Method and system for defending an area
A virtual model-based defense system using affordable drones and guns, trained through simulations, addresses the high cost of existing systems by providing accurate and adaptable aerial threat interception.
Patent Information
- Application Number
- PCT/EP2025/066839
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-05
- Filing Date
- 2025-06-17
- Publication Date
- 2026-11-12
AI Technical Summary
Existing defense systems that use self-guiding missiles or drones are expensive, necessitating a more cost-effective alternative for defending an area.
A method and system utilizing a virtual model to emulate a physical area and its airspace, employing affordable drones and guns, trained through simulations to intercept threats, with a stereo camera system for accurate tracking and engagement.
Enables effective defense against aerial threats using cost-effective drones and guns, with high accuracy and adaptability to varying environmental conditions, through immersive virtual training and real-time threat engagement.
Smart Images

Figure EP2025066839_12112026_PF_FP_ABST
Abstract
Description
[0001] METHOD AND SYSTEM FOR DEFENDING AN AREA
[0002] TECHNICAL FIELD
[0003] The present invention relates to a method and a system for defending an area, in particular by intercepting hostile obj ects entering the airspace above the area .
[0004] PRIOR ART
[0005] From the prior art defense systems are known that use intelligent proj ectiles, such as self-guiding missiles, or that use self-guiding drones, both of which are expensive .
[0006] DESCRIPTION OF THE INVENTION
[0007] One obj ect of the present invention is to provide a method for defending an area and a defense system that do not require expensive proj ectiles or expensive drones .
[0008] This obj ect is achieved by a method for defending at least one area with the features of claim 1. Further embodiments of the method, as well as a system for defending an area, are defined by the features of further claims .
[0009] A method according to the invention for defending at least one area comprises the steps of :
[0010] Setting up a defense system by:
[0011] Defining the at least one area and defining a corresponding airspace above the area;Defining system components dedicated to the at least one area, selected from the group comprising digital stereo camera, computer, gun and / or drone that are controlled by the computer;
[0012] Scanning a surface of the area or scanning the surface of the area and scanning the corresponding airspace ;
[0013] Creating a virtual model of the surface of the area and the corresponding airspace by:
[0014] Transferring data from the scanning to the computer;
[0015] Emulating the digital stereo camera, the gun and / or the drone in the virtual model on the computer; Emulating the surface of the area or the surface of the area and the corresponding airspace in the virtual model on the computer;
[0016] Segmenting the corresponding airspace into finite volumes ;
[0017] Training the virtual model by:
[0018] Introducing virtual targets in the virtual model; Intercepting the virtual targets with the gun and / or with the drone;
[0019] Deploying the defense system by:
[0020] Positioning the digital stereo camera, the gun and / or the drone in the area;Calibrating the trained virtual model with current images from the digital stereo camera;
[0021] Activating the defense system by using the trained virtual model to intercept actual targets .
[0022] Such a method allows the use of guns and / or drones that are not expensive . For example, drones without optical equipment can be used. Such drones use "Biological-style navigation" — where the drone doesn' t know its exact position but navigates by matching what it feels to what it remembers . This is bio-inspired, Al-guided flight by sensation and memory — no need to "see" or "communicate, " just to remember and adapt . Such drones only need an Inertial Measurement Unit ( IMU) , barometric sensor, compass, microcontroller unit (MCU) , memory matching, memory (Flash / SSD) for storing data and the virtual model, deep learning model for matching live sensor stream to stored paths and a control logic that translates the position estimate to motor output PID-loop .
[0023] For emulating the surface of the area, a modeling and rendering platform, such as a 3D-modeling software or game engine, can be used, configured to ingest and process scanned environmental data .
[0024] The virtual model can generate virtual scenes based on the area and the corresponding airspace . The scanned data are converted into digital elevation models (DEMs) or terrain meshes mapped to a virtual coordinate space . Textural data extracted from the images is applied as terrain texture maps to provide realistic surface appearance .Vegetation modeling is performed using data derived from spectral analysis or classified satellite imagery. This data is used to procedurally generate or manually place three-dimensional flora assets . Scanned airspace data, is proj ected onto a virtual airspace, aligned to match sun positions, cloud formations, and lighting conditions relative to the terrain.
[0025] The alignment of the virtual airspace with the virtual area ensures consistency in illumination, shadow behavior, and environmental reflections . Once imported and aligned, no further manual calibration is necessary, thereby enabling a seamless integration of terrain and atmosphere into a unified, immersive virtual environment .
[0026] The emulated digital stereo camera is configured to emulate the optical and behavioral characteristics of the physical digital stereo camera .
[0027] The emulated digital stereo camera includes configurable parameters corresponding to real-world camera specifications, including focal length, aperture, shutter speed, and ISO sensitivity. These parameters govern field of view, depth of field, exposure, and motion blur within the virtual environment . Post-processing filters replicate lens artifacts such as chromatic aberration, vignetting, and barrel or pincushion distortion.
[0028] Autofocus behavior may be simulated using depth-based algorithms or predefined focal distances and blur radii . Camera positioning and motion are defined using keyframe animation, spline-based motion paths, or kinematic rigs to simulate handheld, dolly, or aerial movements .To match real-world pixel density, the system allows calibration of virtual resolution (e . g. , 1920x1080 pixels) and ensures accurate scaling within the environment (e . g. , 1 unit = 1 meter) . Obj ects are tested for pixel coverage at known distances to validate optical parity with real cameras . The resulting virtual camera output provides a synthetic video stream that mirrors real-world imaging behavior .
[0029] The system incorporates a calibrated stereo vision setup and a virtual camera environment designed to replicate real-world optical behavior . Within the 3D engine, the virtual stereo camera is configured to match the physical properties of the stereo camera system, including resolution, focal length, and baseline . The field of view is computed from the focal length and sensor size, ensuring geometric consistency with the real camera . Aperture, shutter speed, and ISO are simulated to match exposure characteristics, while depth of field and lens effects— such as distortion and chromatic aberration— are optionally emulated through post-processing layers . This virtual stereo camera is embedded into a scene where 1 unit corresponds to 1 real-world meter, preserving spatial scale and enabling accurate pixel-based measurements .
[0030] To reconstruct segmented multi-layer airspaces with a tip height of e . g. 20 meters using stereo vision and pixel data, the stereo camera system captures left and right images of the scene . After calibration, image rectification ensures that corresponding points lie along horizontal epipolar lines, simplifying disparity computation.Disparity d is defined as the horizontal difference in pixel coordinates between corresponding points in the left and right images :
[0031] d = x_left - x_right
[0032] With known focal length f (in pixels) and baseline distance B (in meters) , the real-world depth Z of a point is calculated as :
[0033] Z = ( f * B) / d
[0034] Using the pixel coordinates of a point (x, y) and the camera' s principal point (c_x, c_y) , the 3D position in camera space is computed:
[0035] X = ( (x - c_x) * Z) / f
[0036] Y = ( (y - c_y) * Z) / f
[0037] To determine the real-world height of the airspace, the system identifies the base and tip of the airspace within the image and computes their respective Y-coordinates . The height is then given by:
[0038] Height = | Y_tip - Y_base |
[0039] The virtual environment is used during a training phase to simulate how the segmented multi-layer airspaces appears under ideal conditions, with accurate depth, scale, and pixel density. By placing the virtual airspace in a scene and capturing its rendered appearance at varying distances, a lookup table is generated that maps pixel height to real-world distance . During deployment, this data is used to validate or refine stereo depth measurements, enablingaccurate placement of the 20-meter virtual segmented multilayer airspace in the augmented reality (AR) scene .
[0040] Once calibrated, the stereo system and computer work in tandem to ensure that the virtual airspace is anchored and scaled accurately within the physical environment . The rendered height and position of the airspace are dynamically adjusted based on the calculated 3D coordinates, producing a spatially consistent augmented reality experience .
[0041] In one embodiment, the system comprises a virtual representation of defensive platforms, specifically antiaircraft guns and autonomous defense drones . These assets are trained within the same virtual model to engage hostile targets based on real-time inputs and threats positions within the virtual airspace . Their behavior is governed by targeting algorithms that process sensor data— such as target velocity, traj ectory prediction, and proximity— and respond with appropriate countermeasures . Anti-aircraft guns are trained to engage fast-moving, high-altitude threats with precision, while drones are optimized for agile pursuit, interception, or electronic disruption.
[0042] To enhance the realism and robustness of the training process, the system is designed to execute millions of simulations, automatically varying parameters such as obj ect position withing multiple layers gridded domes, atmospheric conditions, lighting, terrain, and background complexity. These simulations can be orchestrated via scripting or Al-controlled agents, enabling large-scale generation of synthetic yet highly realistic datasets .By embedding both hostile targets and defensive systems within the virtual environment, and linking them to the stereo camera vision, the training system produces an integrated dataset that supports :
[0043] Target classification and traj ectory prediction Engagement decision-making logic
[0044] Real-time tracking and system activation
[0045] Post-engagement analysis .
[0046] To bring the trained model into the real world using the same scanned area, start by aligning the virtual environment' s coordinate system with real-world GPS coordinates, ensuring that the virtual map and actual terrain match in scale and orientation. Place the real stereo camera at the same fixed position it had in the virtual world, using precise latitude, longitude, and elevation data . Calibrate the camera' s internal settings ( focal length, sensor size, angle, etc . ) to match those used in the simulation. When the stereo camera operates in the real world, its live feed is processed by the trained model, which recognizes, and tracks obj ects based on the countless virtual scenarios it was exposed to . Since the model already understands how these obj ects appear and move relative to the environment and the virtual multiple layers domes, it can accurately interpret real-world data, adjusting for lighting and atmospheric differences, and apply its detection and tracking abilities effectively in the actual location.The anti-aircraft cannons and defense drones are fully integrated into the central command system and are associated with precise geospatial data . Each asset is registered with its exact latitude and longitude coordinates, allowing the system to maintain an accurate and real-time understanding of their spatial distribution across the operational zone .
[0047] Threats are identified by their positions in the airspace, enabling the system to determine their location with high accuracy .
[0048] Given this dual-positional awareness— of both defensive units and incoming threats, the system is capable of performing real-time threat engagement analysis . Based on proximity, orientation, and position on grid, the system evaluates which defensive asset (anti-air gun / cannon or defense drone) is optimally positioned to neutralize a given threat . It then automatically selects and activates the appropriate response mechanism to intercept or disable the incoming threat at the most effective interception point of the segmented airspace .
[0049] In one embodiment, two or more areas with their corresponding airspaces are defined and the virtual model of each surface area and each corresponding airspace is created and trained by one single computer or by two or more computers that are operatively connected to one another .
[0050] In one embodiment, the corresponding airspace is domeshaped, hemispherical, cuboid, cube-shaped, or prismshaped .In one embodiment, for each of the at least one area, the system comprises two or more dedicated digital stereo cameras, and / or two or more dedicated computers that are operatively connected to one another and / or two or more dedicated guns and / or two or more dedicated drones .
[0051] In one embodiment, the scanning is done by LiDAR systems on fixed or mobile platforms and / or wherein the scanning is done by unmanned aerial vehicles equipped with high-resolution RGB and / or multispectral imaging systems and / or wherein the scanning is done by multispectral or hyperspectral imaging sensors and / or high-dynamic-range imaging systems and / or time-lapse cameras .
[0052] In one embodiment, emulating the surface of the area comprises integrating publicly available topographic data . In one embodiment, segmenting the airspace comprises dividing the airspace into superimposed and adjoining shells, wherein a first shell defines an outside of the airspace and wherein a last shell defines an inside of the airspace, wherein each shell is defined by a difference in radius, and wherein each shell is divided azimuthal and polar into finite volumes, wherein each finite volume is defined by a difference in azimuth angle and by a difference in polar angle or wherein each finite volume is defined by a difference in a first direction, by a difference in a second direction, and by a difference in a third direction.
[0053] In one embodiment, the finite volumes of different shells have the same size or have different sizes, or wherein thefinite volumes of different shells have the same difference in azimuth angle and the same difference in polar angle . In one embodiment, introducing virtual targets is creating virtual images of the emulated digital stereo camera and wherein these virtual images are used to predict the velocity and traj ectory of the virtual targets .
[0054] This enables the stereo camera-based systems to anticipate the future behavior and traj ectory of dynamic aerial or ballistic obj ects, such as drones or artillery shells .
[0055] In one embodiment, the intercepting is done by using the predicted velocities and traj ectories to control the emulated gun and / or emulated drone .
[0056] The system initiates with a series of simulations where virtual camera sensors observe the motion of virtual obj ects under varying initial conditions . These conditions include, but are not limited to, launch angles, velocities, environmental variables (e . g. , wind, lighting, cloud cover) , and obj ect types .
[0057] Each simulation run produces annotated video sequences and metadata such as frame-by-frame obj ect positions in the grid, or finite volume, velocities, and classifications . These outputs are compiled into a dataset and input to a machine learning engine configured to analyze spatiotemporal patterns and develop predictive models of obj ect motion.
[0058] Over successive iterations and high-volume simulation runs; the predictive model is refined through supervised or reinforcement learning techniques . This enables the virtualcamera system to infer, based on early motion cues, position on multiple layer-gridded or layer-segmented airspaces and environmental context, the likely traj ectory and final position of an observed obj ect, or target .
[0059] In operational use, the virtual camera' s perception pipeline, trained on the synthetic data or virtual model, can detect a moving obj ect, analyze its motion history, and forecast its traj ectory in real time .
[0060] In one embodiment, the system comprises a dynamic weather simulation module configured to emulate atmospheric phenomena, including wind, temperature, humidity, cloud cover, rain, snow, and lighting variations corresponding to different times of day and seasons .
[0061] The weather simulation module comprises :
[0062] A wind field generator, producing spatially and temporally varying vector fields representing wind direction, speed, and turbulence at multiple altitudes and terrain-relative positions .
[0063] Precipitation modules, capable of simulating particle effects (rain, snow, hail) and their influence on visibility, obj ect drag, and ground friction.
[0064] A solar and cloud model, calculating sun position, shadow direction, and diffuse lighting effects based on geographic coordinates and time of day.
[0065] Thermal gradient emulation, introducing convection effects or turbulence zones that influence aerial obj ect behavior .These conditions are applied in real-time to affect obj ect dynamics . For instance, a drone ' s path may be altered by wind gusts, and an artillery shell ' s arc may deviate due to simulated crosswinds or atmospheric drag changes .
[0066] The system can perform batch simulation runs, each with randomized or systematically varied weather configurations, enabling the generation of diverse training datasets . Each simulation instance comprises :
[0067] Obj ect kinematics under specified weather conditions Visual scene variations (e . g. , foggy, overcast, glare) Sensor performance degradation (e . g. , reduced contrast, motion blur in rain)
[0068] The resulting labeled data— comprising input image sequences, obj ect traj ectories, and environmental metadata— is used to train machine learning models with built-in domain robustness .
[0069] Machine learning models are trained not only with raw obj ect traj ectories but also with:
[0070] Wind vectors and environmental metadata as auxiliary input channels
[0071] Sensor degradation simulations to enforce resilience against visual obstructions
[0072] The model learns to disentangle obj ect motion from atmospheric distortion, enabling more accurate tracking and prediction under real-world conditions .
[0073] During deployment, the trained model can receive real-time environmental data from external sources (e . g. , weatherAPIs, on-site sensors) . The inference engine dynamically adjusts traj ectory prediction based on:
[0074] Live wind speed and direction
[0075] Temperature and barometric pressure (affecting shell drag coefficients)
[0076] Visibility and lighting conditions (affecting camera reliability)
[0077] Alternatively, when environmental data is unavailable or unreliable, the model ' s weather-invariant training allows it to infer probable conditions based on visual cues (e . g. , rain streaks, haze) and adjust predictions accordingly. In one embodiment, deploying of the defense system comprises entering the position of the digital stereo camera, the gun and / or the drone into the computer to precalibrate the virtual model .
[0078] The embodiments mentioned of the method for defending at least one area can be used in any combination within the scope of the claims, provided they do not contradict each other .
[0079] A defense system according to the invention for defending at least one area comprises a digital stereo camera, a computer, a gun and / or a drone that are controlled by the computer, all of which are dedicated to the at least one area, wherein the computer comprises a virtual model that is created and trained as claimed in claim 1.In one embodiment, the system comprises two or more digital stereo cameras, and / or two or more computers that are operatively connected to one another and / or two or more guns and / or drones .
[0080] The embodiments mentioned of the defense system for defending at least one area can be used in any combination within the scope of the claims, provided they do not contradict each other .
[0081] BRIEF DESCRIPTION OF THE FIGURES
[0082] Embodiments of the present invention are explained in more detail below using figures . These are for explanation purposes only and are not to be interpreted as limiting. It shows
[0083] Fig. 1 a schematic perspective representation of a system according to the invention for defending at least one area; and
[0084] Fig. 2 a flow chart of a method for defending an area according to the invention.
[0085] DETAILED DESCRIPTION OF THE INVENTION
[0086] Fig. 1 shows a schematic perspective representation of a system according to the invention for defending at least one area 1. The shown area 1 is a square . However, a circular area is indicated. Above the area 1, a dome-shaped airspace 2 is indicated.The airspace 2 comprises superimposed and adjoining shells 20; 21, wherein a first shell 20 defines an outside of the airspace 2 and wherein a last shell 21 defines an inside of the airspace (2 ) . Each shell 20; 21 is defined by a difference in radius r, and each shell 20; 21 is divided azimuthal and polar into finite volumes 200, wherein each finite volume 200 is defined by a difference in azimuth angle 0 and by a difference in polar angle . The finite volumes 200 of different shells 20; 21 have the same difference in azimuth angle 0 and have the same difference in polar angle . A digital stereo camera 3 is arranged in a central region of the area 1, oriented towards the airspace 2. A computer 4 by the digital stereo camera 3. Two guns 5 are arranged in the area 1, off the central region and two drones 6 are hovering in a region above the central region of the area 1. The digital stereo camera 3, the guns 5 and the drones 6 being operatively connected to the computer 4. A target 7 is entering the first shell 20 of the airspace 2.
[0087] Fig. 2 shows a flow chart of a method according to the invention for defending at least one area 1.
[0088] Setting up 80 a defense system is done by:
[0089] Defining 800 the at least one area and defining 800 a corresponding airspace above the area;
[0090] Defining 801 system components dedicated to the at least one area, selected from the group comprising digital stereo camera, computer, gun and / or drone that are controlled by the computer;Scanning 802 a surface of the area or scanning the surface of the area and scanning the corresponding airspace .
[0091] Creating 81 a virtual model of the surface of the area and the corresponding airspace is done by:
[0092] Transferring 811 data from the scanning 802 to the computer;
[0093] Emulating 812 the digital stereo camera, the gun and / or the drone in the virtual model on the computer;
[0094] Emulating 813 the surface of the area or the surface of the area and the corresponding airspace in the virtual model on the computer;
[0095] Segmenting 814 the corresponding airspace into finite volumes .
[0096] Training 82 the virtual model is done by:
[0097] Introducing 821 virtual targets in the virtual model; Intercepting 822 the virtual targets with the gun and / or with the drone .
[0098] Deploying 83 the defense system is done by:
[0099] Positioning 831 the digital stereo camera, the gun and / or the drone in the area;
[0100] Calibrating 832 the virtual model with current images from the digital stereo camera;
[0101] Activating 833 the defense system by using the virtual model to intercept actual targets .REFERENCE SIGN LIST
[0102] 1 Area 813 Emulating airspace 2 Airspace 814 Segmenting airspace 20 First shell 82 Training virtual 200 Finite volume model
[0103] 21 Last shell 821 Introducing virtual 3 Digital stereo camera targets
[0104] 4 Computer 822 Intercepting the 5 Gun virtual targets 6 Drone 83 Deploying the system 7 Target 831 Positioning system components
[0105] 80 Setting up the system 832 Calibrating the 800 Defining area and virtual model airspace ; 833 Activating defense 801 Defining system
[0106] components 0 Azimuth angle
[0107] 802 Scanning 4? Polar angle
[0108] 81 Creating virtual r Radius
[0109] model X First direction 811 Transferring data Y Second direction 812 Emulating system Z Third direction
[0110] components
Claims
CLAIMS1 . A method for defending at least one area ( 1 ) , comprising the steps of :Setting up ( 80 ) a defense system by :Defining ( 800 ) the at least one area ( 1 ) and defining ( 800 ) a corresponding airspace ( 2 ) above the area ( 1 ) ;Defining ( 801 ) system components dedicated to the at least one area ( 1 ) , selected from the group comprising digital stereo camera ( 3 ) , computer ( 4 ) , gun ( 5 ) and / or drone ( 6 ) that are controlled by the computer ( 4 ) ;Scanning ( 802 ) a surface of the area ( 1 ) or scanning the surface of the area ( 1 ) and scanning the corresponding airspace ( 2 ) ;Creating ( 81 ) a virtual model of the surface of the area ( 1 ) and the corresponding airspace ( 2 ) by :Trans ferring ( 811 ) data from the scanning ( 802 ) to the computer ( 4 ) ;Emulating ( 812 ) the digital stereo camera ( 3 ) , the gun ( 5 ) and / or the drone ( 6 ) in the virtual model on the computer ( 4 ) ;Emulating ( 813 ) the surface of the area ( 1 ) or the surface of the area ( 1 ) and the corresponding airspace ( 2 ) in the virtual model on the computer ( 4 ) ;Segmenting ( 814 ) the corresponding airspace ( 2 ) into finite volumes ;Training ( 82 ) the virtual model by :Introducing ( 821 ) virtual targets ( 7 ) in the virtual model ;Intercepting ( 822 ) the virtual targets ( 7 ) with the gun ( 5 ) and / or with the drone ( 6 ) ;Deploying ( 83 ) the defense system by :Positioning ( 831 ) the digital stereo camera ( 3 ) , the gun ( 5 ) and / or the drone ( 6 ) in the area ( 1 ) ; Calibrating ( 832 ) the trained virtual model with current images from the digital stereo camera ( 3 ) ; Activating ( 833 ) the defense system by using the trained virtual model to intercept actual targets ( 7 ) .2 . The method of claim 1 , wherein two or more areas ( 1 ) with their corresponding airspaces ( 2 ) are defined and wherein the virtual model of each surface area ( 1 ) and each corresponding airspace is created ( 81 ) and trained ( 82 ) by one single computer ( 4 ) or by two or more computers ( 4 ) that are operatively connected to one another .3 . The method of claim 1 or 2 , wherein the corresponding airspace ( 2 ) is dome-shaped, hemispherical , cuboid, cubeshaped, or prism-shaped .4 . method of one of claims 1 to 3 , wherein for each of the at least one area ( 1 ) , the system comprises two or more dedicated digital stereo cameras ( 3 ) , and / or two or more dedicated computers ( 4 ) that are operatively connected to one another and / or two or more dedicated guns ( 5 ) and / or two or more dedicated drones ( 6 ) .5 . The method of one of claims 1 to 4 , wherein the scanning ( 802 ) is done by LiDAR systems on fixed or mobile platforms and / or wherein the scanning ( 802 ) is done by unmanned aerial vehicles equipped with high-resolution RGB and / or multispectral imaging systems and / or wherein the scanning ( 802 ) is done by multispectral or hyperspectral imaging sensors and / or high-dynamic-range imaging systems and / or time-lapse cameras .6 . The method of one of claims 1 to 5 , wherein emulating ( 813 ) the surface of the area ( 1 ) comprises integrating publicly available topographic data .7 . The method of one of claims 1 to 6 , wherein segmenting ( 814 ) the airspace ( 2 ) comprises dividing the airspace ( 2 ) into superimposed and adj oining shells ( 20 ; 21 ) , wherein a first shell ( 20 ) defines an outside of the airspace ( 2 ) and wherein a last shell ( 21 ) defines an inside of the airspace ( 2 ) , wherein each shell ( 20 ; 21 ) is defined by a di f ference in radius ( r ) , and wherein each shell ( 20 ; 21 ) is divided azimuthal and polar into finite volumes ( 200 ) , wherein eachfinite volume (200) is defined by a difference in azimuth angle (0) and by a difference in polar angle (4?) or wherein each finite volume (200) is defined by a difference in a first direction (X) , by a difference in a second direction (Y) , and by a difference in a third direction (Z) .
8. The method of claim 7, wherein the finite volumes (200) of different shells (20; 21 ) have the same size or have different sizes, or wherein the finite volumes (200) of different shells (20; 21 ) have the same difference in azimuth angle (0) and the same difference in polar angle (<t) .
9. The method of one of claims 1 to 8, wherein introducing ( 821 ) virtual targets (7 ) is creating virtual images of the emulated digital stereo camera (3) and wherein these virtual images are used to predict the velocity and traj ectory of the virtual targets (7 ) .
10. The method of claim 9, wherein the intercepting ( 822 ) is done by using the predicted velocities and traj ectories to control the emulated gun (5) and / or emulated drone ( 6) .
11. The method of one of claims 1 to 10, wherein deploying ( 83) of the defense system comprises entering the position of the digital stereo camera (3) , the gun (5) and / or the drone ( 6) into the computer (4 ) to pre-calibrate the virtual model .12 . A defense system for defending at least one area ( 1 ) , comprises a digital stereo camera ( 3 ) , a computer ( 4 ) , a gun ( 5 ) and / or a drone ( 6 ) that are controlled by the computer ( 4 ) , all of which are dedicated to the at least one area ( 1 ) , wherein the computer ( 4 ) comprises a virtual model that is created ( 81 ) and trained ( 82 ) as claimed in claim 1 .13 . The defense system of claim 12 , wherein the system comprises two or more digital stereo cameras ( 3 ) , and / or two or more computers ( 4 ) that are operatively connected to one another and / or two or more guns ( 5 ) and / or drones ( 6 ) .
