Drone and UAV traffic management system to address detection of drones and UAVs, unlawful trespassing by drones, collision avoidance, route monitoring and much more.
Won the cash prize of ₹10,00,000 and received an invitation to attend the Global Mobility Summit 2018 at Vigyan Bhawan, Delhi by NITI AAYOG to meet major CEOs across automobiles, aviation, mobility organisations and receive the award by Prime Minister of India, Narendra Modi.
NITI Aayog launched MoveHack a global mobility hackathon to crowdsource solutions aimed at the future of mobility in India.
Envisaged to be one of the largest hackathons globally, Move Hack was focused on 10 themes and structured over three legs: idea phase, online screening, and the finals at Vigyan Bhawan in New Delhi.
Dronacharya was submitted as a solution to cater the needs of UAV Management from the air traffic control to detection and reporting of illegal flights.
Solution involved practices like Machine Learning to tackle the issue of detection, MQTT to stream the real time data for accurate geolocation of a given drone, Blockchain network to create an approval pipeline with publically readable scope for anyone with a smartphone can check for geofence violations, And many more
Using a lightweight protocol (MQTT) for transmission and processing of data to achieve realtime-ness and reliability.
Detecting Drones Using Fast R-CNN of resnet50 model trained with publically available images.
Using optical signalling for identification of a given drone.
Smart-contract based Flight plan approval subject to risk factors (evaluated programmatically).
Multiple entities (DGCA, city/state governments) will digitally sign the approval and publish on a permissioned blockchain. (indium.network)
Shows all drones and geofences on satellite map, In case of violation, alerts both the operator and authorities.
MoveHack has been one of the best hackathons that I have participated in, the opportunity to meet a lot of talented and experienced people from different domains (Blockchain, Smart Contracts, etc) was simply amazing. My partner in action (Nilesh Trivedi) without whom our success wouldn't be possible helped me throughout the event in tackling and understanding the concepts of the blockchain (He runs his blockchain network called, indium.network) while I worked on the data-science implementations.
My work in Dronacharya primarily revolved around building Detection (ML) system, Monitoring (dashboard) and writing the smart contract for flight plan approval.
Detection: The task of detecting drones (computer vision problem) was done by utilizing the Fast R-CNN model of resnet50 by training publically available drone images and then doing hyperparameter tuning to get accurate results. Live demo was deployed using Flask API and a webpage to showcase the live detection of drones.
Monitoring: Using D3.js and Mapbox API, a dashboard was built to demonstrate how a drone can be monitored through a web interface and various alerts, emergency operations can be executed on the fly.
Smart Contract: Wrote a smart contract in Solidity (An object-oriented, high-level language for implementing smart contracts) where multiple entities could approve a flight-plan. This multi-signature approval will be published on a blockchain so that it becomes publicly readable.
Overall I'd consider the entire experience to be worthwhile where I could meet with people experts in their given domains of the mobility sector. Top 10 participants were invited to the final event of Global Mobility Summit where we got the chance to meet with Prime Minister, Narendra Modi and various top executives around leading companies in the mobility sector like Uber, Cisco, Ola, etc.