High-precision optical pose measurement method based on tower type recovery

Abstract

The invention provides a high-precision optical pose measurement method based on tower recovery, which adopts a close-range bidirectional optical cooperation mode, measures a mark point of a recovery tower through an optical detector on a rocket, and determines the distance and the pose of the rocket relative to the recovery tower; and the optical measurement system of the recovery tower synchronously detects and identifies the rocket and controls a capture arm and a buffer mechanism of the recovery tower to move, and optical measurement data of the recovery tower are fed back to the rocket and are integrated with optical measurement data on the rocket to control landing of the rocket. According to the invention, optical measurement and navigation are introduced into rocket tower type recovery, and the method has the advantages of high measurement precision, high data frequency, interference resistance and the like. The relative relation between the rocket and the recovery tower can be directly obtained through optical pose measurement, compared with inertial navigation, the coordinate conversion process can be reduced, and the advantage of no accumulated error is achieved. Compared with satellite navigation, optical pose measurement does not depend on external signals, and higher independence and electromagnetic resistance are achieved.

Description

Technical Field

[0001] The present invention relates to the field of measurement and testing technology, and in particular to a high-precision optical posture measurement method based on tower-type recovery. Background Art

[0002] The US Space Shuttle was the world's first reusable space vehicle, capable of traveling between Earth and space. It utilized a vertical launch and horizontal landing method, relying on its aerodynamic shape for lift, and a glide landing similar to an airplane during the return landing phase. However, high launch costs, significant safety risks, and a high structure-to-weight ratio limited the space shuttle's development. Vertical launch and recovery have become key developments in launch vehicle recovery. The Falcon 9 is currently the only launch vehicle to achieve vertical recovery of an orbital stage. It utilizes a support leg recovery method, with four support legs installed at the base of the first stage. These legs deploy before landing and support the rocket's weight during landing. Due to the significant impact on the rocket body during landing, maintenance and repair of the support legs is time-consuming. Furthermore, due to their strength design, the support legs significantly increase the rocket's structural weight. A new, more technically challenging method, known as a "chopstick gripper," has been proposed since the support leg recovery method. In this method, a recovery tower clamps the rocket, eliminating the need for support legs and associated servo systems. Compared with using rocket support legs for recovery, tower recovery can reduce the weight of the rocket structure and increase reliability, shorten the post-recovery maintenance time, increase the payload weight, and help improve the rocket's launch efficiency. However, tower recovery is more difficult and requires much higher positioning accuracy for the rocket than recovery leg recovery.

[0003] The position and attitude information of a rocket during recovery and landing is a crucial parameter in the reentry phase, directly impacting the success of the recovery. Therefore, obtaining high-precision, real-time position and attitude information is a key technical challenge facing rocket recovery. Tower-based recovery requires consideration of tower structural interference, and the recovery accuracy requirements are far greater than those of leg-based recovery. Summary of the Invention

[0004] The present invention aims to solve the problem of positioning accuracy in rocket recovery and provides a high-precision optical posture measurement method based on tower-type recovery. It adopts a close-range two-way optical cooperation method, and measures the landmark points of the recovery tower through the optical detector on the rocket to determine the distance and posture of the rocket relative to the recovery tower; the optical measurement system of the recovery tower synchronously detects and identifies the rocket, controls the movement of the recovery tower capture arm and buffer mechanism, and the optical measurement data of the recovery tower is also fed back to the rocket and integrated with the optical measurement data on the rocket to control the landing of the rocket.

[0005] The present invention provides a high-precision optical posture measurement method based on tower-type recovery, comprising the following steps:

[0006] S1. Install the recovery tower optical measurement system, the onboard optical measurement system, the rocket control system, and the recovery tower control system on the recovery tower and the rocket, respectively. Connect the grid fins and landing hook to the outside of the rocket, and connect the first capture arm, the second capture arm, and the buffer mechanism to the recovery tower.

[0007] The recovery tower optical measurement system includes a first optical detector connected to the recovery tower and its surroundings, a first processor, and an onboard beacon connected to the rocket;

[0008] The onboard optical measurement system includes a second optical detector attached to the rocket, an image processor, and a cooperative beacon attached to the recovery pylon;

[0009] S2. During rocket recovery, when the rocket reaches a specified distance from the recovery tower, the recovery tower optical measurement system and the onboard optical measurement system begin operating. The rocket is always within the field of view of the first optical detector, and the recovery tower is within the field of view of the second optical detector. The onboard optical measurement system establishes an optical measurement channel with the recovery tower optical measurement system.

[0010] S3, the second optical detector and the image processor transmit the pose solution results obtained by the cooperative beacon to the rocket control system, determine the distance and attitude of the rocket relative to the recovery tower, and adjust the position through the rocket power and attitude control system to gradually approach the recovery tower;

[0011] S4. The first optical detector and the first processor synchronously measure the rocket's posture. By identifying the rocket body and the beacon on the rocket, the posture is calculated to obtain the position deviation between the landing hook and the predetermined point. The feedback is fed back to the rocket so that the rocket gradually approaches the recovery tower.

[0012] S5. The optical measurement system of the recovery tower simultaneously transmits the measurement data to the recovery tower servo system to adjust the angles of the first capture arm and the second capture arm and to make the buffer mechanism match the current position of the rocket;

[0013] S6. When the first capture arm and the second capture arm provide rigid support for the landing hook, the rocket recovery is completed, and a high-precision optical posture measurement method based on tower-type recovery is completed.

[0014] The high-precision optical posture measurement method based on tower-type recovery described in the present invention is preferably as follows: in step S1, the beacon on the arrow is arranged axially and radially positioned at the landing hook;

[0015] The first optical detectors consist of at least two groups, which perform image processing and pose calculation by identifying beacons on the arrows. The first optical detectors have a built-in gimbal, and multiple groups are arranged to cover all airspace around the recovery tower.

[0016] The cooperative beacons are arranged in three dimensions;

[0017] The grid fins perform aerodynamic attitude adjustment of the rocket;

[0018] The upper ends of the first capture arm and the second capture arm are both connected to the buffer mechanism;

[0019] The recovery tower is an integrated launch-recovery device that can support the rocket and provide propellant filling and electrical connections before launch. The rocket returns to the recovery tower after launching and completing its flight mission.

[0020] The high-precision optical attitude measurement method based on tower-type recovery described in the present invention is preferably configured such that the number of grid vanes is four, evenly connected around the rocket body in a circular manner, and aerodynamic attitude adjustment is performed;

[0021] There are two landing hooks, both located under the grid wing;

[0022] The arrow beacons include a first arrow beacon and a second arrow beacon located sequentially below the landing hook, and the number of the first arrow beacon and the second arrow beacon are both two;

[0023] The cooperative beacons include two groups of beacons arranged along a 180° angle; the cooperative beacons include a first cooperative beacon, a second cooperative beacon, a third cooperative beacon, a fourth cooperative beacon, a fifth cooperative beacon, and a sixth cooperative beacon; the first cooperative beacon, the second cooperative beacon, and the third cooperative beacon are all connected to the top of the recovery tower and identified according to a triangle scheme; the fourth cooperative beacon and the fifth cooperative beacon are respectively arranged between the first capture arm and the tower root and the second capture arm and the tower root, and do not change position with the movement of the first capture arm and the second capture arm; the sixth cooperative beacon is connected to the lower part of the first capture arm or the second capture arm.

[0024] The present invention describes a high-precision optical posture measurement method based on tower-type recovery, as a preferred method. In step S2, the optical measurement system on the rocket is in a dynamic-photographing-static mode, and a second optical detector installed on the moving rocket body is used to photograph a cooperative beacon stationary on the ground. A simulation analysis is performed based on the rocket's flight trajectory, and the camera field of view angle and installation angle of the second optical detector are selected so that the recovery tower is always within the field of view of the second optical detector during the tower-type recovery mission; the recovery tower optical measurement system is in a static-photographing-dynamic mode, and a first optical detector installed on the stationary recovery tower is used to photograph the moving rocket.

[0025] The high-precision optical attitude measurement method based on tower-type recovery described in the present invention is preferably configured such that, in step S2, when the rocket reaches the vicinity of the recovery tower and the distance is 20 times the length of the first stage of the rocket, the optical measurement system of the recovery tower and the optical measurement system on the rocket start working;

[0026] In step S3, the image processor filters, performs threshold segmentation, target recognition and extraction on the optical image obtained by the second optical detector to obtain optical image information, and then performs posture solution to obtain the position and attitude of the rocket relative to the recovery tower and calculate the rocket motion parameters; the rocket motion parameters are navigation parameters, which are output to the rocket control system, adjust the engine thrust and direction, and use online guidance to control the rocket to gradually approach the recovery tower.

[0027] In the high-precision optical posture measurement method based on tower-type recovery described in the present invention, as a preferred embodiment, the posture solution data of the first optical detector in step S4 is divided into two paths, one path is transmitted to the second optical detector as a feedback signal to correct the error of the rocket optical system; the other path is used to adjust the recovery tower, including changing the angle between the first capture arm and the second capture arm and controlling the three-axial movement of the buffer mechanism;

[0028] The height positions of the first and second capture arms can be set in advance according to the size of the rocket, and the angle between the first and second capture arms is adjusted based on optical measurement data to complete the capture. The buffer mechanism is composed of a metal body and a non-metallic material wrapped on the surface, which provides a buffer when the landing hook is mounted on the first and second capture arms.

[0029] In step S5, the first capture arm and the second capture arm provide rigid support for the landing hook and trigger a feedback signal. The rocket shuts down the power system and attitude control system, and the recovery tower capture arm stops working.

[0030] The high-precision optical posture measurement method based on tower-type recovery described in the present invention, as a preferred embodiment, further includes step SA between step S5 and step S6:

[0031] SA, the recovery tower optical measurement system, and the onboard optical measurement system all use a deep learning network architecture. The test and actual recovery data are used as training samples. The deep learning network is used to learn and save the obtained data parameters to the image processor for subsequent recovery. At the same time, the learning results based on the rocket's motion state will also be used to adjust the movement of the first capture arm, the second capture arm, and the buffer mechanism to form a correction network. During the recovery process, the flight measurement results will be compared with the database to perform data fusion and optimize the measurement data.

[0032] The high-precision optical pose measurement method based on tower-type recovery described in the present invention is preferably configured such that the deep learning network architecture includes an input end, a backbone network, a neck, and a detection head, wherein the backbone network includes a target recognition and segmentation network;

[0033] The target recognition and segmentation network performs a weighted fusion of the results of classification and regression to determine whether the sample is positive or negative; the output of the target recognition and segmentation network only has a classification branch and a regression branch;

[0034] The post-processing operations of the target recognition and segmentation network include: scale restoration, non-maximum suppression, and instance segmentation based on recognition location.

[0035] In the high-precision optical pose measurement method based on tower-type recovery described in the present invention, as a preferred embodiment, the loss function of the target recognition and segmentation network adopts the TaskAlignedAssigner strategy:

[0036] t=s α +u β ;

[0037] Among them, s is the prediction score corresponding to the labeled category, u is the intersection-over-union ratio of the predicted box and the true box, and the degree of alignment is measured by multiplying s and u; α and β are weight parameters, and t is the weighted judgment metric value;

[0038] The loss function of the classification branch is the BCE loss function, and the loss function of the regression branch is the CIOU loss function and the DFL loss function;

[0039] The deep learning network architecture uses an iterative training method, training through a variety of environmental samples, and continuously using the latest collected data to train and optimize the network as experiments and tests proceed.

[0040] The present invention proposes a bidirectional synchronous optical posture measurement method for tower recovery, which belongs to the field of navigation, positioning and control technology. It is a high-precision posture measurement method in rocket recovery navigation technology. This method not only has the advantages of no cumulative error and high positioning accuracy, but also can synchronize and coordinate the movement of the rocket and tower capture arm in real time, thereby improving the reliability of recovery.

[0041] The present invention is mainly used in the return landing phase of commercial rocket recovery missions. Based on the optical measurement equipment on the rocket and the recovery tower, a high-precision optical posture measurement method for tower-type rocket recovery is proposed. This method adopts a close-range bidirectional optical cooperation method. The optical detector on the rocket measures the landmark points of the recovery tower to determine the distance and posture of the rocket relative to the recovery tower. The optical measurement system of the recovery tower synchronously detects and identifies the rocket, controls the movement of the recovery tower's capture arm and buffer mechanism, and the optical measurement data of the recovery tower is also fed back to the rocket. It is integrated with the optical measurement data on the rocket to control the rocket's landing. This method is mainly used in tower-type recovery of vertically landing rockets.

[0042] The brief steps for optical posture measurement during rocket recovery are as follows: the onboard optical equipment establishes an optical measurement channel with the tower, using close-range optical recognition and high-precision posture algorithms to achieve rocket positioning; the onboard optical equipment transmits the calculation results to the control system, which adjusts the posture through the rocket's power and attitude control systems; the recovery tower's optical equipment measures the rocket, and by identifying the optical cooperation points arranged on the rocket and the rocket, it performs high-precision posture calculations to obtain the position deviation between the rocket's landing hook and the predetermined point, and feeds this back to the rocket. The recovery tower's optical measurement data is transmitted to the recovery tower's servo system to adjust the angle between the two capture arms. The buffer structure installed on the capture arm can move in three directions in conjunction with the rocket's current posture. The recovery tower's capture arm provides rigid support for the rocket's landing hook, completing the recovery. The optical posture measurement system introduces artificial intelligence, and can use test and actual recovery data as training samples for learning through a deep learning network.

[0043] The high-precision real-time posture measurement method of the present invention is as follows: when the rocket approaches the recovery tower, a high-resolution optical detector on the rocket is used to measure the optical cooperation point arranged on the recovery tower;

[0044] Arrange cooperative beacons on the recovery tower. Due to the spatial layout of the recovery tower, the beacons can be arranged in three dimensions, which is better than the two-dimensional pattern on the ground;

[0045] The recovery tower optical measurement system requires the placement of optical cooperation beacons on the rocket. Optical beacons are arranged axially on the rocket and radially positioned at the rocket's recovery hook, with two sets of beacons arranged at an angle of 180 degrees. The recovery tower is equipped with multiple sets of optical detectors, which identify the optical beacons on the rocket and perform image processing and pose calculation. The data measured by the recovery tower optical measurement system is transmitted back to the rocket via the ground telemetry system, and the recovery tower arm is adjusted to ensure support and recovery.

[0046] The optical measurement data of the recovery tower is used by the servo system to adjust the position of the capture arm and the buffer mechanism, adjust the capture arm angle, and move the buffer mechanism in three directions. In accordance with the actual position of the rocket, the recovery tower capture arm and the rocket recovery landing hook are in rigid contact and support the weight of the rocket, thus completing the recovery.

[0047] After the recovery tower capture arm is rigidly connected to the rocket landing hook, the feedback signal is triggered, the rocket shuts down the power system and attitude control system, and the recovery tower capture arm stops working.

[0048] The present invention has the following advantages:

[0049] (1) The present invention introduces optical measurement and navigation into rocket tower recovery, which has the advantages of high measurement accuracy, high data frequency, and anti-interference. Optical position measurement can directly obtain the relative relationship between the rocket and the recovery tower. Compared with inertial navigation, it can reduce the coordinate conversion process and has the advantage of no cumulative error. Compared with satellite navigation, optical position measurement does not rely on external signals and has stronger independence and anti-electromagnetic capabilities.

[0050] (2) The present invention has bidirectional measurement capabilities. In addition to the optical equipment on the rocket identifying and measuring the recovery tower, the optical equipment on the tower can also measure the rocket. Based on this bidirectional measurement principle, the onboard measurement information guides the rocket's adjustments, while the tower's optical measurement information controls the tower's capture arm and buffer structure adjustments. This provides greater flexibility and safety compared to support-leg recovery, where the recovery site is fixed and only the rocket can be adjusted.

[0051] (3) The present invention has a synchronous measurement function. On the basis of bidirectional measurement, the optical measurement results of the tower can be transmitted to the rocket as rocket measurement information.

[0052] (4) The present invention has the advantage of spatially arranging beacon points. If optical measurement methods are introduced into a landing leg recovery rocket, the ground cooperative beacon is usually a two-dimensional scheme, while it is easy to realize spatial beacon arrangement on the tower, thereby improving the flexibility of beacon scheme design and image recognition accuracy.

[0053] (5) The optical measurement data of the tower of the present invention is used to control the movement of the capture arm buffer mechanism. Through the three-axis movement, the actual landing and recovery accuracy can be made to have a certain degree of fault tolerance.

[0054] (6) The present invention introduces artificial intelligence into the optical measurement system of the recovery tower. The test and actual recovery data can be used as training samples. Through deep learning network learning, the image recognition rate can be greatly increased. As the number of recovery samples increases, the accuracy gradually improves. At the same time, the learning results based on the motion state of the rocket will also be used to adjust the motion of the capture arm and the buffer mechanism to form a correction network and enhance the motion stability of the servo mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a high-precision optical pose measurement method based on tower-type recovery;

[0056] Figure 2 A schematic diagram of the grid fin, landing hook, and on-rocket beacon based on a high-precision optical attitude measurement method for tower-type recovery.

[0057] Figure 3 A schematic diagram of the composition of an onboard optical measurement system based on a high-precision optical attitude measurement method using tower-type recovery.

[0058] Figure 4 A schematic diagram of a recovery tower and capture arm based on a high-precision optical pose measurement method for tower-type recovery;

[0059] Figure 5 A schematic diagram of the movement direction of a buffer mechanism based on a high-precision optical posture measurement method for tower-type recovery;

[0060] Figure 6 A schematic diagram of the recovery tower beacon layout based on a high-precision optical pose measurement method for tower-type recovery;

[0061] Figure 7 This is a workflow diagram of a rocket-borne measurement system using a high-precision optical posture measurement method based on tower-type recovery.

[0062] Reference numerals:

[0063] 1. Recovery tower optical measurement system; 11. On-rocket beacon; 111. First on-rocket beacon; 112. Second on-rocket beacon; 2. On-rocket optical measurement system; 21. Second optical detector; 22. Image processor; 23. Cooperation beacon; 231. First cooperation beacon; 232. Second cooperation beacon; 233. Third cooperation beacon; 234. Fourth cooperation beacon; 235. Fifth cooperation beacon; 236. Sixth cooperation beacon; 3. Grid wing; 4. Landing hook; 5. First capture arm; 6. Second capture arm; 7. Buffer mechanism. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0065] Example 1

[0066] like Figures 1 to 7 As shown, a high-precision optical posture measurement method based on tower-type recovery includes the following steps:

[0067] S1. Install the recovery tower optical measurement system 1, the onboard optical measurement system 2, the rocket control system, and the recovery tower control system on the recovery tower and the rocket, respectively. Connect the grid fin 3 and the landing hook 4 to the outside of the rocket, and connect the first capture arm 5, the second capture arm 6, and the buffer mechanism 7 to the recovery tower.

[0068] The recovery tower optical measurement system 1 includes a first optical detector connected to the recovery tower and its surroundings, a first processor, and an on-arrow beacon 11 connected to the rocket;

[0069] The onboard optical measurement system 2 includes a second optical detector 21 connected to the rocket, an image processor 22, and a cooperation beacon 23 connected to the recovery tower;

[0070] S2. During rocket recovery, when the rocket reaches a specified distance from the recovery tower, the recovery tower optical measurement system 1 and the onboard optical measurement system 2 begin to operate. The rocket is always within the field of view of the first optical detector, and the recovery tower is within the field of view of the second optical detector 21. The onboard optical measurement system 2 establishes an optical measurement channel with the recovery tower optical measurement system 1.

[0071] S3, the second optical detector 21 and the image processor 22 transmit the posture solution results obtained based on the cooperative beacon 23 to the rocket control system, determine the distance and posture of the rocket relative to the recovery tower, and adjust the posture through the rocket power and attitude control system to gradually approach the recovery tower;

[0072] S4. The first optical detector and the first processor synchronously measure the rocket's posture. By identifying the rocket body and the beacon 11 on the rocket, the posture is calculated to obtain the position deviation between the landing hook 4 and the predetermined point, and the feedback is fed back to the rocket so that the rocket gradually approaches the recovery tower.

[0073] S5. The optical measurement system 1 of the recovery tower simultaneously transmits the measurement data to the servo system of the recovery tower to adjust the angle between the first capture arm 5 and the second capture arm 6 and to make the buffer mechanism 7 match the current posture of the rocket;

[0074] SA, the recovery tower optical measurement system 1 and the onboard optical measurement system 2 all use a deep learning network architecture. The test and actual recovery data are used as training samples. The deep learning network is used to learn and save the obtained data parameters to the image processor 22 for subsequent recovery. At the same time, the learning results based on the rocket's motion state will also be used to adjust the motion of the first capture arm 5, the second capture arm 6 and the buffer mechanism 7 to form a correction network. During the recovery process, the flight measurement results are compared with the database, data fusion is performed, and the measurement data is optimized.

[0075] S6. When the first capture arm 5 and the second capture arm 6 provide rigid support to the landing hook 4, the rocket recovery is completed, and a high-precision optical posture measurement method based on tower-type recovery is completed.

[0076] A recovery tower rocket recovery primarily consists of two components: the rocket (the vehicle) and the ground-based recovery tower. Rocket recovery involves both the first-stage recovery and the final stage recovery. After the first-stage rocket ejects the final stage from the dense atmosphere, it uses reverse thrust from the rocket engine and aerodynamic deceleration to reduce its speed. The three surfaces of the grid fins are manipulated to adjust the flight attitude, gradually decelerating and returning to a landing. The recovery tower is a single-stage launch and recovery device that supports the rocket before launch and provides auxiliary functions such as propellant filling and electrical connections. After the rocket completes its mission, it returns to the recovery tower.

[0077] When the rocket reaches the vicinity of the recovery tower, at a distance of about 20 times the length of the first stage of the rocket, the optical measurement system 1 of the recovery tower and the optical measurement system 2 on the rocket start to work and feed back the measurement data to the rocket control system and the tower servo system. At this time, the speed of the rocket is relatively low and the pitch angle is about 90 degrees. The optical measurement system 2 on the rocket is in the "dynamic shooting static" mode, that is, the optical device 21 on the rocket is installed on a moving carrier to shoot targets stationary on the ground. According to the simulation analysis of the flight trajectory, the field of view angle and installation angle of the camera 21 are selected to ensure that the recovery tower is always within the field of view of the camera 21 during the mission. The optical measurement system 1 of the recovery tower is in the "static shooting dynamic" mode, that is, the optical device of the recovery tower is installed on a stationary ground facility to shoot the moving rocket. The optical system of the recovery tower (the first optical detector) has a built-in pan-tilt head, which is arranged in multiple groups to cover all airspace around the recovery tower and can capture the rocket in the landing phase during the mission.

[0078] The rocket's optical measurement system 2 consists of an optical detector 21 mounted on the rocket, an image processor 22, and a cooperative beacon 23 mounted on the launch tower. Optical measurement achieves higher accuracy by photographing calibrated beacons 23. The optical cooperative beacons 23 are arranged on the recovery tower. Due to the tower's spatial layout, the cooperative beacons 23 can be arranged in three dimensions, achieving higher accuracy and speed than two-dimensional ground-based patterns. The multiple optical detectors 21 mounted on the rocket can perform high-resolution, high-frame-rate optical imaging and transmit the images to the onboard processor 22. The onboard processor 22 performs operations such as filtering, threshold segmentation, and target identification and extraction on the optical images. The extracted optical image information is used for pose calculation, which can calculate the rocket's position and attitude relative to the recovery tower and, from this, the complete rocket motion parameters. The rocket motion parameters calculated by the optical processor 22 are navigation parameters with an output frequency greater than 30 Hz. These parameters are fed back to the rocket control system, which adjusts engine thrust and direction, using online guidance to gradually guide the rocket toward the recovery tower.

[0079] The recovery tower optical measurement system 1 consists of an optical detector (first optical detector) installed on the recovery tower, a processor, and a beacon 11 installed on the rocket. During the working period of the optical measurement system, the rocket is always within the field of view of the tower optical detector. The tower optical measurement system calculates the relative position of the rocket body by capturing it. To improve reliability, the beacon arranged on the rocket is also identified. Figure 2 The figure shows a schematic diagram of a rocket and landing hook 4. The rocket body is a cylindrical structure with four grid fins 3 mounted on its upper surface for aerodynamic attitude adjustment. There are two rocket landing hooks 4, which are installed at a radial angle of 180 degrees and are located below the grid fins 3. Two optical beacons 11 (the first beacon 111 and the second beacon 112) are arranged below them. The engine is installed below the rocket body. The simplest and most efficient beacon layout used on the tower is Figure 6 The illustrated arrangement includes six beacons (a first cooperative beacon 231, a second cooperative beacon 232, a third cooperative beacon 233, a fourth cooperative beacon 234, a fifth cooperative beacon 235, and a sixth cooperative beacon 236). First, second, and third cooperative beacons 231, 232, and 233 are located at the top of the recovery tower and can be identified using a triangular pattern. Fourth and fifth cooperative beacons 234, 235 are located between the capture arm and the base of the tower and do not change position with the movement of the capture arm. Sixth cooperative beacon 236 is located below one of the capture arms. This minimizes the number of beacon points in the spatial arrangement.

[0080] The position and posture data of the tower optical measurement system are divided into two channels. One channel is transmitted to the rocket's optical system as a feedback signal to correct the errors of the rocket's optical system. The other channel is used to adjust the recovery tower, including changing the angle of the recovery tower's capture arm (first capture arm 5, second capture arm 6) and the three-axis movement of the capture arm buffer mechanism 7. The recovery tower schematic is shown in Figure 4 As shown, the height position of the recovery tower capture arm can be set in advance according to the size of the rocket, and the angle between the first capture arm 5 and the second capture arm 6 is adjusted according to the optical measurement data to complete the capture. The capture arm buffer mechanism 7 is composed of a metal body and a non-metallic material wrapped on the surface, which is used to buffer the rocket landing hook 4 when it is mounted on the capture arm. The schematic diagram of the movement direction of the buffer mechanism 7 is shown in FIG. Figure 5 As shown, by controlling the operation of the buffer mechanism 7 in three axes through the recovery tower servo system, the rocket landing buffer can be effectively performed while reducing the rocket position accuracy requirement and improving safety.

[0081] A deep learning network is also used on the rocket and in the recovery tower image processor 22. Using early-stage tests and actual recovery data as training samples, a database is built. Learning is performed through the deep learning network, and the acquired data parameters are saved to the image processor 22 for subsequent recovery. Simultaneously, the learned results based on the rocket's motion state are used to adjust the motion of the first capture arm 5, the second capture arm 6, and the buffer mechanism 7, forming a correction network. During the recovery process, flight measurement results are compared with the database, and data fusion is performed to optimize the measurement data.

[0082] The deep learning network architecture constructed by this method can be divided into four parts: input end, backbone network, neck and detection head. The overall network adopts a lightweight design, making it suitable for carrier embedded computing environment.

[0083] The loss function of the target recognition and segmentation network is proposed to adopt the TaskAlignedAssigner strategy, as shown in the following formula, which performs a weighted fusion of the results obtained by classification and regression to determine whether the sample is positive or negative.

[0084] t=s α +u β

[0085] Where s is the predicted score corresponding to the labeled category, and u is the intersection-over-union ratio (IoU) between the predicted box and the ground-truth box. Multiplying the two measures the degree of alignment. The network output lacks a confidence branch, leaving only a classification branch and a regression branch. The classification loss uses the BCE loss function, while the regression branch uses the CIOU loss and the DFL loss function.

[0086] The post-processing operations of the target recognition and segmentation network include: scale restoration, non-maximum suppression, and instance segmentation based on the recognition location. At the same time, in order to fully ensure that the falling area can be accurately identified and segmented under different weather and environmental conditions, this solution will also adopt an iterative training method. On the one hand, as many environmental samples as possible are collected for training, on the other hand, as the experiments and tests proceed, the latest collected data is continuously used for training, so that the network can be continuously optimized and robust.

[0087] When the first capture arm 5 and the second capture arm 6 of the recovery tower are in rigid contact with the rocket landing hook 4 and support the weight of the rocket, the recovery is completed. After the recovery tower capture arm is rigidly connected to the rocket landing hook 4, the feedback signal is triggered, the rocket shuts down the power system and attitude control system, and the recovery tower capture arm stops working.

[0088] Through theoretical calculations and simulation analysis, the high-precision optical pose measurement method proposed in this paper can be effectively applied to rocket pylon recovery. At a distance of 200 meters, the relative measurement error can reach 0.29%, and at a distance of 50 meters, the relative measurement error can reach 0.1%, which meets the accuracy requirements for landing and recovery.

[0089] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A high-precision optical posture measurement method based on tower-type recovery, characterized by: The following steps are involved: S1. Installing a recovery tower optical measurement system (1), an onboard optical measurement system (2), a rocket control system, and a recovery tower control system on the recovery tower and the rocket, respectively; connecting a grid wing (3) and a landing hook (4) to the outside of the rocket; and connecting a first capture arm (5), a second capture arm (6), and a buffer mechanism (7) to the recovery tower; The recovery tower optical measurement system (1) comprises a first optical detector connected to the recovery tower and its surroundings, a first processor, and an on-rocket beacon (11) connected to the rocket; The onboard optical measurement system (2) comprises a second optical detector (21) connected to the rocket, an image processor (22), and a cooperation beacon (23) connected to the recovery tower; S2, when the rocket is recovered, when the rocket reaches a specified distance from the recovery tower, the recovery tower optical measurement system (1) and the on-rocket optical measurement system (2) start working, the rocket is always in the field of view of the first optical detector, the recovery tower is in the field of view of the second optical detector (21), and the on-rocket optical measurement system (2) establishes an optical measurement channel with the recovery tower optical measurement system (1); S3, the second optical detector (21) and the image processor (22) transmit the posture solution result obtained according to the cooperation beacon (23) to the rocket control system, determine the distance and posture of the rocket relative to the recovery tower, and adjust the posture through the rocket power and attitude control system to make the rocket gradually approach the recovery tower; S4, the first optical detector and the first processor synchronously measure the rocket's posture, calculate the posture of the rocket body and the beacon (11) on the rocket, obtain the position deviation between the landing hook (4) and the predetermined point, and feed it back to the rocket so that the rocket gradually approaches the recovery tower; S5, the recovery tower optical measurement system (1) simultaneously transmits the measurement data to the recovery tower servo system to adjust the angle between the first capture arm (5) and the second capture arm (6) and to make the buffer mechanism (7) match the current posture of the rocket; S6. When the first capture arm (5) and the second capture arm (6) rigidly support the landing hook (4), the rocket recovery is completed, and a high-precision optical posture measurement method based on tower-type recovery is completed.

2. The high-precision optical posture measurement method based on tower-type recovery according to claim 1 is characterized in that: In step S1, the arrow beacon (11) is arranged in the axial direction and radially positioned at the landing hook (4); The first optical detectors are at least two groups, and perform image processing and posture calculation by identifying the beacon (11) on the arrow; the first optical detectors have a built-in pan / tilt platform, and are arranged in multiple groups to cover all airspaces around the recovery tower; The cooperation beacons (23) are arranged in three dimensions; The grid wing (3) performs aerodynamic attitude adjustment of the rocket; The upper ends of the first capture arm (5) and the second capture arm (6) are both connected to the buffer mechanism (7); The recovery tower is an integrated launch-recovery device that can support the rocket and provide propellant filling and electrical connections before launch. The rocket returns to the recovery tower after launching and completing its flight mission.

3. The high-precision optical posture measurement method based on tower-type recovery according to claim 2 is characterized in that: The number of the grid wings (3) is four, and they are evenly connected around the rocket body in a circle to perform aerodynamic attitude adjustment; There are two landing hooks (4), both of which are located below the grid wing (3); The arrow beacon (11) comprises a first arrow beacon (111) and a second arrow beacon (112) sequentially located below the landing hook (4), and the number of the first arrow beacon (111) and the second arrow beacon (112) are both two; The cooperative beacon (23) includes two groups of beacons arranged along an angle of 180 degrees; the cooperative beacon (23) includes a first cooperative beacon (231), a second cooperative beacon (232), a third cooperative beacon (233), a fourth cooperative beacon (234), a fifth cooperative beacon (235), and a sixth cooperative beacon (236); the first cooperative beacon (231), the second cooperative beacon (232), and the third cooperative beacon (233) are all connected to the top of the recovery tower and identified according to a triangle scheme; the fourth cooperative beacon (234) and the fifth cooperative beacon (235) are respectively arranged between the first capture arm (5) and the tower root, and the second capture arm (6) and the tower root, and do not change position with the movement of the first capture arm (5) and the second capture arm (6); the sixth cooperative beacon (236) is connected to the lower part of the first capture arm or the second capture arm (6).

4. The high-precision optical posture measurement method based on tower-type recovery according to claim 1 is characterized in that: In step S2, the optical measurement system (2) on the rocket is in a dynamic-photography-static mode, and the cooperative beacon (23) stationary on the ground is photographed by the second optical detector (21) installed on the moving rocket body, and simulation analysis is performed according to the rocket flight trajectory. By selecting the camera field of view angle and installation angle of the second optical detector (21), the recovery tower is always in the field of view of the second optical detector (21) during the tower-type recovery mission; the recovery tower optical measurement system (1) is in a static-photography-dynamic mode, and the moving rocket is photographed by the first optical detector installed on the stationary recovery tower.

5. The high-precision optical posture measurement method based on tower-type recovery according to claim 1 is characterized in that: In step S2, when the rocket reaches the vicinity of the recovery tower and the distance is 20 times the length of the first stage of the rocket, the recovery tower optical measurement system (1) and the onboard optical measurement system (2) start working; In step S3, the image processor (22) filters, thresholds, segments, identifies and extracts the optical image obtained by the second optical detector (21) to obtain optical image information, and then performs posture calculation to obtain the position and attitude of the rocket relative to the recovery tower and calculates the rocket motion parameters; the rocket motion parameters are navigation parameters, which are output to the rocket control system, adjust the engine thrust and direction, and use online guidance to control the rocket to gradually approach the recovery tower.

6. The high-precision optical posture measurement method based on tower-type recovery according to claim 1 is characterized in that: In step S4, the position and posture solution data of the first optical detector are divided into two paths, one path is transmitted to the second optical detector (21) as a feedback signal to correct the error of the rocket optical system; the other path is used for adjusting the recovery tower, including changing the angle between the first capture arm (5) and the second capture arm (6) and controlling the three-axial movement of the buffer mechanism (7); The height positions of the first capture arm (5) and the second capture arm (6) can be set in advance according to the size of the rocket, and the angles between the first capture arm (5) and the second capture arm (6) are adjusted according to optical measurement data to complete the capture; the buffer mechanism (7) is composed of a metal body and a non-metallic material wrapped on the surface, and performs buffering when the landing hook (4) is mounted on the first capture arm (5) and the second capture arm (6); In step S5, the first capture arm (5) and the second capture arm (6) rigidly support the landing hook (4) and trigger a feedback signal, the rocket shuts down the power system and attitude control system, and the recovery tower capture arm stops working.

7. The high-precision optical posture measurement method based on tower-type recovery according to claim 1 is characterized in that: Step SA is also included between step S5 and step S6: SA, the recovery tower optical measurement system (1) and the onboard optical measurement system (2) all use a deep learning network architecture, using test and actual recovery data as training samples, learning through a deep learning network and saving the obtained data parameters to the image processor (22) for subsequent recovery use. At the same time, the learning results based on the rocket's motion state will also be used to adjust the motion of the first capture arm (5), the second capture arm (6) and the buffer mechanism (7) to form a correction network; during the recovery process, the flight measurement results are compared with the database, data fusion is performed, and the measurement data is optimized.

8. The high-precision optical posture measurement method based on tower-type recovery according to claim 7 is characterized in that: The deep learning network architecture includes an input end, a backbone network, a neck, and a detection head, wherein the backbone network includes an object recognition and segmentation network; The target recognition and segmentation network performs weighted fusion of the results obtained from classification and regression to determine whether the sample is positive or negative; The output of the target recognition and segmentation network only has a classification branch and a regression branch; The post-processing operations of the target recognition and segmentation network include: scale restoration, non-maximum suppression, and instance segmentation based on recognition location.

9. The high-precision optical posture measurement method based on tower-type recovery according to claim 8, characterized in that: The loss function of the target recognition and segmentation network adopts the TaskAlignedAssigner strategy: t=s α +u β ; Among them, s is the prediction score corresponding to the labeled category, u is the intersection-over-union ratio of the predicted box and the true box, and the degree of alignment is measured by multiplying s and u; α and β are weight parameters, and t is the weighted judgment metric value; The loss function of the classification branch is the BCE loss function, and the loss function of the regression branch is the CIOU loss function and the DFL loss function; The deep learning network architecture uses an iterative training approach, training through diverse environmental samples, and continuously using the latest collected data to train and optimize the network as experiments and tests proceed.

Images