WO2022172217A1

WO2022172217A1 - System and method for rogue drone detection and interception

Info

Publication number: WO2022172217A1
Application number: PCT/IB2022/051245
Authority: WO
Inventors: Kamalakar DEVAKI; Jesudas Victor FERNANDES
Original assignee: Devaki Kamalakar; Fernandes Jesudas Victor
Priority date: 2021-02-11
Filing date: 2022-02-11
Publication date: 2022-08-18

Abstract

According to the invention there is provided a system and a method to comprehensively analyse and predict a rogue drone. It uses machine learning and artificial intelligence as a part of the system, thereby digitizing the drone detection technology. The subject invention provides a system and method consisting of a hardware and software component. The hardware components of the system automatically capture various data from multiple drone detection systems and provides the data to the artificial intelligence software. The software module interfaces with the hardware components and applies several deep learning techniques to achieve detection, recognition of rogue drones and deciding the interception technique to defect the rogue drones.

Description

SYSTEM AND METHOD FOR ROGUE DRONE DETECTION AND INTERCEPTION

Field of the Invention:

The invention relates generally to the field of drones or unmanned aerial vehicles. More specifically, the invention relates to systems and methods for detecting and intercepting rogue drones using artificial intelligence.

Background of the Invention

The term ‘drone’ or an ‘unmanned aerial vehicle’ (UAV) is usually defined as an aircraft which does not have a pilot onboard, or is controlled remotely, or is an autonomous drone. Drones operates through a combination of technologies, including computer vision, artificial intelligence, object avoidance tech etc. and can also be ground or sea vehicles that operating autonomously. Originally applied in military and aerospace field, drones are now being widely harnessed for a variety of commercial purposes across industries.

While there are many legitimate uses for UAVs, their misuse can interfere with or threaten the safety or security of other aerial vehicles, functions performed by public safety officers and can also cause public nuisance and/or public safety concerns. Unregulated drones, UAVs and remotely-piloted aircraft system are a "potential threat" to vital installations, sensitive locations and specific events and a "compatible solution" is required to counter them. According to a report, India has an estimated over six lakh unregulated unmanned aerial vehicles (UAVs) of various sizes and capacities, are present in the country and anyone of them can be used for launching a nefarious act by disruptive element. Security agencies are analyzing modern anti-drone weapons like 'sky fence' and 'drone gun' to counter terror or similar sabotage bids by these aerial platforms.

There are already various drone detection devices available in prior art. Most of the conventional drone detection methods and systems use one or more sensors in order to detect a drone. Detecting rogue drones is challenging given the fly-zone of drone and also, to add to the complexity, drones are being used by general public for various purposes like shooting events like marriages, functions etc. In the near future, the delivery of goods by businesses will be done using drones. India has already started issuing licenses for such drones and enforcing certain regulations to be followed for the drone hardware and also for users & operators. Given the fly-zone of drone which starts from 50-100 feet till 400ft which based on the VLOS (Visual Line of Sight) in India. Detecting object at this fly-zone throws multiple challenges be it technology choices we have in place and practicality issues with respect to identifying the rogue drone.

While in the urban environments with domestic buildings around important assets / installations and also people moving around, the techniques to detect the rogue drones leave us with few choices. The rural scenario which are at outskirts of cities / towns, villages or even those in deserts, forests and difficult terrains gives the flexibility of deploying various techniques but offers a different set of challenges.

Rogue Drone Interception - Challenges

1. Buildings and other infra with in the cities cause a lot of objects to be detected, which are not of our interest.

2. Rural / Remote deployments are more vulnerable to attacks when it comes to rogue drones.

3. Usage of satellites is not feasible.

4. Centralized solution poses a challenge due to size and diversity of weather, topology of India, it needs to be federated so that there is no single point of failure.

5. With the new regulations from DGCA the incidences of rogue drones will be deterred but cannot be ruled out, DGCA can effectively handle the registered drones, while those which are un-registered are still a threat.

6. Rogue drones flown with malicious intent will always be a challenge to handle.

The subject invention addresses the limitations of the prior art by proposing an intelligent, and precision Rogue drone detection and recognition solution which works in a coordinated manner to comprehensively analyse and predict a rogue drone. It uses machine learning and artificial intelligence as a part of the system, thereby digitizing the drone detection technology.

The subject invention provides a system and method consisting of a hardware and software components. The hardware components of the system automatically capture various data from multiple drone detection systems and provides the data to the artificial intelligence software. The software module interfaces with the hardware components and applies several deep learning techniques to achieve detection, recognition of rogue drones and deciding the interception technique to defect the rogue drones.

Problems with the Prior Art:

United States Patent No. 09704508 titled as “Drone detection and classification methods and apparatus” deals with a system, method, and apparatus for drone detection and classification. An example method includes receiving a sound signal in a microphone and recording, via a sound card, a digital sound sample of the sound signal, the digital sound sample having a predetermined duration. The method also includes processing, via a processor, the digital sound sample into a feature frequency spectrum. The method further includes applying, via the processor, broad spectrum matching to compare the feature frequency spectrum to at least one drone sound signature stored in a database, the at least one drone sound signature corresponding to a flight characteristic of a drone model. The method moreover includes, conditioned on matching the feature frequency spectrum to one of the drone sound signatures, transmitting, via the processor, an alert.

The aforementioned invention is using the drone sound to create a signature library, and not the sensor fusion technology for drone detection and interception as the present invention.

Patent Application No. PCT/SG2016/050434 titled as “System and Method for detecting, intercepting and taking over control of multiple rogue drones simultaneously” deals with a system for detecting, intercepting and taking over control of multiple drones simultaneously, comprising a scanner with multiple antenna array that supports scanning frequencies of 433 MHz, 2.4 GHz and 5.0-5.8 GHz, a GPS spoofer, a radio frequency (RF) jammer and a main system on a Linux based operating system. The scanner receives and transmits signals, such as de authentication signal to the multiple drones. The GPS spoofer spoofs the location information. The RF jammer jams drones that cannot be intercepted and taken control of. The main system is loaded with central management software for managing the detecting, intercepting and taking over control of the drone using the scanner, GPS spoofer and RF jammer.

Radio-frequency scanners / monitoring system mentioned in the aforementioned invention is used as one of the inputs to the drone detection system of the subject invention and the RF jammers are used as an interception or disruption device for the detected rogue drones. Patent Application No. PCT/EP2017/082258 titled as “Drone Detection Method” deals with a drone detection method and a drone detection system, a drone being detected by means of a sensor and sensor data being acquired. In order to improve said drone detection method and said drone detection system, sensor data are compared to reference data stored in a data processing system and on the basis of this comparison it is detected whether the drone carries a payload or not.

The aforementioned invention discusses how a drone can be detected by means of a sensor.

It does not use the PCL system for drone detection and tracking. Also, there is no mention of corroboration of inputs from all sensors for the purpose of Rogue classification and threat perception. This invention also does not cover on the interception or disruption part to defeat the detected rogue drones.

Patent Application No. CA3000005 titled as “Drone Detection Systems” deals with Methods, systems, and apparatus, including computer programs encoded on storage devices, for drone- augmented emergency response services. In one aspect, a device includes a network interface, one or more sensors, one or more processors, and one or more storage devices that include instructions that are operable to perform operations. The operations include monitoring a predetermined geographic area that surrounds a particular property, determining that a drone device is within the predetermined geographic area that surrounds the particular property, determining whether the drone device that is detected within the predetermined geographic area that surrounds the property is an unauthorized drone device, and in response to determining that the drone device that is detected within the predetermined geographic area that surrounds the property is an unauthorized drone device, transmitting a signal indicating the detection of the unauthorized drone device within the predetermined geographic area that surrounds the property.

The aforementioned invention discusses drone detection systems but does not claims PCL and Passive radars system which is used like the present invention. The AI box feature and sensor fusion methodology is also not mentioned herein.

None of the prior art references combine all three aspects of the subject invention:

1. AI engine to corroborate multiple input sources / sensor inputs for Drone detection and Rogue classification 2. Multiple sources for drone detection, like RADAR, PCL + Passive RADAR + RGB camera + Thermal camera + RF Scanner

3. Multiple output systems for interception, disruption of Rogue drones.

Objects of the Invention:

• The main object of the invention is aimed at detection of the flying objects.

• A further object of the invention is recognition of the flying object as a drone/UAV.

• A further object of the invention is tracking of the detected drone/UAV path.

• A further object of the invention is to provide the tracking of the drones in real-time.

• A further object of the invention is to calculate the distance and other parameters of the detected drone/UAV.

• A further object of the invention is to enable the decision for interception & neutralisation of the detected drone/UAV.

• A further object of the invention is to execute the detection/tracking process in an economical and cost-effective manner.

• A further object of the invention is to use the already existing infrastructure to repurpose or extend in achieving the solution.

• A further object of the invention is to ensure that the solution should be adoptable across the terrain, geography protecting both urban & rural important deployments, assets and personnel.

• A further object of the invention is to send alerts and notifications to the regulatory authorities on the detection of a rogue drone/UAV.

• A further object of the invention is to ensure that the solution follows DGCA rules and regulations are also covered as part of the solution thus enhancing the effectiveness of the regulations.

Statement and Summary of the Invention:

According to the invention there is, therefore, provided a device, a system and a method to comprehensively analyse and predict a rogue drone. It uses machine learning and artificial intelligence as a part of the system, thereby digitizing the drone detection technology. The subject invention provides a system and method consisting of a hardware and software components. The hardware components of the system automatically capture various data from multiple drone detection systems and provides the data to the artificial intelligence software.

The software module interfaces with the hardware components and applies several deep learning techniques to achieve detection, recognition of rogue drones and deciding the interception technique to defect the rogue drones.

The subject invention consists of:

(a) A primary mechanism or means for detecting the flying object and recognition of the same as a drone/UAV.

(b) A primary mechanism or means for tracking of the detected drone/UAV.

(c) An auxiliary mechanism or means of generating alerts on the breach of a pre-set threshold values for various parameters to the regulatory authorities. The alerts are generated on the detection of a rogue drone/UAV.

(d) A primary mechanism or means of interception & neutralisation of the detected drone/UAV.

Description:

The description of the preferred embodiment is meant to demonstrate the broad working principles of the invention without limitation as to possible adaptations, extensions, applications etc., which would be obvious to a person skilled in the art. This invention is illustrated in the accompanying drawings, throughout which, like reference numerals indicate corresponding parts in the various figures. In the interest of brevity and for the purposes of exemplary explanation, references have been made to a system, depicted in figures 1 to 8 herein without limitation, to describe the invention which is essentially directed towards providing the user with option to comprehensively analyse and predict a rogue drone.

Figure 1 gives a schematic overview of the functional units that comprise the system and comprises a plurality of components such as the AI Box (1), RF UAV Jammers (2), RF Scanners (3), UAV/Drone detection Radar (4), RGB+ Thermal Camera (5), AI Server (6) PCL System (7), Passive Radars (8), DGCA- Drones/UAV (9), AI Server (10), Zonal Command (11), EMP generator (12), Net Guns (13), Net Thrower (14) and Laser Guns (15).

Figure 2 illustrates the input sources to the AI Box.

Figure 3 illustrates the output sources which the AI box interfaces with.

Figure 4 represents the AI box and the interfacing components.

Figure 5 illustrates the general steps of the rogue drone/UAV detection and recognition and the interception and neutralisation of the detected rogue drone.

Figure 6 describes the step-by-step flowchart for the data processing in the method of flying object detection and drone recognition and threat perception and prediction.

Figure 7 illustrates the RF Scanner and UAV Protocol Jammer system.

Components of the Invention

The subject invention comprises the following:

1. Artificial Intelligence box

This hardware component (herein referred to as AI box) is the heart of the system, which controls the entire system functionality. It processes the input signals from various drone detection equipment and generates the output signals to command and operate drone interception equipment. The AI box supports interfaces to all the drone detection equipment and drone interception equipment listed. The software components of the system execute on the AI box.

The AI box supports interfaces to the following equipment.

2. RF scanner / monitoring equipment:

Radio-frequency (RF) scanners detects and track UAVs / drones based on their communication signature. Deep learning and machine learning algorithms scan known radio frequencies and find and geolocate RF-emitting drones, in all weather conditions. With multiple RF scanners triangulation is possible.

This is a cost-effective solution than radar sensors with a medium-range up to 600 m, which can be increased with better antenna and sensor configurations. It can provide Early warning capability even before UAV takes-off, if the RF communication signal is on. Drones use different frequencies for control, telemetry and video or audio transmission. They can also switch to custom frequency channels for communication which make them robust against jamming.

The RF scanners will sense the RF communication signals and then extract features like Signal power, Noise to Signal threshold value, OFDM parameters like Cyclic prefix length, FFT size, Sub carrier spacing, Symbol duration, etc. These features are then gathered for training a Deep learning neural networks which can then classify a given drone type.

RF Scanner system:

There are two types of RF Scanners, one which scans the airspace for RF bands used by UAV / drones, second which scans RF bands for specific protocols, which are generally used by the UAV / Drones.

Radiofrequency scanner constantly analyzes wide frequency bands, classifies and decodes signals and provides early warnings often even before the drone becomes airborne. As soon as drone in the detection radius of the RF Sensor establishes connection with its remote controller, radiofrequency scanners detect this communication. The Scanning may be for a specific frequency band (carrier) or for signals (protocol) being carried over the band. RF protocol scanning can also identify the UAV. RF Scanners can detect an UAV up to 5 kms.

RF Scanners can detect a presence of a UAV but will provide the location of the UAV. Direction may be available based on direction of RF scanners. Multiple RF scanners can be used to cover 360 deg. RF scanners provide a “UAV detected” trigger signal, along with details of the type of UAV, to the RDDS engine for further action.

3. Radar:

Radars work by transmitting a signal in a particular direction. When the signal meets an object, it reflects off of it, and the radar receives this reflected signal. When that happens, the object that has reflected the signal back to the radar appears on the radar. Judging by the time it takes for the signal to return the distance between the radar and the object is calculated. Radars that can run on a higher resolution consume more power but can provide a lot more accurate and detailed detection. With radars that are dialed up to use very high resolution or frequency, you will be able to see almost anything like Birds, Clouds, Rain, Snow, Auroras, and Meteors.

With high frequency, accurate, and multiple Radars, detection systems can detect objects (and drones) that have a radar cross-section (RCS) of just 0.01 m2 at a distance of 5 kms to 10 kms. Popularly used drones usually have an average RCS of about 0.01 to 0.02 m2 as compared to birds, which have a radar cross-section of about 0.01 to 0.001 m2. This poses the biggest challenge in detecting whether the small flying object is a drone or a bird.

Challenges faced in drone detection using Radars are as follows: Drone not in line of sight (LOS), or covered by obstacles like trees or buildings, drone is too small, difficult terrain, is covered de earth’s curvature, etc.

Over the horizon Radars can help overcome these challenges where Radar signal reflected of the ionosphere can be used to detect object which are not in LOS and covered by difficult terrain.

UAV detection RADARs

With the best technology for target identification and rejection filters (Doppler Signature Analysis), these RADARs are meant for detection of the low flying targets. These RADARs employ the state-of-the-art Single pulse elevation determination and stereo Doppler channels.

The RADAR detects the drones and provides the precise location of the drone. The RADARs also incorporate smart micro doppler filtering mechanism to avoid false alarms and improve detection. These RADARs contain multiple transmit and receive sections, and the radar-beam focusing its energy on detection of targets. The output of the RADAR is provided to AI box for further action.

Multiple RADARs will be used for get a full coverage of the airspace which is to be protected. The output of the RADAR will be provided to the RDDS engine for further action.

Passive radar: Passive Radars do not transmit any signal on its own, but depends on the signals from other broadcast and communication transmitters for deducing target location. Broadcast and communications transmitters are at higher locations and hence cover a broad area. Since the PCL system makes use of existing transmitters, the cost of a passive radar is likely to be much lower than a conventional radar.

Passive radar allows the use of frequency bands (particularly VHF and UHF) which is not used in Radars. Such frequencies may be beneficial in detecting stealthy targets, since the wavelength is of the same order as the physical dimensions of the target, and forward scatter gives a relatively broad angular scatter. Since the Passive radar does not emits any signal, and as long as the receive antenna is inconspicuous, the passive radar receiver may be undetectable and covert. Hence it is difficult to deploy counter measures against a passive radar.

The signals of broadcast transmissions are not optimized for radar purposes, hence it is required to select the signal source and process them in a very optimum manner to meet the objective.

4. Passive Coherent Location (PCL) System:

PCL system uses Passive radars to receive direct and target reflected signals from other sources like Cellular base station or FM/Digital audio broadcasting. Since the number of the cellular base stations and FM transmitter may be more in a given region, there will be a lot of signals which include the direct signals as well. PCL system receives all the signals, processes, declutters, eliminates to choose the ones of interest.

The design of a PCL system goes through the following steps.

Choose the location with critical infrastructure to be protected.

Survey of the location for FM transmitters or cellular base stations and get the list of all the transmitter.

Study all the ambient radio signals and detect reflections from moving and stationary targets and understand pattern.

Choose a best transmitting source based on location, signal strength illumination and beam width. Decide the location of the receiver and design the receiver antenna array (configuration dimensions excitation elements).

Synthesis of the pattern of the signal environment, design the null placements for reducing interference.

Deployment of the PCL system and detection of target passively steer nulls to reduce interference.

Passive RADAR and Passive Coherent location

This is a stealth radar technology which work passively and relying on the RF transmissions of FM, Digital audio or Digital broadcasts to detect a target in the protected airspace.

This system has two parts, the Passive RADARs which are generally an array of antennas, and the PCL (passive coherent location) unit which works on the inputs from the RADARs to localize and find the location of the target. The PCL unit employs algorithms like A-CFAR (adaptive averaging constant false alarm rate), Pulse compressing (PC), Digital beam forming (DBF), Radial velocity estimation and judgement module to get better detections.

The elevation and the location details of the UAV, from the PCL system is provided to the RDDS engine for further processing and actions.

5. RGB Camera / Thermal Camera

The camera provides the raw image / video stream of the airspace. The camera will be installed on a prop which can rotate and capture the video stream of the airspace. The camera video stream will be accessed by the AI box and pre-processed before it is fed into algorithms and neural nets for detection of low flying drones/ UAVs. Apart from the detection, the neural nets will also help in recognising a drone apart from birds or kites, track the flying drone in its flight, gauge the speed of the drone.

Thermal cameras can also be used to detect UAVs and provide all-day all-weather coverage. Thermal camera stream together with the AI vision engine it can detect a UAV/ drone in all weather conditions. 6. RF Jammers:

These devices will jam all the frequencies in a given range and within a configured radial distance. This can jam all possible frequency bands, which are used by UAV/ Drones, frequency bands for GPS/GLONASS, etc. The power of the RF signal and the frequency range can be configured.

7. Net Guns:

Useful when the drone is visible and at close range. The user can shoot the drone /UAV with a net gun, which shoot a light weight, thin net over the drone. The net entangles the rotor blades thus disabling the drone/UAV.

8. Net throwers:

These are drones with are fitted with nets which can be unfurled on command and dropped. These drones fly above the rogue drones and drop the nets on them. The nets can be also fitted with parachute to bring the drone safely to the ground.

9. Laser guns:

Guns which send Laser beams toward the target which raises the temperature of the target to 1000 deg C, and the target get fried. All this happens at the speed of light.

10. EMP (Electromagnetic pulse) Suicide drones:

Drones fitted with EMP pulse generators will be used for suicide missions to neutralise drone or a swarm of drones. The suicidal drone once in the vicinity of the rogue drone/s will generate an EMP pulse to neutralise the rogue drone.

The drone carries and EMP generator which can generate an EMP pulse to disturb or destroy electronic components of the rogue drone and it will destroy the suicide drone as well.

11. DGCA server

DGCA, is the Directorate General for Civil Aviation in India. The DGCA has prepared an extensive framework of rules and regulations, together with systems for acquiring the necessary permissions and licenses for operating a drone and flying a drone in India. The main objective of DGCA is to ensure the skies and enable safe flying of drones/UAVs. The AI box will interact with the DGCA information service to corroborate the information of the detected drone to classify it as Rogue.

12. Zonal command center

Zonal command centre - Command centres which may be government based or setup for protected airspaces, which houses one or many drone interception equipment. The AI box will communicate / notify the Zonal command centre whenever a rogue drone has been detected and classified.

13. AI Server

This server shall be used for internal system transactions and storage of artefacts used and generated by the system.

System for drone detection and interception

The Drone detection and interception system comprises of the hardware and software modules. The hardware module of the system consists of the components as covered above.

The software module of the system, executes on the ‘Artificial Intelligence (AI) box’ has components like Rogue drone detection system (RDDS), AI vision engine for drone recognition and neutralisation.

The software accepts inputs and provides outputs to and from the following elements.

INPUTS:

1. RGB camera

2. Thermal camera

3. RF scanner

4. PCL UAV detector

5. Radar UAV detector

6. AI Server

7. Zonal command centre

8. DGCA OUTPUTS

1. RF Protocol j ammer

2. RF jammer

3. Laser guns

4. EMP suicide drones

5. AI Server

6. Notifications to Zonal command centre

7. Interception and Neutralisation commands through Zonal command centre

8. Notifications to DGCA

9. Alarm and Notifications system

Multiple input sources for the drone detection and prediction are interfaced to the AI box, which are then processed appropriately to detect the presence of the unidentified drone/UAV in the protected airspace. Multiple sources are used in combination to increase the accuracy and performance of the solution.

The AI box is enabled with an AI Vision engine, a set of algorithms and Deep learning neural nets which shall be used for features such as: -

1. Background subtraction

2. Drone detection

3. Flying object tracking using optical flow, re-identification

4. Classification between drones and other flying objects

5. Speed of flight of the flying object

The AI vision engine detects a low flying object and confirms if it is a drone or not. The detected drone is tracked during its flight, and the tracked frames and the angle of the camera rotation are sent to the RDDS engine for further action.

Method of drone detection and interception

The drone detection and interception system comprises: a) A primary mechanism or means for detecting the flying object and recognition of the same as a drone/UAV. b) A primary mechanism or means for tracking of the detected drone/UAV. c) An auxiliary mechanism or means of generating alerts on the breach of a pre-set threshold values for various parameters to the regulatory authorities. The alerts are generated on the detection of a rogue drone/UAV. d) A primary mechanism or means of interception & neutralisation of the detected drone/UAV.

The means of detecting the flying object and recognition of the same as a drone/UAV is achieved by the following combination of hardware and software components as follows: a. RGB and Thermal cameras capture images and video stream data of the airspace under protection, and this is fed to the software component of the AI box. b. RF Scanner system scans the frequency spectrum for frequency or frequency bands used by drones and if the frequency band was detected in the protected airspace, then the frequency or band code is sent to the software component. c. Similarly, the UAV RADAR and PCL system with Passive RADAR can be used as detectors which provide the elevation, radial angle of the detected drone d. The Deep learning neural nets predict the presence of the flying object and classify the object between a bird or a drone, from the images and video stream data received from RGB or Thermal cameras e. The result is further corroborated with the RF scanner input indicating whether the frequency band used by Drones was detected. f. The result can also be corroborated with the inputs from UAV RADAR and PCL system with Passive RADAR.

The tracking of the detected drone/UAV is achieved by the following combination of hardware and software components as follows: a. RGB and Thermal cameras capture video streams of the airspace under protection, and this is fed to the software component of the AI box b. The cameras can be mounted on a rotational prop in order to track the drone 360 degrees. c. The Deep learning neural nets predict the presence of the flying object and classify the object between a bird or a drone d. Once it is classified as a drone, the Object tracking algorithm starts tracking the drone from the video stream received from the cameras.

The means of generating alerts on the breach of a pre-set threshold values for various parameters to the regulatory authorities is achieved by the following combination of hardware and software components as follows: a. Using the interface with DGCA server the software component requests the data on the permitted drone flights, flight paths and the drone details from the DGCA server b. The Deep learning neural nets predict the presence of the flying object and classify the object between a bird or a drone from the video stream received from the cameras c. Once it is classified as a drone, the Object tracking algorithm starts tracking the drone and constructs the flight path. d. The Deep learning neural net will also predict the possible brand and model of the drone detected, which use the lookup table from the AI server e. In case the permitted and actual flight paths do not match, an alert is raised to the DGCA server with the necessary details f. The software component - RDDS takes inputs like Flight path, Drone brand and model, speed of drone, flight duration and make a threat prediction and this prediction is sent to DGCA and Zonal command center.

The interception & neutralisation of the detected drone/UAV is achieved by the following combination of hardware and software components as follows: a. Inputs received from drone detection components like Cameras, UAV RADAR, RF Scanners, PCL system with Passive RADARs is processed and analysed by RDDS module of the software component b. Once the RDDS component make a threat prediction level high, it issues alerts to Zonal command center and DGCA c. Based on the threat level and data received from drone detection components, the RDDS suggest the drone interception method d. The RDDS sends out intercept command with drone interception technique to be employed for the same. e. The drone interception systems like EMP - suicidal drones, Net throwers, Laser Guns, then execute the command and intercept the drone. f. The RDDS sends out intercept command to the Zonal command centre for employing drone interception.

The method of drone detection and interception thus, majorly consists of two modules: Drone detection module and Threat identification and escalation module.

In the ‘Drone detection module’, the AI engine for the RDDS (Rogue drone detection) embedded in the AI box takes inputs from multiple input sources of the system. Every input source (sensors) provides an event trigger of a drone detection to the RDDS engine. The input source details are processed and corroborated with other input sources to verify a detected flying object as a drone. The RDDS employs a Sensor Fusion methodology to make the accurate judgement of drones/UAVs. Sensor Fusion is the art of combining multiple sensors to produce an accurate “ground truth”. This is a method for integrating data provided by various sensors, in order to obtain the best Situational Awareness (SA). Sensor fusion algorithms are particularly useful in Unmanned System applications, where performance and reliability are desired, given a limited set of inexpensive sensors. Since no single technology has the capability of providing a total and reliable solution for UAV detection under all conditions, thus to increase the overall availability of the system, it may be necessary to include a number of sensors in the package. Typically, this could consist of a combination of optical, thermal camera, Radar, PCL, RF scanners.

In the ‘Threat identification and escalation module’, the RDDS engine of the AI box corroborates the data from the DGCA drone server to classify the detected drone as rogue drone. The AI box solution will always be in sync with the DGCA server and will get the following data from the DGCA in order to corroborate the RDDS output. a) List of drones registered with their types, weight and other details. b) List of drones registered for take-off with the start and destination locations, flight path data, time details.

The DGCA data is one of the criteria the RDDS engine uses along with the drone detection events from the system to classify if the drone detected is a rogue drone. The RDDS engine informs the DGCA and the Zonal command centre that a drone has been detected in the protected airspace and provides the details of the same. The zonal command checks their database for the list of drones which have been issued licenses and have permission for take-off. It verifies the current drone flight path against the permissible area of flight. In case the verification fails the zonal command classifies it as a rogue drone and tracks the drone’s movement. In parallel the zonal command predicts the direction of travel and the possible hit criteria. The threat detection and perception engine of the AI box also provides a score for threat level for the detected rogue drone. Additionally, the system of the present invention maintains its own secure database, which includes the data used for verification of the detected drones. The RDDS engine uses its secure database to make decisions on detected drone verification and interceptive methods to be employed in case of rogue drones. The database is also used by the RF scanner interface module to identify the drone type based on the drone/UAV RF protocol detected.

The RDDS engine readies the interceptive mechanism to counter the rogue drone/UAV. If the rogue drone flight continues to be threat and the threat level is termed as ‘high’, the interceptive mechanism is given a “GO” to counter / destroy the drones.

If the drone has not been classified as rogue and the threat possibility is not indicated, then the flight time and distance covered by the drone is monitored. If the drone travels for more 15 mins, the zonal command classifies the drone as “probable threat”. This is because in domestic drones the battery capacity is generally for a 15 mins flight duration and a limited distance.

In case the probable threat continues to persist, the zonal command AI prepares the catcher drones deployed at police stations as for rogue drones. In case of swarm of drones, the AI server prompts the Zonal commands of multiple rogue drones and the Zonal commands prepares the equipment to catch / shoot the drones accordingly.

Claims

We claim:

1. A drone detection and interception system comprising the following components:

(i) An artificial intelligence (AI) box

(ii) Drone detection systems including: a. RGB camera b. Thermal camera c. RF scanner d. PCL system with Passive RADARs e. UAV RADARs

(iii) Drone interception systems including: a. RF Jammers b. Laser guns c. Net throwers d. EMP suicide drones.

1. The system of claim 1 wherein the AI box is a hardware component embedded with software.

2. The system of claim 1 wherein the software component of the AI box consists of AI vision engine, Rogue drone detection (RDDS) algorithm and Sensor fusion algorithm.

3. The system of claim 1 wherein the AI box captures data from at least one of the drone detection systems mentioned in claim 1 and passes it down to its software component.

4. The system of claim 1 wherein the software component of the AI box provides command and control data to at least one drone interception system.

5. The system of claim 1 wherein the AI box interfaces with Zonal command centre(s) which may be government based or setup for protected airspaces and houses drone interception equipment.

6. The system of claim 1 wherein the AI box interfaces with DGCA - the Directorate General for Civil Aviation in India.

7. The system of claim 1 wherein the software component of the AI box requests transactional data and provides transactional data, command and control data to at least one of the following: a. Zonal command center b. AI server c. DGCA server

8. The system of claim 1 wherein the AI box uses an AI Server for internal transactions and storage as required.

9. The system of claim 4 wherein the data captured from RF Scanners is a code for a specific frequency or set of frequency bands used by drones detected in the airspace under protection and the code for the frequency or frequency band if detected.

10. The system of claim 4 wherein the data captured from the UAV Radar is whether a drone/s are detected in the airspace under protection using RADAR technologies and the elevation and location of the detected drone if detected.

11. The system of claim 4 wherein the data captured from the PCL system using Passive RADARs is the elevation and the location details of the drone detected in the airspace under protection.

12. The system of claim 4 wherein the data captured from the RGB and Thermal Cameras are the images and video data of airspace under protection.

13. The system of claim 5 wherein the command and control data provided to the to the RF Jammer system include the frequency or frequency band code being used by the drone and radial angle.

14. The system of claim 5 wherein the command and control data provided to the Laser guns,

Net throwers and EMP (Suicidal drones) is the location and flight path of the detected drone.

15. The system of claim 8 wherein the data provided by the software component of the AI box includes: a. Elevation, flight path and location details of the detected drone to Zonal command center, with the mode of drone interception to be employed b. Alerts and Notifications to the DGCA server regarding non-compliance of the rules by the rogue drone c. Rogue drone details like location information, brand and version, actual flight path followed to DGCA server d. Permitted flight and drone details along with date, time, flight path data from DGCA.

16. A drone detection and interception system comprising: a) A primary mechanism or means for detecting the flying object and recognition of the same as a drone/UAV. b) A primary mechanism or means for tracking of the detected drone/UAV. c) An auxiliary mechanism or means of generating alerts on the breach of a pre-set threshold values for various parameters to the regulatory authorities. The alerts are generated on the detection of a rogue drone/UAV. d) A primary mechanism or means of interception & neutralisation of the detected drone/UAV.

17. The system of claim 17 wherein the means of detecting the flying object and recognition of the same as a drone/UAV is achieved by components like RGB and Thermal cameras, RF Scanner system, UAV RADAR and PCL system with Passive RADAR and the AI box.

18. The system of claim 17 wherein tracking of the detected drone/UAV is achieved by the components like RGB and Thermal cameras and the AI box.

19. The system of claim 17 wherein generating alerts on the breach of a pre-set threshold values for various parameters to the regulatory authorities is achieved by the components like the DGCA server, cameras, AI server, RDDS and Zonal command center.

20. The system of claim 17 wherein interception & neutralisation of the detected drone/UAV is achieved by input components like Cameras, UAV RADAR, RF Scanners, PCL system with Passive RADARs, RDDS, Zonal command center and DGCA and drone interception systems like EMP - suicidal drones, Net throwers and Laser Guns.

21. A method of detecting a flying object and recognition of the same as a drone/UAV comprising of the following steps: a. Collection of inputs from multiple input sources of the system by the AI engine embedded in the AI box. b. Processing of the input source details and corroborating with other input sources to verify a detected flying object as a drone. c. Tracking of the detected drone during its flight, and sending the tracked frames and the angle of the camera rotation to the RDDS engine for further action. d. Use of Sensor Fusion methodology to make the accurate judgement of drones/UAVs by the RDDS.

22. A method of interception & neutralisation of the detected drone/UAV comprising the following steps: a) Corroborating the drone detection data from the DGCA drone server to classify the detected drone as rogue drone by the RDDS engine of the AI box. b) Generating alerts to the DGCA and the Zonal command centre that a rogue drone has been detected in the protected airspace and providing the details of the same by the RDDS engine. c) Suggesting the drone interception method by the RDDS based on the threat level and data received from drone detection components, d) Sending out the drone interception command with drone interception technique to be employed for the same by the RDDS to the Zonal command centre. e) Execution of the command by the drone interception systems like EMP - suicidal drones, Net throwers, Laser Guns.

23. The methods of claims 22 or 23 wherein the software component of the AI box applies the following techniques to achieve detection, recognition of rogue drones (RDDS) and deciding the interception technique to defeat the rogue drones: a. Deep learning algorithms for drone detection and localisation b. Machine learning algorithms for sensor data fusion and analysis. c. Image processing algorithms like Background subtraction, Re-identification.

24. The method of claim 23 wherein in case of swarm of drones, the AI server prompts the Zonal commands of multiple rogue drones and the Zonal commands prepares the equipment to catch / shoot the drones accordingly.

25. The method of claim 23 wherein the zonal command receives command and control data which consists of the direction of travel of the detected drone and the possible hit criteria.

26. The method of claim 23 wherein the threat detection and perception engine of the AI box also provides a score for threat level for the detected rogue drone.

27. method of claim 27 wherein the score is calculated by the RDDS software component, which take multiple inputs and parameters like inputs like Flight path, Drone brand and model, speed of drone, flight duration, inputs from DGCA for permitted drone flights.

28. The system of claim 1 maintains its own secure database, which includes the data used for verification of the detected drones.

29. The methods of claims 22 or 23 wherein the RDDS engine uses its secure database to make decisions on detected drone verification and interceptive methods to be employed in case of rogue drones.

30. The methods of claims 22 or 23 wherein the database is also used by the RF scanner interface module to identify the drone type based on the drone/UAV RF protocol detected.

31. The method of claim 22 wherein the AI box solution will always be in sync with the DGCA server and will get the following data from the DGCA in order to corroborate the RDDS output. a) List of drones registered with their types, weight and other details. b) List of drones registered for take-off with the start and destination locations, flight path data, time details.

32. The method of claim 23 wherein the RDDS engine uses the DGCA data as a criteria along with the drone detection events from the system to classify if the drone detected is a rogue drone.