Completed Projects
Note:
* Any additional & elaborate technical details about the projects tagged as Confidential are restricted due to security and proprietary conditions.
Government of India, Department of Space,
Indian Space Research Organization (ISRO)
- First time in Indian history, a 15 years old student, was invited to work on a confidential project for ISRO.
- Provided with a time period of 2 years to complete the assignment. However, it was completed, within 35 days.
- Post-completion, the project was presented to ISRO council members & submitted to Govt. of India, through ISRO at my 15 years of age.
Government of India, Ministry of Defence,
Defence Research & Development Organization (DRDO)
- Entrusted with a Confidential Project for Indian Defence, through DRDO & completed it.
Visalini's Solution for Demonetization Against Corruption (ViSDAC)
- Suggested a possible solution for implementation during demonetization to make the process smoother, named ViSDAC.
Solution for Gandcrab Ransomware v5.0 (SGRv5.0)
- Implemented a working solution for Gandcrab Ransomware V5.0, and successfully recovered the data from an affected PC, within 70 hours without paying any ransom and without using any data recovery tools.
- Received an official call from Microsoft Corporation, Seattle, USA, to provide the solution to them.
SOS Equipment for Women
- Plenty of mobile apps and wearables available in the market today, work in conjunction with smartphones. However, majority of the crimes against women happen in forbidden locations, without cellular connectivity. Often, the victims are snatched off their mobile phoned. This nullifies the presence of SOS apps dependent on mobile phones.
- My project's uniqueness lies in its ability to automatically sense the emergency, even in forbidden locations and alert local Police stations and family. Removing the need to manually generate alerts, independent of cellular connectivity.
Wild-animal intrusion detection, recognition & notification using Computer Vision
- Wild animals intruding into civilian regions foraging food and water, happen frequently these days. They damage agricultural lands and pose perils to human & domestic lives. Nevertheless, not all animals that intrude pose a severe threat. Consequently, simple motion detection becomes inadequate.
- My computer vision-based AI solution detects movement, recognizes the animal and cautions the neighbouring people and forest rangers, only if that particular animal trespassing into the zone is precariously dangerous. It simply requires a small camera positioned on a tree. The deep learning-based model was capable to produce about 85% accurate results even during dark nights.
Sensor Placement Algorithm with Range Constraints for Precision Farming (SPARC)
- Precision agriculture, is a modern farming practice, in which fusing low resolution Remote Sensing data with proximal sensors provided by Wireless Sensor Networks (WSN) can help increase productivity, reduce cost, and optimize resources. However, sensor cost and their range emerge as a major bottleneck. Typically optimal sensor placement is a combinatorial problem which is computationally intensive (NP10 hard) and range constraints add another complex dimension to the problem. This article proposes the SPARC algorithm for optimally placing the sensors considering their range.
- In our two-step approach, first step solves the optimal sensor placement problem by using the optimization models. In second step, the sensor locations are adjusted to address the range constraints such that the error covariance is reduced. As a result, sensor cost, physical constraints of farming practices, and maintenance cost can be reduced drastically without compromising on the monitoring capability.
- Collaborators:
- Dr. Seshadhri Srinivasan, Berkeley Education Alliance for Research in Singapore (BEARS Lab, Singapore)
- Dr. Korkut Bekiroglu, Assistant Professor, Department of Electrical Engineering Technology, SUNY Polytechnic Institute, New York, US
- Dr. Giovanni Palmieri, GRACE Research Group, University of Sannio, Benevento, Italy
On-going Projects
Assistive Technology for Visually-Challenged
- Currently, visually impaired persons use Braille script as the sole medium for learning & communication throughout their lives. Producing hand-written Braille script is tiresome and affects the visually-challenged person physically.
- While, Braille is an inflexible generalized methodology; The levels of Visual impairment vary from person-to-person. For a person, with partial blindness, Braille deprives them from all realizations of their available visual capabilities.
- My novel AI-powered solution has the potential to replace Braille throughout the world. It is dynamic and totally customizable to each person according to their sight-capabilities.
Assistive Technology for Mental Disorders
- AI-assisted therapy for people with neuro-cognitive disorders like, Intellectual Disability, Alzheimer’s disease, ADHD, Autism & etc.
- Harnessing advanced technologies like Virtual Reality and ageless methods like meditation together and understanding various facets that they provide to the ailments of serious disorders especially, PTSD.
- Taking advantage of AI, to assist doctors and special educators in monitoring & predicting the cognition & decision-making processes of the intellectual disabled persons. It can help in enhancing their physical, emotional and social well-being.
- AI that helps to handle people of any age group with various levels of Intellectual Disabilities and can assist the doctors and special educators in handling the person. It provides much better level of analysis and therapeutic conditions for behaviour, sensory conditioning and self-care.
Vision based Navigation using Deep Reinforcement Learning
- Hitherto, Autonomous vehicles have performed well in controlled environments, but the same is not observed in natural real-time environments.
- My key contributions so far:
- Propose reinforcement learning (Q-learning) based approach for an autonomous vehicle to navigate in an unknown real-time environment. Allow the agent to update the reward function spontaneously by comparing each episode of learning to drive.
- This is a process meticulous as the agent learns new moves against the existing reward functions.
SOS Equipment for Fishermen
- The Indian Peninsular fishermen face the uncertain wild face of the oceans during storms and are unable to reach out for help due to the unavailability of cellular network inside the sea.
- This project allows to get alerted about the upcoming calamity, to reach out coast guards and nearby ships for help & rescuers to track their live location.