Method and system for AI-based autonomous security agent
A system for an automated processing of security data including a processor of a Home Security Agent (HAS) node configured to host a machine learning (ML) module and connected to security data capture array and to at least one controller of a drone and a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to: acquire security data from the security data capture array entity reflecting a threat subject extract a set of classifying features reflecting positioning and movements of the threat subject relative to the protected home; generate a feature vector based on the set of classifying features; receive a plurality of security parameters from a security predictive model generated by the ML module using outputs of the ANN based on the classifier feature vector; and generate a threat subject verification verdict based on the classifier feature vector.
1 . A system for an automated processing of security data, comprising:
a processor of a Home Security Agent (HAS) node configured to host a machine learning (ML) module and connected to security data capture array and to at least one controller of a drone over a wireless network connection; and
a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to:
acquire security data from the security data capture array entity reflecting a threat subject in a vicinity of a protected home, the security data comprising a plurality of signals;
normalize the plurality of signals for depth consistency;
extract a set of classifying features reflecting positioning and movements of the threat subject relative to the protected home from the normalized signal;
generate at least one classifier feature vector based on the set of classifying features;
provide the at least one classifier feature vector to the ML module coupled to an Artificial Neural Network (ANN);
receive a plurality of security parameters from a security predictive model generated by the ML module using outputs of the ANN based on the at least one classifier feature vector;
generate a threat subject verification verdict based on the at least one classifier feature vector;
generate a control command for operation of the drone based on the threat subject verification verdict; and
send the control command to a target controller of the drone.
2 . The system of claim 1 , wherein the machine-readable instructions that when executed by the processor, cause the processor to record the plurality of security parameters along with the threat subject verification verdict on a permissioned blockchain for creation of a security audit log.
3 . The system of claim 1 , wherein the machine-readable instructions that when executed by the processor, cause the processor to derive the set of classifying features comprising any of: presence of the threat subject within a range from the protected home, a speed of movement of the threat subject, a trajectory of the movement of the threat subject, a distance from the of the threat subject to at least one entry point of the protected home, and an orientation of the threat subject relative to the at least one entry point of the protected home.
4 . The system of claim 2 , wherein the machine-readable instructions that when executed by the processor, cause the processor to continually track the distance from the threat subject to the at least one entry point to produce the threat subject verification verdict responsive to the distance reaching a pre-set threshold distance value.
5 . The system of claim 3 , wherein the machine-readable instructions that when executed by the processor, cause the processor to send the control command for activation of the drone responsive to the distance reaching a pre-set threshold distance value.
6 . The system of claim 4 , wherein the machine-readable instructions that when executed by the processor, cause the processor to acquire additional security data from the drone.
7 . The system of claim 5 , wherein the machine-readable instructions that when executed by the processor, cause the processor to generate the at least one classifier feature vector based on the set of classifying features and the additional sensory data acquired from the drone.
8 . The system of claim 6 , wherein the machine-readable instructions that when executed by the processor, cause the processor to adjust the threshold distance based on outputs of the security predictive model generated based on the speed of movement of the threat subject and the trajectory of movement of the threat subject.
9 . The system of claim 6 , wherein the machine-readable instructions that when executed by the processor, cause the processor to increase the threshold distance value based on at least one security parameters generated by the ML module based on the speed of movement of the threat subject and the trajectory of the movement of the threat subject combined with the orientation of the threat subject relative to the at least one entry point.
10 . The system of claim 1 , wherein the machine-readable instructions that when executed by the processor, cause the processor to continuously monitor the threat subject by the security data capture array to provide an updated current set of classifying features to generate the at least one classifier feature vector to be ingested into the ML module configured to generate an updated set of security parameters for generation of an updated threat subject verification verdict in real-time.
11 . The system of claim 10 , wherein the machine-readable instructions that when executed by the processor, cause the processor to generate the at least one classifier feature vector based on the set of classifying features and the local remote historical property security data combined with the remote historical property security data.
12 . The system of claim 1 , wherein the machine-readable instructions that when executed by the processor, cause the processor to retrieve local and remote historical property security data from at least one local and at least one remote database based on the set of classifying features, wherein the remote historical data is collected at remote locations employing the security data capture array for collecting security data.
13 . A method for an automated processing of security data, comprising:
acquiring, by a Home Security Agent (HAS) node configured to host a machine learning (ML) module, security data from a security data capture array entity reflecting a threat subject in a vicinity of a protected home, the security data comprising a plurality of signals;
normalizing, by the HAS node, the plurality of signals for depth consistency;
extracting, by the HAS node, a set of classifying features reflecting positioning and movements of the threat subject relative to a protected home from the normalized signal;
generating, by the HAS node, at least one classifier feature vector based on the set of classifying features;
providing, by the HAS node, the at least one classifier feature vector to the ML module coupled to an Artificial Neural Network (ANN);
receiving, by the HAS node, a plurality of security parameters from a security predictive model generated by the ML module using outputs of the ANN based on the at least one classifier feature vector;
generating, by the HAS node, a threat subject verification verdict based on the at least one classifier feature vector;
generating, by the HAS node, a control command for operation of the drone based on the threat subject verification verdict; and
sending, by the HAS node, the control command to a target controller of the drone.
14 . The method of claim 13 , further comprising recording the plurality of security parameters along with the threat subject verification verdict on a permissioned d blockchain for creation of a security audit log.
15 . The method of claim 14 , further comprising sending the control command for activation of the drone responsive to the updated threat subject verification verdict.
16 . The method of claim 13 , further comprising deriving the set of classifying features comprising any of: presence of the threat subject within a range from the protected home, a speed of movement of the threat subject, a trajectory of the movement of the threat subject, a distance from the of the threat subject to at least one entry point of the protected home, and an orientation of the threat subject relative to the at least one entry point of the protected home.
17 . The method of claim 16 , further comprising acquiring additional security data from the drone.
18 . The method of claim 13 , further comprising continually tracking the distance from the threat subject to the at least one entry point to produce an updated threat subject verification verdict responsive to the distance reaching a pre-set threshold distance value.
19 . The method of claim 18 , further comprising continuously monitoring the threat subject by the security data capture array to provide an updated current set of classifying features to generate the at least one classifier feature vector to be ingested into the ML module configured to generate an updated set of security parameters for generation of an updated threat subject verification verdict in real-time.
20 . A non-transitory computer-readable medium comprising instructions, that when read by a processor, cause the processor to perform:
acquiring security data from a security data capture array entity reflecting a threat subject in a vicinity of a house, the security data comprising a plurality of signals;
normalizing the plurality of signals for depth consistency;
extracting a set of classifying features reflecting positioning and movements of the threat subject relative to a protected home from the normalized signal;
generating at least one classifier feature vector based on the set of classifying features;
providing the at least one classifier feature vector to a machine learning (ML) module coupled to an Artificial Neural Network (ANN);
receiving a plurality of security parameters from a security predictive model generated by the ML module using outputs of the ANN based on the at least one classifier feature vector;
generating a threat subject verification verdict based on the at least one classifier feature vector; and
generating a control command for operation of the drone based on the threat subject verification verdict; and
sending the control command to a target controller of the drone.