Design and develop a drone-based aircraft scanning system. The National Security Guard has very limited surveillance capability in counter hijack operations. In case the window shutters have been pulled down by the hijacker, there is no way to detect & identify the location of hostages / hijackers. There is accordingly a need to develop drone base aircraft scanners. There is a need to develop essential capability to see the situation inside the hijacked aircraft.
The requisite system should have the following features:-
- It should be a drone-based solution
- It should have the capability to perform an X-Ray scan of the hijacked aircraft by flying over it
- It should have the capability to record data for post-operation analysis
- It should have the capability to transmit live video feed during scanning of the aircraft
- It should have low aural signature so as to be able to operate undetected by the hijacker
- It should have the battery backup with a minimum of 60 minutes or longer endurance
- Team name - MindHawks
- Led my team to a 2nd place finish at the National Level.
- As the head of the Machine Learning & Software team, my role in the object detection software I built, team management, and external communication were pivotal in securing our victory.
- Played a crucial role in developing a drone-based solution named प्रतिष्क: to aid NSG in counter-terrorist tasks efficiently.
- Framework/Model: YOLOv5 (Windows 10), SSD MobileNet(Raspberry Pi 4b)
- Detection Classes: Orange_Gun, Black_Gun, COCO Dataset Classes
Choose from various detection categories
- Chassis: Tarot T810
- Flight Controller: Pixhawk Orange Cube
- Motor: T Motor MN4014
- ESC: T Motor Air 40A
- Propeller: Eolo 15*55
- Transmitter and Receiver:
- Battery: 12600mAh 6S Lithium Ion
- Camera Module: SJCAM SJ4000
- Gimbal: 3-Axis Brushless Gimbal(RTF)
- Video Transmitter: TS835 FPV 5.8G 600MW 48CH
- Video Receiver: 5.8G UVC OTG Android
Final UAV model named प्रतिष्क:
In December 2021, we submitted our concept paper for the above-mentioned problem statement to NSG. Our concept paper included innovative ideas for the solution, focusing on enhancing surveillance capabilities inside hijacked aircraft.
- Innovative Object Detection: We proposed real-time object detection using YOLOv5 and TFLite models, a feature not initially mentioned in the problem statement. This capability would enable the system to detect and locate potential threats inside the aircraft.
- Explored advanced technologies such as Wi-Vi System and explored the use of X-ray and IR cameras for through-wall scanning.
We had a discussion of our solution with NSG commandos, and they appreciated our innovative approach, especially the object detection aspect.
- Emphasized the object detection module for enhanced security.
- Explored solutions for low aural signature and extended battery endurance.
- Discussed potential technology options for real-time surveillance.
During this phase, we focused on building our first prototype of the drone and collecting data.
- We continued to develop and refine the object detection system for improved accuracy, further enhancing our solution.
- Demonstrated our prototype's capabilities, including real-time object detection on day to day items.
In this phase, we conducted extensive research to improve and apply technology for real-time object detection.
- Developed the starting phase of our final prototype
- Successfully tested custom YOLOv5 and TFLite models, tailored to identify potential threats inside the aircraft with high accuracy.
The last review at Manesar provided valuable insights. We focused on object detection, drone improvements, and data quality.
- Improved upon previous prototype while keeping in mind the review points
- Provided the NSG Team with a comprehensive booklet explaining the model and its operation, enhancing our presentation.
We were awarded 2nd place in the competition during the final felicitation ceremony, highlighting the success of our object detection-focused approach.
- Special thanks to Team MindHawks for their dedicated efforts and contributions.
- Team Members:
- Siddhi Marvaniya (siddhimarvaniya@gmail.com)
- Niraj Lalwani (niraj311202@gmail.com)
- Anuj Shukla (anujmshukla2002@gmail.com)
- Smit Vidja (smitvidja2604@gmail.com)
For inquiries, collaborations, or to learn more about our projects, please don't hesitate to contact us. We look forward to connecting with you!
I'd love to hear from you and answer any questions you may have about the applications or any other projects. Feel free to reach out to me:
