.

[0014] Furthermore, the distance between the acoustic array and the transponder signal is R, as shown in the following formula:

[0015] R = cΔT

[0016] Where ΔT is the time difference between the acoustic array and the transponder signal, c is the speed of sound in water, and ΔT is the time difference. The distance between the two acoustic units is l, and the phase difference of the received signals is... If the angle of incidence of the signal to the hydrophone is θ, then the relationship between the three is:

[0017]

[0018] The phase difference is obtained from the measured phase difference of the ultra-short baseline (USBL) line:

[0019]

[0020] The relative position of the carrier is calculated using the slant distance R and the azimuth angle θ. xyz Representing the acoustic array coordinate system, the relationship between the slant distance R and the carrier depth h is easily obtained as follows:

[0021] h=Rsinθ

[0022] The horizontal distance between the acoustic array and the carrier is:

[0023] s=Rcosθ

[0024] The mean square error of the horizontal distance relative to the slope distance R is:

[0025]

[0026] The motion control unit analyzes acoustic information and adjusts the vehicle's motion direction based on a two-layer tracking control system of fuzzy PID controller and S-plane controller, moving it to the vicinity of the shore-based acoustic array.

[0027] Furthermore, the formula for calculating the brightness of the light source is:

[0028] I d =I0·e kd

[0029] I d I0 is the brightness of the LED light at a depth of d, k is a constant representing the absorption and scattering coefficients of water, and d is the depth.

[0030] The image processing unit acquires a frame of image through an optical camera and uses a pre-trained YOLOv8s object detection model to detect light sources. If the target confidence score (conf) is greater than a certain value, the target is considered a light source. The formula for calculating conf is as follows:

[0031] conf=(1-gr)+gr·CIoU

[0032] Where gr represents the probability of the presence of the light source target, and CIoU is the bounding box loss, calculated as follows:

[0033]

[0034] Where ρ is the distance between the center points of the predicted bounding box and the label box, c is the diagonal length of the minimum bounding rectangle of the predicted bounding box and the label box, ν is the aspect ratio similarity between the predicted bounding box and the label box, α is the influence factor of ν, and IoU is the ratio of the intersection area and the union area of ​​the predicted bounding box and the label box, as shown in the following formula:

[0035]

[0036] For targets identified as light sources, the NMS algorithm is used for filtering. First, the light source targets are sorted in descending order of confidence and placed into the input list. The highest value is removed and placed into the output list. The intersection-union ratio (IoU) of the removed highest value and the second highest value is calculated. If the IoU is greater than the threshold TH, the second highest value is removed. The above steps are repeated until the input list is empty, and the NMS processing is completed. The threshold TH is a predetermined value. Whether the guiding light source is found is determined by whether the threshold is reached. If not, proceed to step 3; if yes, proceed to step 4.

[0037] Furthermore, the formulas for calculating the position and orientation of the center of the guidance light source array relative to the vehicle in step 4 are as follows:

[0038]

[0039] in These are the detection boxes (e.g.) Figure 4 The coordinates of the top left and bottom right points, The coordinates of the points within the detection frame are given. The number of detected light sources is determined; if the number is 1, its position and direction relative to the vehicle are calculated. When there are two light sources, the midpoint between the two light sources is the target point, and the calculation formula is:

[0040]

[0041] When the number of light sources is 3, the midpoint of the longest side of each light source is the target point, calculated using the following formula:

[0042]

[0043] in When the number of light sources is 4, the center position and direction of the light source array are calculated using the quadrilateral feature. Here, the intersection of the diagonal lines is assumed to be the center. Substituting the coordinates of the two points on the diagonal, the equation of the diagonal is obtained, as shown in the following formula:

[0044]

[0045] The intersection of the two diagonals is the target point. If the number of light sources is 5, the center position of the light source array is calculated using the characteristics of a regular pentagon. First, the distance from any point to the two farthest points is calculated, and then the coordinates of the midpoint between the two farthest points are calculated using the following formula:

[0046]

[0047] According to the properties of a regular pentagon, its center point is:

[0048]

[0049] The angle between the bow of the aircraft and the center of the detected light source array is:

[0050]

[0051] The distance between the center of the optical camera and the center of the light source array is:

[0052]

[0053] Among them The image center coordinates, h is the depth, and f is the focal length of the optical camera. s and θ are sent to the motion control unit. After successful transmission, proceed to step 5.

[0054] The beneficial effects of this invention are:

[0055] 1. In this invention, a combined ultra-short baseline and optical vision guidance method is used. The invention proposes to use the ultra-short baseline response mode for guidance at long distances. The end effector uses an upward-looking optical camera to acquire images of the guidance light source. The center position of a single light source is obtained through a real-time target detection method based on deep learning, which improves the retrieval accuracy and ensures high accuracy and high success rate of the retrieval process.

[0056] 2. The present invention arranges a regular pentagonal recovery guide light source array inside the ice well. The side length can be adjusted according to the actual application requirements to achieve a suitable effective guidance depth. The pentagonal deployment increases the guidance area and provides more recovery orientations. Furthermore, the recovery device is placed on the surface of the ice well, which is conducive to monitoring the recovery status and ensures the high robustness and high safety of the recovery process.

[0057] 3. In this invention, a comb-shaped search is used to find the light source, a trajectory offset method is used to move towards the target position, a fuzzy PID controller is used to compensate for the heading angle, and an S-plane controller is used to control each thruster, so that the vehicle reaches the target position with the desired heading angle and speed, ensuring high robustness and high stability of the recovery process. Attached Figure Description

[0058] Figure 1 It is an underwater unmanned vehicle used for combined acoustic and optical guidance and recovery.

[0059] Figure 2 It is a regular pentagonal light source array used for optical guidance and recovery;

[0060] Figure 3These are images captured during the guided retrieval process;

[0061] Figure 4 This is a schematic diagram of the light source detection results during the guided recovery process;

[0062] Figure 5 This is a schematic diagram of the comb search simulation trajectory;

[0063] Figure 6 This is a flowchart of the acoustic-optical combined guidance and recovery method for underwater unmanned vehicles;

[0064] Figure 1 The underwater unmanned vehicle (UUV) suitable for acoustic-optical combined guidance and recovery has a typical underactuated form of rudder-propeller combined control. An optical camera 1 is installed above its bow, which is vertically mounted to the horizontal plane. An image processing unit 2 is installed in the sealed instrument compartment of the UUV. The optical camera 1 is controlled to focus and zoom via Ethernet TCP protocol. The image data of the optical camera 1 is acquired and processed via coaxial video signal cable 3. A motion control unit 4 is also installed. At the same time, the motion control unit 4 acquires data from altimeter 5 via RS232 serial port connection 6. Data exchange and communication with the image processing unit 2 are carried out via Ethernet UDP protocol 7. Lateral thrusters 8 and 9 are installed in the transverse pipes at the bow and stern to assist in steering. Main thrusters 10 are installed on both sides of the stern. An ultra-short baseline transponder 11 is installed inside, which transmits acoustic data to the motion control unit 4 via RS232 serial port connection 12.

[0065] Figure 2 Five rigid rod-shaped components 14 extend from the guide recovery platform 13, and five LED lights 15 arranged in a regular pentagon are installed on them as guide light sources to accurately mark the docking and recovery position. A part of the long side of the platform extends out to facilitate it to be placed on the ice surface to maintain stability. A shore control box 16 is placed on one side and an acoustic array 17 is installed on the other side. Detailed Implementation

[0066] Specific implementation method one: as follows Figures 1 to 6 As shown, a method for the combined acoustic and optical guidance and recovery of an underwater unmanned vehicle under ice specifically includes:

[0067] Step 1: The acoustic array near the shore-side recovery well sends an interrogation signal to the transponder inside the underwater unmanned vehicle. After receiving the interrogation signal, the transponder sends a response signal. The distance and angle are calculated by calculating the time difference and phase difference between sending the interrogation signal and receiving the response signal.

[0068] Step 2: When the underwater unmanned vehicle reaches the vicinity of the recovery platform, it encodes the depth information and sends it to the shore control box via underwater acoustic communication. The light source is then adjusted to the guidance brightness via the serial port server.

[0069] Step 3: The motion control unit of the underwater unmanned vehicle adjusts to a suitable search depth by collecting data from the altimeter. The vehicle performs a comb search through the motion control unit. The image processing unit acquires and detects images through the upward-looking optical camera. Proceed to Step 4.

[0070] Step 4: The image processing unit acquires the latest frame image for light source detection and calculates the position and orientation of the center of the guiding light source array relative to the vehicle.

[0071] Step 5: The motion control unit adjusts the motion direction of the underwater unmanned vehicle and moves it toward the target point based on the information provided by the image processing unit. The trajectory deviation is controlled by a two-layer tracking control system, which includes top-level heading control and bottom-level execution control. For heading control, a fuzzy PID controller is used to compensate for the heading angle. For execution control, an S-plane controller is used to operate the rudder and thrusters so that the vehicle reaches the target position with the desired heading angle and speed. After the adjustment is completed, proceed to step 6.

[0072] Step 6: The image processing unit acquires the latest frame image to detect light sources, calculates the center position and direction of the light source array, and determines the number of detected light sources. If the vehicle reaches the center of the light source array, proceed to step 7; otherwise, repeat step 4.

[0073] Step 7: The image processing unit acquires multiple consecutive frames of images for light source detection. If the number of light sources in the acquired 30 frames is 5, the motion control unit completes the vehicle's turning based on the S-plane controller using the information provided by the image processing unit. After the turning is completed, the bow of the vehicle is facing a vertex and perpendicular to the side opposite the vertex. Proceed to Step 8. If the number of light sources in the acquired 30 frames is less than 5, the motion control unit controls the vehicle to dive to a certain depth and encodes the depth information acquired by the altimeter and sends it to the shore control box via underwater acoustic communication. The light source is linearly increased to a suitable brightness. Proceed to Step 4.

[0074] Step 8: Move the center of the vehicle to the center of the light source array. After the adjustment is completed, the unmanned vehicle arrives at the predetermined docking and recovery position below the guidance and recovery platform. The motion control unit collects data from the altimeter and rises in stages. The underwater unmanned vehicle emerges from the ice well surface, and the operation ends.

[0075] Specific implementation method two: such as Figures 1 to 6 As shown, the distance between the acoustic array and the transponder signal is R, and the formula is as follows:

[0076] R = cΔT

[0077] Where ΔT is the time difference between the acoustic array and the transponder signal, c is the speed of sound in water, and ΔT is the time difference. The distance between the two acoustic units is l, and the phase difference of the received signals is... If the angle of incidence of the signal to the hydrophone is θ, then the relationship between the three is:

[0078]

[0079] The phase difference is obtained from the measured phase difference of the ultra-short baseline (USBL) line:

[0080]

[0081] The relative position of the carrier is calculated using the slant distance R and the azimuth angle θ. xyz Representing the acoustic array coordinate system, the relationship between the slant distance R and the carrier depth h is easily obtained as follows:

[0082] h=Rsinθ

[0083] The horizontal distance between the acoustic array and the carrier is:

[0084] s=Rcosθ

[0085] The mean square error of the horizontal distance relative to the slope distance R is:

[0086]

[0087] The motion control unit analyzes acoustic information and adjusts the vehicle's motion direction based on a two-layer tracking control system of fuzzy PID controller and S-plane controller, moving it to the vicinity of the shore-based acoustic array.

[0088] Specific implementation method three: such as Figures 1 to 6 As shown, the formula for calculating the brightness of a light source is:

[0089] I d =I0·e kd

[0090] I d I0 is the brightness of the LED light at a depth of d, k is a constant representing the absorption and scattering coefficients of water, and d is the depth.

[0091] The image processing unit acquires a frame of image through an optical camera and uses a pre-trained YOLOv8s object detection model to detect light sources. If the target confidence score (conf) is greater than a certain value, the target is considered a light source. The formula for calculating conf is as follows:

[0092] conf=(1-gr)+gr·CIoU

[0093] Where gr represents the probability of the presence of the light source target, and CIoU is the bounding box loss, calculated as follows:

[0094]

[0095] Where ρ is the distance between the center points of the predicted bounding box and the label box, c is the diagonal length of the minimum bounding rectangle of the predicted bounding box and the label box, ν is the aspect ratio similarity between the predicted bounding box and the label box, α is the influence factor of ν, and IoU is the ratio of the intersection area and the union area of ​​the predicted bounding box and the label box, as shown in the following formula:

[0096]

[0097] For targets identified as light sources, the NMS algorithm is used for filtering. First, the light source targets are sorted in descending order of confidence and placed into the input list. The highest value is removed and placed into the output list. The intersection-union ratio (IoU) of the removed highest value and the second highest value is calculated. If the IoU is greater than the threshold TH, the second highest value is removed. The above steps are repeated until the input list is empty, and the NMS processing is completed. The threshold TH is a predetermined value. Whether the guiding light source is found is determined by whether the threshold is reached. If not, proceed to step 3; if yes, proceed to step 4.

[0098] Specific implementation method four: such as Figures 1 to 6 As shown, the formulas for calculating the position and orientation of the center of the guidance light source array relative to the vehicle in step 4 are as follows:

[0099]

[0100] in These are the detection boxes (e.g.) Figure 4 The coordinates of the top left and bottom right points, The coordinates of the points within the detection frame are given. The number of detected light sources is determined; if the number is 1, its position and direction relative to the vehicle are calculated. When there are two light sources, the midpoint between the two light sources is the target point, and the calculation formula is:

[0101]

[0102] When the number of light sources is 3, the midpoint of the longest side of each light source is the target point, calculated using the following formula:

[0103]

[0104] in When the number of light sources is 4, the center position and direction of the light source array are calculated using the quadrilateral feature. Here, the intersection of the diagonal lines is assumed to be the center. Substituting the coordinates of the two points on the diagonal, the equation of the diagonal is obtained, as shown in the following formula:

[0105]

[0106] The intersection of the two diagonals is the target point. If the number of light sources is 5, the center position of the light source array is calculated using the characteristics of a regular pentagon. First, the distance from any point to the two farthest points is calculated, and then the coordinates of the midpoint between the two farthest points are calculated using the following formula:

[0107]

[0108] According to the properties of a regular pentagon, its center point is:

[0109]

[0110] The angle between the bow of the aircraft and the center of the detected light source array is:

[0111]

[0112] The distance between the center of the optical camera and the center of the light source array is:

[0113]

[0114] Among them The image center coordinates, h is the depth, and f is the focal length of the optical camera. s and θ are sent to the motion control unit. After successful transmission, proceed to step 5.

[0115] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A method for the recovery of an underwater unmanned vehicle using combined acoustic and optical guidance under ice, characterized in that, Specifically, it includes: Step 1: The acoustic array near the shore-side recovery well sends an interrogation signal to the transponder inside the underwater unmanned vehicle. After receiving the interrogation signal, the transponder sends a response signal. The distance and angle are calculated by calculating the time difference and phase difference between sending the interrogation signal and receiving the response signal. Step 2: When the underwater unmanned vehicle reaches the vicinity of the recovery platform, it encodes the depth information and sends it to the shore control box via underwater acoustic communication. The light source is adjusted to the guidance brightness via the serial port server. Five rigid rod-shaped components 14 extend from the guidance recovery platform 13, and five LED lights 15 arranged in a regular pentagon are installed on them as guidance light sources. Step 3: The motion control unit of the underwater unmanned vehicle adjusts to a suitable search depth by collecting data from the altimeter. The vehicle performs a comb search through the motion control unit. The image processing unit acquires and detects images through the upward-looking optical camera. Proceed to Step 4. Step 4: The image processing unit acquires the latest frame image for light source detection and calculates the position and orientation of the center of the guiding light source array relative to the vehicle. Step 5: The motion control unit adjusts the motion direction of the underwater unmanned vehicle and moves it toward the target point based on the information provided by the image processing unit. The trajectory deviation is controlled by a two-layer tracking control system, which includes top-level heading control and bottom-level execution control. For heading control, a fuzzy PID controller is used to compensate for the heading angle. For execution control, an S-plane controller is used to operate the rudder and thrusters so that the vehicle reaches the target position with the desired heading angle and speed. After the adjustment is completed, proceed to step 6. Step 6: The image processing unit acquires the latest frame image to detect light sources, calculates the center position and direction of the light source array, and determines the number of detected light sources. If the vehicle reaches the center of the light source array, proceed to step 7; otherwise, repeat step 4. Step 7: The image processing unit acquires multiple consecutive frames of images for light source detection. If the number of light sources in the acquired 30 frames is 5, the motion control unit completes the vehicle's turning based on the S-plane controller using the information provided by the image processing unit. After the turning is completed, the bow of the vehicle is facing a vertex and perpendicular to the side opposite the vertex. Proceed to Step 8. If the number of light sources in the acquired 30 frames is less than 5, the motion control unit controls the vehicle to dive to a certain depth and encodes the depth information acquired by the altimeter and sends it to the shore control box via underwater acoustic communication. The light source is linearly increased to a suitable brightness. Proceed to Step 4. Step 8: Move the center of the vehicle to the center of the light source array. After the adjustment is completed, the unmanned vehicle arrives at the predetermined docking and recovery position below the guidance and recovery platform. The motion control unit collects data from the altimeter and rises in stages. The underwater unmanned vehicle emerges from the ice well surface, and the operation ends.

2. The method for ice-based acoustic-optical combined guidance and recovery of an underwater unmanned vehicle according to claim 1, characterized in that, The distance between the acoustic array and the transponder signal is R The formula is as follows: in The time difference between the acoustic array and the transponder signal. c The speed of sound in water, The time difference is; the distance between the two acoustic units is... l The phase difference of the received signal is The angle of incidence of the signal to the hydrophone is The relationship between the three is as follows: The phase difference is obtained from the measured phase difference of the ultra-short baseline (USBL): By slope distance R and azimuth Calculate the relative position of the carrier. Representing the acoustic array coordinate system, the slant range is easily obtained. R With carrier depth h The relationship is: The horizontal distance between the acoustic array and the carrier is: The mean square error of the horizontal distance relative to the slope distance R is: The motion control unit analyzes acoustic information and adjusts the vehicle's motion direction based on a two-layer tracking control system of fuzzy PID controller and S-plane controller, moving it to the vicinity of the shore-based acoustic array.

3. The method for ice-based acoustic-optical combined guidance and recovery of an underwater unmanned vehicle according to claim 1, characterized in that, The formula for calculating the brightness of a light source is: Is the depth as d The brightness of the LED light It is the detectable brightness of the light source. k It is a constant representing the absorption and scattering coefficients of water. d It is depth; The image processing unit acquires a frame of image through an optical camera and uses a pre-trained YOLOv8s object detection model to detect the light source. If the target confidence level... conf If the value is greater than a certain threshold, the target is considered a light source. conf The calculation formula is as follows: in gr This represents the probability of the presence of a target light source. CIoU The bounding box loss is calculated using the following formula: in The distance between the center points of the predicted bounding box and the label bounding box. This is the diagonal length of the minimum bounding rectangle of the prediction box and the label box. To predict the aspect ratio similarity between the bounding box and the label box, for Influence factors IoU The ratio of the intersection area and the union area of ​​the prediction box and the label box is given by the following formula: For targets identified as light sources, the NMS algorithm is used for filtering. First, the light source targets are sorted in descending order of confidence and placed in the input list. The highest value is removed and placed in the output list. The intersection-union ratio (IUU) of the removed highest and second-highest values ​​is calculated. IoU ,like IoU> threshold TH If the second highest value is removed, repeat the above steps until the input list is empty. NMS processing is then complete, and the threshold is determined. TH The threshold is set to a predetermined value; whether the guide light source is detected is determined by whether the threshold is reached. If not, proceed to step 3; if yes, proceed to step 4.

4. The method for ice-based acoustic-optical combined guidance and recovery of an underwater unmanned vehicle according to claim 1, characterized in that, The formulas for calculating the position and orientation of the center of the guidance light source array relative to the vehicle in step 4 are as follows: in , These are the coordinates of the top-left and bottom-right points of the detection box, respectively. The coordinates of the points in the detection frame are given; the number of detected light sources is determined. If the number is 1, its position and direction relative to the vehicle are calculated, which is... When the number of light sources is 2, the midpoint between the two light sources is the target point, and the calculation formula is: When the number of light sources is 3, the midpoint of the longest side of each light source is the target point, calculated using the following formula: in When the number of light sources is 4, the center position and direction of the light source array are calculated using the quadrilateral feature. Here, the intersection of the diagonal lines is assumed to be the center. Substituting the coordinates of the two points on the diagonal, the equation of the diagonal is obtained, as shown in the following formula: The intersection of the two diagonals is the target point. If the number of light sources is 5, the center position of the light source array is calculated using the characteristics of a regular pentagon. First, the distance from any point to the two farthest points is calculated, and then the coordinates of the midpoint between the two farthest points are calculated using the following formula: According to the properties of a regular pentagon, its center point is: The angle between the bow of the aircraft and the center of the detected light source array is: The distance between the center of the optical camera and the center of the light source array is: Among them Image center coordinates h For depth, f For the focal length of the optical camera, , , Send to the motion control unit. After successful transmission, proceed to step 5.

Images