” Artificial Intelligence platform developed by indigenous software Company with product engineering DNA can enhance the capabilities of the security force”
The combination of Artificial intelligence (AI) and Unmanned aerial vehicle(UAV) / drone can be a deadly cocktail. I am not talking about using Artificial Intelligence for flying ,monitoring or navigating the UAV, this is something what Artificial Intelligence has to offer and machine learning ensures. Let me paint the scenario – 02.00 am Drone on night surveillance mission and the pilot controlling from the ground control station is looking for say infiltrators along the border. He is focussed on the monitor screen straining his eyes and using the thermal imaging to look for human activity indication in mountain terrain or pinpoint foot marks in the snow or the rugged hilly terrain. Now, how can digital disruption enhance the capability of the mission .It is pertinent to mention here that the Artificial Intelligence all by itself cannot solve problem but the Artificial Intelligence tool can be bought up to speed by self learning which we call as machine learning. The Artificial Intelligence platform can quickly consume huge amount of data and quickly join the dots to predict based on huge data bank and assist the operator in quicker identification of suspect target.
Taking a step further the platform can be trained to identify whether the target is carrying a weapon and also can plot his movements, following footmarks on the snow/attempts to obliterate it. Another use case can be to detect/identify any activity near the launching pad or any type of silhouette changes or fresh tracks. Similarly any freshly broken branches in thickly wooded area can indicate intruder activity.
Let me now come to the crux – the drone is going to provide video streaming from a standoff distance to the control station. The Artificial Intelligence program will function by using the videos streaming from the drone and applying a series of algorithms . Algorithms are basically mathematical procedures that are like a flow chart to compare the object seen with thousands of stored reference images of target in different postures/angles/movements/positions. So reference images giving height /shape or action whether it has a head /hands/legs is compared. On the lighter side a mountain goat seen by the camera will immediately compare with the data bank and tell it to be not a human . Here using the model of human brain , machine learning teaches Artificial Intelligence learn new things, recognize patterns and make decisions and raise alert in near real time for the security forces to take proactive action to neutralize the threat.
The value proposition here is the ability of the Artificial Intelligence tool to consume vast amount of data(big data) in short time and drawing conclusions based on certain rule sets followed .In typical existing scenario -assuming that the operator has identified (?) the activity and now he will rewind/pause/zoom many times to be sure. But the Artificial Intelligence tool in near real time can raise a alert (even if a false positive).
Human brain and Artificial Intelligence
Such Artificial Neural Network platform is designed to consume vast quantity of structured and unstructured data and then analyse it for desired results.
The video analytics platform will have the capability to identify the human form/tank/bunker or any selected object . When such a object say in this case intruder violates a preset rule[say any human movement in marked area] then an alert is generated .And based on the alert say a red box is generated on the suspicious target and it follows it and thereby his direction of movement can also be predicted.
It is characteristic of such program that it is self learning So the programmer needs to set rules for all things that the operator wishes to be alerted. The vast experience of the human operator can further enhance the learning of the recognition tool. the images recognized and the silhouette are classified. Just imagine the impact of such a Artificial Intelligence platform integrated with Machine learning in operation for say couple of months. The tool will self learn and become a credible support system for the forces against infiltration . Every such sighting will improve the tool and the results will tilt in favour of the security forces .
Detect > Capture description > Match with the data bank
Just like humans, machines will continue to improve as they learn … by doing.
“The biggest advantage to machine learning is that it allows us to do things much more quickly than we’d be able to do otherwise,” said Singh.
Artificial neural networks (ANN) use algorithms to learn things, recognize patterns and make decisions. They mimic the way the human brain solves a problem — by taking in inputs, processing them and generating an output.
Learning algorithms come in three subsets: supervised, semi-supervised and unsupervised learning. The first two require large sets of training data to guide the desired results.
Author : Lt Col Avadhut Deshpande
Senior Analyst Persistent India,Pune