A 3D twin intelligent examination device and system for lance training

Abstract

The present application relates to a kind of 3D twin intelligence examination device and system to stab training, belong to intelligent training examination technical field, including inductive protective clothing, no wearing motion capture terminal, smart bracelet, multi-source data synchronization module, intelligent simulator, triplex screen host, the inductive protective clothing includes multiple combined armors and abdominal armor, multiple the combined armors are equipped with protective bending part and fixing belt part, several the fixing belt part is formed by the fastening structure of elastic band and velcro combination, multiple protective bending part includes fixed bending plate, elastic protective pad and conversion convex shell, the present application solves the problem of single data dimension in traditional assassination training examination, big artificial error in judgment, no tactical analysis, training injury prevention and control loss, AI guidance blank and poor review effect, realize the digitization, standardization, intelligentization of assassination training examination, while guaranteeing data local processing, satisfy information security requirement.

Description

Technical Field

[0001] This invention relates to the field of intelligent training and assessment technology, specifically to a 3D twin intelligent assessment device and system for needle-based training. Background Technology

[0002] Stabbing drills are an important part of daily training. Traditional stabbing drill assessments mainly rely on sensor-based protective gear and human judgment, which has many technical shortcomings. Data collection is only achieved through contact sensors, which can only obtain basic data such as whether a stab has occurred, the force, and the location. The data dimensions are limited and cannot capture key information such as the trainee's limb movements and tactical maneuvers. Assessment and judgment rely on human observation, which is prone to human error and difficult to resolve disputes. There is a lack of professional tactical analysis capabilities, and it is impossible to statistically analyze the trainee's movement trajectory and offensive and defensive posture, making it difficult to guide tactical optimization. There is no dedicated training injury prevention mechanism, and it is impossible to monitor high-risk movements during training, which can easily lead to sports injuries to the trainee's joints, waist, and other parts of the body. There is no AI-assisted guidance function, and training improvement relies solely on experience accumulation, which is inefficient. The review method is only a numerical list, which is not intuitive and cannot accurately locate training problems.

[0003] At the same time, some existing systems pose a risk of data leakage and are difficult to meet information security requirements. Therefore, there is an urgent need for an intelligent, standardized, and high-dimensional training and testing system to solve the pain points of the aforementioned traditional technologies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a 3D twin intelligent assessment device and system for bayonet training, which solves the problems of single data dimension, large human error in judgment, lack of tactical analysis, lack of training injury prevention and control, lack of AI guidance, and poor review effect in traditional bayonet training assessment. It realizes the digitalization, standardization, and intelligentization of bayonet training assessment, while ensuring local data processing and meeting information security requirements.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a 3D twin intelligent assessment device for stabbing training, comprising a sensor protective suit, a wearless motion capture terminal, a smart bracelet, a multi-source data synchronization module, an intelligent simulator, and a triple-screen host. The sensor protective suit comprises multiple combined armor plates and an abdominal armor. Each of the multiple combined armor plates is provided with a protective bend and a fixing strap. Each of the fixing straps adopts a fastening structure formed by a combination of elastic bands and hook and loop fasteners.

[0006] Each of the multiple protective bends includes a fixed bend plate, an elastic protective pad, and a conversion convex shell. The elastic protective pad, the fixed bend plate, and the conversion convex shell are arranged sequentially from the inside to the outside. A conversion component is provided inside several of the conversion convex shells.

[0007] The conversion assembly includes sealing push plates slidably mounted on opposite sides of the inner wall of the conversion convex shell. Several spring pieces are fixedly installed on the inner wall of the conversion convex shell near the sealing push plates. Several contact rods are fixedly installed on the elastic protective pad near the fixed bending plate. Several connecting ports are opened inside the fixed bending plate on one side of the contact rods. One end of each contact rod extends through into the interior of the corresponding connecting port. A first piston block is fixedly installed at one end of the contact rod in the connecting port. Several pulling components are provided on the inner wall of the conversion convex shell near the sealing push plates.

[0008] Furthermore, the sealing push plate is in contact with the inner wall of the conversion convex shell, and the two sealing push plates divide the internal space of the conversion convex shell into three regions. The elastic force of multiple spring plates can realize the automatic reset of the sealing push plate after displacement.

[0009] Furthermore, the side of several of the contact rods near the elastic protective pad is enlarged to improve wearing comfort. The surface of the first piston block is in close contact with the inner wall of the connecting port. Multiple connecting ports are disposed through one side of the conversion convex shell and extend into its interior.

[0010] The connection port at the fixed bending plate and the conversion convex shell are combined with each other, so that when the elastic protective pad moves, it can drive the first piston block to move. The gas change caused by the displacement is transmitted to the interior of the conversion convex shell. Multiple connection ports are located between the two sealing push plates, that is, in the middle area of ​​the conversion convex shell, so as to push the sealing push plates to move to both sides.

[0011] Furthermore, the pulling assembly includes a pulling bend tube fixedly installed on one side of the inner wall of the conversion convex shell. One end of the pulling bend tube penetrates the interior of the conversion convex shell and the surface of the fixed bend plate, and extends to one side of the surface of the elastic protective pad.

[0012] A second piston block is movably installed on the inner wall of the bent tube near the elastic protective pad. A first connecting ring block is fixedly installed on the surface of the second piston block near the elastic protective pad. A second connecting ring block is rotatably installed on one end of the first connecting ring block. One end of the second connecting ring block is fixedly connected to one side of the surface of the elastic protective pad. A gas filling valve pipe is fixedly installed at the bottom of the conversion convex shell to realize gas filling inside the conversion convex shell. Nitrogen can be filled into the conversion convex shell through the gas filling valve pipe to ensure the gas conversion effect when the structure is working.

[0013] Furthermore, the wearable motion capture terminal includes a supporting tripod and a data acquisition camera threadedly connected to the top of the supporting tripod. The data acquisition camera transmits the acquired information to the triple-screen host through a multi-source data synchronization module, thereby realizing the acquisition and aggregation of training data.

[0014] Furthermore, the triple-screen host includes a fixed main rod, on which three data processing screens are rotatably mounted at the middle position of the fixed main rod wall, and a fixing clamp is threadedly connected to the bottom of the fixed main rod.

[0015] A system for using a dual-twin intelligent assessment device for stabbing training, wherein the multi-source data synchronization module is integrated into the platform terminal control and AI central system, and is used to fuse and process the multi-source data collected by the physical perception layer. The physical perception layer is wirelessly connected to a wearable motion capture terminal, an intelligent simulator and a sensory protective suit. The wearable motion capture terminal is used to output the three-dimensional coordinates of 23 joints of the entire body of two people in real time, with a spatial positioning accuracy of ≤2cm, and supports data fusion with IMU data to continuously output data under occlusion. The intelligent wristband is worn on the trainee's wrist and is used to collect and wirelessly transmit physiological information including heart rate, blood oxygen saturation and blood pressure in real time.

[0016] The platform terminal control and AI central system is connected to the physical perception layer to receive multi-source data and includes a 3D twin construction module to construct a 1:1 twin digital examination room based on the received data. The model error is <0.5%, and a 3D digital human with a synchronization delay of ≤50ms with the physical action is generated in real time. The twin digital examination room displays the skeleton, view rotation and zoom, field ruler and distance ring auxiliary marks.

[0017] The multi-source data synchronization module is used to perform spatiotemporal fusion of multi-source data collected by the physical sensing layer, and is seamlessly taken over by the IMU when there is occlusion.

[0018] Basic functional modules are used to achieve real-time synchronization of 3D digital humans, construction of 1:1 twin digital examination rooms, wearable 3D motion capture, and support for multi-mode needle-assisted training.

[0019] The professional function modules include a standardized automatic assessment unit, a training injury prevention unit, a local AI large model guidance unit, and a data analysis, archiving, and review unit.

[0020] Furthermore, the standardized automatic evaluation unit includes a digital monitoring subunit, which is used to capture 6DoF data of the limbs and weapons of two people in real time, and dynamically present 3D skeleton model, foot trajectory, distance ring of 1.5m thrust danger zone and 2.5m tactical control zone, and visual elements of hit point. Among them, the skeleton changes color to warn when the joint angle exceeds the safe range, the distance ring changes color from green to yellow to red according to the confrontation distance, and the light effect and force value are displayed when hit.

[0021] The automated evaluation subunit is used to record and display seven core indicators in real time: number of hits / effective hit rate, number of hits conceded / effective defense rate, number of dangerous hits, dominant player in the confrontation, thrust speed, dynamic trajectory, and stamina percentage. The evaluation rule thresholds are adjusted in stages according to training and assessment needs.

[0022] The multimodal alert subunit is used to provide real-time alerts for hits, dangers, and violations through voice broadcasts and colored alarm bars.

[0023] Furthermore, the local AI large model guidance unit adopts a local deployment mode and has real-time tactical analysis capabilities, identifying tactical problems and providing prompts based on real-time adversarial data;

[0024] The personal technical diagnosis and recommendation function combines historical data to generate a quantitative technical weakness diagnosis report and recommends targeted training programs;

[0025] The dynamic injury risk assessment function integrates data from multiple fields, combines physiological information and movement frequency and intensity collected by the smart bracelet, and uses a machine learning model to predict the cumulative injury risk of high-risk areas and generate intervention suggestions.

[0026] The intelligent question answering and model evolution features support professional knowledge-based question answering in natural language and continuously optimize the model as training data accumulates.

[0027] Furthermore, the data analysis and archiving review unit supports 3D replay of the entire adversarial process, automatically marks key events such as hits, dangers, and violations, enables comparison of historical performance of individuals and units and generates training curves, supports personnel information input and assessment data export, and incorporates the current training data into the AI ​​model training library to achieve model iterative evolution.

[0028] Compared with the prior art, the present invention provides a 3D twin intelligent assessment device and system for needle training, which has the following beneficial effects:

[0029] 1. This device uses vision combined with IMU and sensor multi-source data fusion and wearless optical motion capture technology to accurately capture the 3D coordinates of 23 joints of the whole body, the 6DoF trajectory and movement path of the firearm. It breaks through the limitations of traditional contact acquisition and provides more comprehensive data dimensions. Relying on built-in standardized evaluation rules and adjustable judgment thresholds, it automatically completes real-time evaluation of 7 core indicators, achieving standardized evaluation, eliminating human error, and eliminating subjective judgment and evaluation disputes.

[0030] 2. This device has intelligent training injury prevention capabilities, monitors joint posture and force characteristics in real time, identifies high-risk movements and issues warnings, and assesses injury risks using models to ensure training safety from the source. It is equipped with a locally deployed AI large model to provide real-time tactical prompts, weakness diagnosis, customized training programs, and intelligent Q&A. All data is processed locally, improving training efficiency while ensuring information security.

[0031] 3. The device supports 3D holographic visualization for precise review, automatically marks key events, generates training curves and compares historical results, intuitively locates problems, and the review effect far exceeds that of traditional methods. The hardware adopts an integrated three-screen folding design, which is lightweight, easy to store, and can be quickly deployed and moved. It is suitable for training venues in multiple scenarios, and has a high degree of system integration and convenient deployment.

[0032] 4. The device utilizes the second piston block in the pulling component and the connecting ring structure to transmit air pressure changes in the opposite direction to the elastic protective pad, enabling it to dynamically adapt to changes in the joint bending angle and always maintain a close fit to the body parts. This adaptive fit design avoids the pressure or hollow feeling caused by traditional hard protective gear when the joint is bent, fundamentally improving the wearing comfort during long-term training.

[0033] 5. The device uses a reset system consisting of springs and pulling components to ensure that the elastic protective pad can quickly return to its initial state after each joint movement. Whether in high-frequency footwork or after intense offensive and defensive transitions, the protective pad can always fit closely to the body curves and will not shift or loosen due to movement inertia, thus achieving continuous comfortable fit during dynamic processes.

[0034] 6. The several connecting ports on the fixed curved plate and the airtight area inside the conversion convex shell of the device are all hollow structures. While ensuring structural strength, the use of materials is minimized, and the protective unit is made lightweight.

[0035] 7. Compared with the traditional pure mechanical linkage transmission structure, the gas transmission of this device does not require complex metal connectors, which greatly reduces the overall weight. At the same time, the air passage design inside the connecting port and the conversion convex shell replaces the bulky mechanical transmission channel, further optimizing the weight distribution. Attached Figure Description

[0036] Figure 1 This is a diagram of the software functional module architecture of the present invention;

[0037] Figure 2 This is a schematic diagram of the hardware process of the present invention;

[0038] Figure 3 This is a plan view of the sensor-activated protective suit of the present invention;

[0039] Figure 4 This is a perspective view of the smart bracelet of the present invention;

[0040] Figure 5 This is a perspective view of the triple-screen host of the present invention;

[0041] Figure 6 This is a three-dimensional view of the wearable motion capture terminal of the present invention;

[0042] Figure 7 This is a three-dimensional view of the combined armor of the present invention;

[0043] Figure 8 This is a perspective view of the protective bend of the present invention;

[0044] Figure 9 This is a perspective view of the fixed curved plate, elastic protective pad, and conversion convex shell of the present invention.

[0045] Figure 10 for Figure 9 A magnified structural diagram of structure A is shown below;

[0046] Figure 11 This is a cross-sectional perspective view of the fixed bending plate of the present invention;

[0047] Figure 12 for Figure 11 A schematic diagram of the enlarged structure of B shown;

[0048] Figure 13 This is a perspective view of the vertical section of the convex shell of the present invention;

[0049] Figure 14 for Figure 13 A schematic diagram of the enlarged structure of C shown;

[0050] Figure 15 This is a three-dimensional view of the cross-section of the pull-bend portion of the present invention.

[0051] In the diagram: 1. Sensor-activated protective suit; 2. Wearable motion capture terminal; 201. Support tripod; 202. Acquisition camera; 3. Smart bracelet; 4. Multi-source data synchronization module; 5. Smart simulator; 6. Triple-screen host; 601. Fixed main pole; 602. Data processing screen; 603. Fixing clamp;

[0052] 7. Combined armor; 701. Protective bend; 702. Fixing strap; 703. Fixing bend plate; 704. Elastic protective pad; 705. Conversion convex shell; 706. Gas valve pipe; 8. Abdominal armor;

[0053] 9. Conversion assembly; 901. Sealing push plate; 902. Spring; 903. Contact rod; 904. Connecting port; 905. First piston block; 706. Gas filling valve pipe;

[0054] 10. Pulling assembly; 1001. Pulling bend; 1002. Second piston block; 1003. First connecting ring block; 1004. Second connecting ring block. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1: A 3D twin intelligent assessment device for needle-stick training

[0057] Please see Figures 2 to 15 The present invention provides a 3D twin intelligent assessment device for stabbing training, including a sensor protective suit 1, a wearless motion capture terminal 2, a smart bracelet 3, a multi-source data synchronization module 4, an intelligent simulator 5, and a three-screen host 6.

[0058] The sensory protective suit 1 includes multiple combination armor plates 7 and an abdominal armor 8. The multiple combination armor plates 7 correspond to key parts of the human body such as the shoulders, elbows, chest, hips, and knees, respectively. The abdominal armor 8 fits snugly against the abdomen. Each combination armor plate 7 is equipped with a protective bend 701 and a fixing strap 702. The fixing straps 702 are all fastened by a combination of elastic bands and hook and loop fasteners. This design allows trainees of different body types to wear it comfortably and securely, and it is also easy to put on and take off.

[0059] Furthermore, each of the multiple protective bends 701 includes a fixed bend plate 703, an elastic protective pad 704, and a conversion convex shell 705. The elastic protective pad 704, the fixed bend plate 703, and the conversion convex shell 705 are arranged sequentially from the inside to the outside. The elastic protective pad 704 is made of high-elasticity memory foam or similar soft material, which directly contacts the human body, provides cushioning, improves wearing comfort, and disperses the stabbing pressure. The fixed bend plate 703 is an arc-shaped plate with a certain rigidity, which plays the role of support and force transmission. The conversion convex shell 705 is a rigid hollow shell, and a conversion component 9 is provided inside it to convert the mechanical energy of the stabbing into a detectable air pressure signal.

[0060] Specifically, the conversion assembly 9 includes sealing push plates 901 that are slidably installed on opposite sides of the inner wall of the conversion convex shell 705. The sealing push plates 901 are tightly fitted to the inner wall of the conversion convex shell 705 to ensure airtightness. The two sealing push plates 901 divide the internal space of the conversion convex shell 705 into three independent airtight areas: the left side area, the middle area, and the right side area. Several spring pieces 902 are fixedly installed inside the inner wall of the conversion convex shell 705 on the side near the sealing push plates 901, i.e. the left and right side areas. The elastic force of these spring pieces 902 can push the sealing push plates 901 to automatically reset after the external force disappears.

[0061] Several contact rods 903 are fixedly installed on the side of the elastic protective pad 704 near the fixed bending plate 703. Several connecting ports 904 are opened inside the fixed bending plate 703. The positions of these connecting ports 904 correspond one-to-one with the contact rods 903. One end of the contact rod 903 extends through into the interior of the corresponding connecting port 904. A first piston block 905 is fixedly installed at one end of the contact rod 903 located in the connecting port 904. The surface of the first piston block 905 is tightly fitted with the inner wall of the connecting port 904 to form a piston structure. Multiple connecting ports 904 are arranged through one side of the conversion convex shell 705 and extend into its interior. Specifically, the ends of these connecting ports 904 all lead to the middle area of ​​the conversion convex shell 705, that is, the area between the two sealing push plates 901.

[0062] Its working principle is as follows:

[0063] When the trainee uses it, the knee bends or an external force is applied to the elastic protective pad 704, causing it to indent inward. The displacement of the elastic protective pad 704 drives the contact rod 903 and the first piston block 905 to move inward within the connecting port 904. Since the first piston block 905 is tightly attached to the inner wall of the connecting port 904, its movement will compress the gas in the middle area of ​​the connecting port 904 and the conversion convex shell 705. The gas inside is pre-charged nitrogen gas added through the gas filling valve pipe 706, which causes the gas pressure in this area to rise. The increased gas pressure then pushes the two sealing push plates 901 to move to the left and right sides respectively, compressing the spring 902. The displacement of the sealing push plates 901 can push the second piston block 1002 to move.

[0064] In addition, a pulling assembly 10 is provided inside the conversion convex shell 705. The pulling assembly 10 includes a pulling bend 1001 fixedly installed on one side of the inner wall of the conversion convex shell 705. One end of the pulling bend 1001 passes through the interior of the conversion convex shell 705 and the surface of the fixed bend plate 703, and extends to one side of the surface of the elastic protective pad 704. A second piston block 1002 is movably installed on the inner wall of the pulling bend 1001 near the elastic protective pad 704. A first connecting ring block 1003 is fixedly installed on the surface of the second piston block 1002 near the elastic protective pad 704. A second connecting ring block 1004 is rotatably installed on one end of the first connecting ring block 1003. One end of the second connecting ring block 1004 is fixedly connected to one side of the surface of the elastic protective pad 704. A gas filling valve pipe 706 is fixedly installed at the bottom of the conversion convex shell 705.

[0065] The main function of the pulling component 10 is to assist the pulling of the elastic protective pad 704. The elastic protective pad 704 can be pulled by the connection of the first connecting ring block 1003 and the second connecting ring block 1004. This ensures that the elastic protective pad 704 better wraps the shoulders, elbows, chest, hips and knees, increasing the comfort of the structure. Inert gases such as nitrogen can be filled into the conversion convex shell 705 through the gas filling valve pipe 706 to improve the sensitivity and stability of the air pressure signal transmission.

[0066] The wearable motion capture terminal 2 includes a support tripod 201 and a capture camera 202 threadedly connected to the top of the support tripod 201. There can be multiple capture cameras 202, which are arranged around the training area to capture the trainee's movements from different angles. The capture cameras 202 transmit the captured visual information to the triple-screen host 6 through the multi-source data synchronization module 4 to realize the collection and summarization of training data.

[0067] The triple-screen host 6 includes a fixed main pole 601. Three data processing screens 602 are rotatably installed in the middle of the wall of the fixed main pole 601. The three screens can be used to display different information, such as 3D twin scenes, real-time assessment data, AI guidance suggestions, etc., which facilitates coaches and trainees to observe from multiple dimensions. The bottom of the fixed main pole 601 is threaded with a fixing clip 603, which makes it easy to fix the host to the side of the table or a mobile stand.

[0068] Example 2: A System for Using a 3D Twin Intelligent Assessment Device for Needle Training

[0069] This embodiment is based on the 3D twin intelligent assessment device for stabbing training in Embodiment 1, and further defines its software system and overall workflow.

[0070] Please see Figures 1 to 15 The system includes a physical perception layer and a platform terminal control and AI central system;

[0071] The physical sensing layer communicates wirelessly with the wearable motion capture terminal 2, the intelligent simulator 5, the sensing protective suit 1, and the smart bracelet 3.

[0072] The wearable motion capture terminal 2, based on AI visual recognition technology, is used to output the three-dimensional coordinates of 23 joints of two people's whole body in real time, with a spatial positioning accuracy of ≤2cm. It integrates an IMU (Inertial Measurement Unit) and supports data fusion with IMU. When the visual signal is blocked, the IMU data can take over seamlessly to ensure the continuity of motion capture.

[0073] The Smart Bracelet 3 is worn on the wrist of the trainee and has built-in physiological sensors for real-time collection and wireless transmission of physiological information including heart rate, blood oxygen saturation and blood pressure.

[0074] The Smart Simulator 5 is designed for simulating training firearms. It has a built-in 6DoF (six degrees of freedom) sensor and can output parameters such as the firearm's spatial position, attitude, and thrusting speed in real time.

[0075] The sensor-activated protective suit 1, as in Embodiment 1, has a pressure sensor connected to its internal conversion component 9 for collecting data on the force and location (hit point) of the stabbing impact.

[0076] The platform terminal control and AI central system communicates with the physical perception layer to receive and process multi-source data, and integrates multiple functional modules:

[0077] Multi-source data synchronization module 4

[0078] This module is responsible for spatiotemporal fusion of visual, sensory, and IMU data collected by the physical perception layer. Spatial fusion converts 2D image coordinates into unified 3D world coordinates through pre-calibrated camera intrinsic and extrinsic parameters. When visual data is lost due to occlusion, this module can be seamlessly taken over by IMU data within ≤0.3 seconds, ensuring that the overall system latency is ≤50ms.

[0079] 3D Twin Building Module

[0080] Based on 3D reconstruction technology, this module pre-generates a 1:1 digital twin of the actual training site through laser scanning or photogrammetry with a model error of <0.5%. During training, it generates a 3D digital human in real time based on motion capture data with a synchronization delay of ≤50ms with the actual physical movements. In the digital twin, it can overlay and display skeleton visibility, view rotation and zoom, site markers, and auxiliary markers such as the 1.5m thrust danger zone and the 2.5m tactical control zone.

[0081] Basic functional modules

[0082] Used to achieve real-time synchronization of 3D digital humans, construction of 1:1 twin digital examination rooms, wearable 3D motion capture, and support for multi-mode (such as single-person training, two-person stabbing, and group confrontation) stabbing training;

[0083] Professional functional modules

[0084] The professional functional modules include a standardized automatic assessment unit, a training injury prevention unit, a local AI large model guidance unit, and a data analysis, archiving, and review unit;

[0085] Standardized automated assessment unit

[0086] This unit further includes a digital monitoring subunit, an automated evaluation subunit, and a multimodal prompting subunit;

[0087] The digital monitoring subunit captures 6DoF data of the two combatants' limbs and weapons in real time, dynamically presenting visual elements such as 3D skeletal models, foot trajectories, distance rings, and hit points. When the joint angle exceeds the preset safe range, the skeletal model will change color as a warning; as the distance between the two combatants changes, the distance ring color will change from green (safe) to yellow (warning) to red (danger); upon a hit, the screen displays light effects and force values.

[0088] The automated evaluation subunit, based on real-time data, automatically calculates and displays seven core metrics: hits / effective hit rate, hits conceded / effective defense rate, dangerous hits, dominant player in the confrontation, thrust speed, dynamic trajectory, and stamina percentage. The thresholds for all metrics can be adjusted according to the training syllabus, eliminating human error.

[0089] The multimodal alert subunit broadcasts voice messages through the speakers of the triple-screen host 6, while simultaneously displaying real-time alerts for hits, dangers, and violations at the bottom of the three screens in the form of scrolling colored alarm bars (green / blue / yellow / red).

[0090] The training injury prevention unit monitors the three-dimensional angles and force characteristics of key joints such as the hip, knee, shoulder, and waist in real time. It uses a biomechanical model to automatically identify high-risk movements such as knee valgus, lumbar compensatory force, and elbow hyperextension, and displays text warnings on the screen. At the same time, this unit combines physiological data (such as sudden increase in heart rate) and movement frequency and intensity collected by the smart bracelet 3, and uses a built-in machine learning model to assess the risk of cumulative musculoskeletal injury and generate targeted warm-up, cool-down, or strength-enhancing intervention suggestions.

[0091] The local AI large model guidance unit adopts a localized deployment mode to ensure that training data is not transmitted outside and to protect data security. Its core functions include real-time tactical analysis and prompts: based on real-time adversarial attack and defense data, the AI ​​model can identify tactical problems (such as excessive defensive gaps or poor timing of thrusts) and provide real-time prompts through voice or text.

[0092] Individual technical weakness diagnosis: After training, the model combines historical data to generate a quantitative diagnostic report, which clearly points out individual technical weaknesses (such as "too high center of gravity when thrusting with the left lunge" and "slow speed of transitioning from defense to offense") and recommends targeted training programs. Dynamic injury risk assessment: By integrating training data from multiple sessions, the model predicts the cumulative injury risk trend of specific parts (such as the knee joint and lower back).

[0093] Intelligent question answering and model evolution support natural language interaction. Trainers or coaches can ask questions directly (such as "How to improve thrust speed?"), and the model will provide answers based on its professional knowledge base. As training data accumulates, the model will continue to optimize and achieve self-evolution.

[0094] Data analysis and archiving review unit: Supports 3D holographic replay of the entire adversarial process, automatically marking key events such as hits, dangers, and violations, facilitating retrospective analysis for coaches and trainees. The system can automatically generate historical performance comparison curves for individuals and units, intuitively displaying training results. Simultaneously, it supports personnel information entry and export / archiving of assessment data, and incorporates the training data into the AI ​​model training library for iterative model evolution.

[0095] S1: System setup and initialization, unfold and start the triple-screen host 6, deploy the wearless motion capture terminal 2 and complete the calibration, trainees wear the sensing protective clothing 1, smart bracelet 3, and hold the smart simulator 5, the system automatically loads the three-dimensional model of the site, constructs a 1:1 twin digital examination room, and completes high-precision time synchronization between all devices.

[0096] S2: Real-time acquisition of multi-source data. After training begins, the wearable motion capture terminal 2 captures the 3D coordinates of the joints of the two people 23, the intelligent simulator 5 outputs the DoF data of the gun 6, the sensing protective suit 1 collects the stabbing force data, and the multi-source data synchronization module 4 performs spatiotemporal fusion of visual, sensing and IMU data. When visual occlusion occurs, the IMU takes over seamlessly.

[0097] S3: Real-time construction of 3D twin scenes. Based on the fused motion data, the system generates a 3D digital human synchronized with the trainee in real time in the twin digital examination room, and dynamically displays visual elements such as skeleton, trajectory, distance loop, and hit light effect, with a synchronization delay of ≤50ms.

[0098] S4: Real-time monitoring and automatic evaluation. The standardized automatic evaluation unit captures 6DoF data in real time and automatically evaluates 7 core indicators. When a life-threatening, dangerous, or violation event occurs, the system immediately provides multimodal prompts through voice broadcast and screen color alarm bars, and generates evaluation results in real time.

[0099] S5: Real-time training injury prevention. The training injury prevention unit continuously monitors the posture of key joints. Once a high-risk movement (such as knee valgus) is detected, a text warning will immediately pop up on the screen. At the same time, combined with the physiological data of the smart bracelet 3, the cumulative risk of injury is dynamically assessed.

[0100] S6: Real-time guidance and post-match analysis from a large AI model. During training, the local large AI model analyzes offensive and defensive data in real time and provides tactical hints. After training, the model automatically generates a technical weakness diagnosis report and recommends subsequent training plans.

[0101] S7: Full-process review and data archiving. The system automatically generates 3D replays of the adversarial process, marking all key events. Coaches and trainers can review the replays through a three-screen setup. All assessment data, diagnostic reports, and training videos are archived and used for the continuous evolution of the AI ​​model.

[0102] The installation, connection, or setting methods disclosed in this embodiment are all common mechanical connection methods. Any method that can achieve its beneficial effect can be implemented. In addition, the electrical components in this embodiment are all electrically connected to the main controller and the power supply. The main controller can be a conventional known device such as a computer that plays a control role. Those skilled in the art can control the electrical components through simple programming. Moreover, the existing disclosed power connection technology is also common knowledge in the field. Therefore, the specific structural composition and working principle will not be described in detail in this embodiment.

Claims

1. A 3D twin intelligent assessment device for stabbing training, comprising a sensor protective suit (1), a wearable motion capture terminal (2), a smart bracelet (3), a multi-source data synchronization module (4), an intelligent simulator (5), and a triple-screen host (6), characterized in that: The inductive protective suit (1) includes multiple combined armor plates (7) and abdominal armor (8). Each of the multiple combined armor plates (7) is provided with a protective bend (701) and a fixing strap (702). Each of the fixing straps (702) adopts a fastening structure formed by a combination of elastic band and hook and loop fastener. Each of the multiple protective bends (701) includes a fixed bend plate (703), an elastic protective pad (704), and a conversion convex shell (705). The elastic protective pad (704), the fixed bend plate (703), and the conversion convex shell (705) are arranged sequentially from the inside to the outside. A conversion component (9) is provided inside several of the conversion convex shells (705). The conversion assembly (9) includes a sealing push plate (901) slidably mounted on opposite sides of the inner wall of the conversion convex shell (705). A plurality of spring pieces (902) are fixedly mounted on the inner wall of the conversion convex shell (705) near the sealing push plate (901). A plurality of contact rods (903) are fixedly mounted on the side of the elastic protective pad (704) near the fixed bending plate (703). A plurality of connecting ports (904) are opened inside the fixed bending plate (703) on one side of the plurality of contact rods (903). One end of the plurality of contact rods (903) extends through to the interior of the corresponding connecting port (904). A first piston block (905) is fixedly mounted on one end of the contact rod (903) at the connecting port (904). A plurality of pulling assemblies (10) are provided on the inner wall of the conversion convex shell (705) near the sealing push plate (901). The pulling assembly (10) includes a pulling bend (1001) fixedly installed on one side of the inner wall of the conversion convex shell (705). One end of the pulling bend (1001) penetrates the interior of the conversion convex shell (705) and the surface of the fixed bend plate (703), and extends to one side of the surface of the elastic protective pad (704). A second piston block (1002) is movably installed on the inner wall of the pull bend (1001) near the elastic protective pad (704). A first connecting ring block (1003) is fixedly installed on the surface of the second piston block (1002) near the elastic protective pad (704). A second connecting ring block (1004) is rotatably installed on one end of the first connecting ring block (1003). One end of the second connecting ring block (1004) is fixedly connected to one side of the surface of the elastic protective pad (704). A gas filling valve pipe (706) is fixedly installed at the bottom of the conversion convex shell (705).

2. The 3D twin intelligent assessment device for stabbing training according to claim 1, characterized in that: The sealing push plate (901) fits against the inner wall of the conversion convex shell (705). The two sealing push plates (901) divide the internal space of the conversion convex shell (705) into three regions. The elastic force of the multiple spring pieces (902) can realize the automatic reset of the sealing push plate (901) after displacement.

3. The 3D twin intelligent assessment device for stabbing training according to claim 2, characterized in that: The side of several of the contact rods (903) near the elastic protective pad (704) is an enlarged structure. The surface of the first piston block (905) is in close contact with the inner wall of the connecting port (904). Multiple connecting ports (904) are provided through one side of the conversion convex shell (705) and extend into its interior. The fixed bending plate (703) and the connecting port (904) at the conversion convex shell (705) are combined with each other so that when the elastic protective pad (704) moves, it can drive the first piston block (905) to generate displacement. The gas change caused by the displacement is transmitted to the interior of the conversion convex shell (705). The multiple connecting ports (904) are located between the two sealing push plates (901), that is, in the middle area of ​​the conversion convex shell (705), so as to push the sealing push plates (901) to move to both sides.

4. The 3D twin intelligent assessment device for stabbing training according to claim 3, characterized in that: The wearable motion capture terminal (2) includes a support tripod (201) and a capture camera (202) threadedly connected to the top of the support tripod (201).

5. The 3D twin intelligent assessment device for stabbing training according to claim 4, characterized in that: The triple-screen host (6) includes a fixed main rod (601), three data processing screens (602) are rotatably installed at the middle position of the wall of the fixed main rod (601), and a fixing clip (603) is threaded to the bottom of the fixed main rod (601).

6. A system for using a 3D twin intelligent assessment device for needle training, characterized in that: The 3D twin intelligent assessment device for stabbing training according to any one of claims 1-5, wherein the multi-source data synchronization module (4) includes a platform terminal control and AI central system and a physical perception layer, wherein the physical perception layer is wirelessly connected to the wearless motion capture terminal (2), the intelligent simulator (5) and the sensing protective clothing (1), wherein the wearless motion capture terminal (2) is used to output the three-dimensional coordinates of 23 joints of the whole body of two people in real time, with a spatial positioning accuracy of ≤2cm, and supports data fusion with IMU data to continuously output data under occlusion conditions, wherein the intelligent wristband (3) is worn on the wrist of the trainee and is used to collect and wirelessly transmit physiological information including heart rate, blood oxygen saturation and blood pressure in real time; The platform terminal control and AI central system is connected to the physical perception layer to receive multi-source data and includes a 3D twin construction module to construct a 1:1 twin digital examination room based on the received data. The model error is <0.5%, and a 3D digital human with a synchronization delay of ≤50ms with the physical action is generated in real time. The twin digital examination room displays the skeleton, view rotation and zoom, field ruler and distance ring auxiliary marks. The multi-source data synchronization module (4) is used to perform spatiotemporal fusion of multi-source data collected by the physical sensing layer and is seamlessly taken over by the IMU when occlusion occurs. Basic functional modules are used to achieve real-time synchronization of 3D digital humans, construction of 1:1 twin digital examination rooms, wearable 3D motion capture, and support for multi-mode needle-assisted training. The professional function modules include a standardized automatic assessment unit, a training injury prevention unit, a local AI large model guidance unit, and a data analysis, archiving, and review unit.

7. The system for using a 3D twin intelligent assessment device for stabbing training according to claim 6, characterized in that: The standardized automatic evaluation unit includes a digital monitoring subunit, which is used to capture 6DoF data of the limbs and weapons of two people in real time, dynamically present 3D skeleton model, foot trajectory, distance ring of 1.5m thrust danger zone and 2.5m tactical control zone, and visual elements of hit point. Among them, the skeleton changes color to warn when the joint angle exceeds the safe range, the distance ring changes color from green to yellow to red according to the confrontation distance, and the light effect and force value are displayed when hit. The automated evaluation subunit is used to record and display seven core indicators in real time: number of hits / effective hit rate, number of hits conceded / effective defense rate, number of dangerous hits, dominant player in the confrontation, thrust speed, dynamic trajectory, and stamina percentage. The evaluation rule thresholds are adjusted in stages according to training and assessment needs. The multimodal alert subunit is used to provide real-time alerts for hits, dangers, and violations through voice broadcasts and colored alarm bars.

8. The system for using a 3D twin intelligent assessment device for stabbing training according to claim 6, characterized in that: The local AI large model guidance unit adopts a local deployment mode and has real-time tactical analysis capabilities, which identify tactical problems and provide prompts based on real-time adversarial data. The personal technical diagnosis and recommendation function combines historical data to generate a quantitative technical weakness diagnosis report and recommends targeted training programs; The damage risk dynamic assessment function integrates data from multiple fields, combines the physiological information and movement frequency and intensity collected by the smart bracelet (3), and predicts the cumulative damage risk of high-risk parts and generates intervention suggestions through machine learning models. The intelligent question answering and model evolution features support professional knowledge-based question answering in natural language and continuously optimize the model as training data accumulates.

9. The system for using a 3D twin intelligent assessment device for stabbing training according to claim 6, characterized in that: The data analysis and archiving review unit supports 3D replay of the entire adversarial process, automatically marks key events such as hits, dangers, and violations, enables comparison of historical performance of individuals and units and generates training curves, supports personnel information input and assessment data export, and incorporates the current training data into the AI ​​model training library to achieve model iterative evolution.

Images