Home / Projects
I have developed so many projects in career. This page is all about my projects. Here is the small description of those projects.
My research focuses on developing an AI-enabled camera-based monitoring system for rivers that integrates deep learning segmentation, machine learning stage prediction, discharge estimation, and edge computing for real-time deployment. The system was tested at two sites in northern Utah (Logan River and Blacksmith Fork River), where state-of-the-art models such as SAM, MobileSAM, and SegFormer were evaluated. Results showed that ensemble machine learning models could accurately predict stage height and discharge, while MobileSAM on Jetson Orin Nano provided the best balance of accuracy and efficiency for edge devices. This work demonstrates the potential of low-cost, scalable, non-contact camera systems to complement traditional hydrologic monitoring and improve water resource management in data-sparse regions.
BRACU Mongol Tori Revolution is a multidisciplinary team formed by a group of enthusiastic students of BRAC University. The rover is a fully functional, stand-alone mobile platform which is capable of acting as an astronaut assistant in various servicing and scientific research task. This year marks the third major update in the Mongol Tori lineup with numerous minor changes along the way. The major updates in the current version of the rover includes rover expansion system, switching from six wheeler to four wheeler design, reconstruction of the entire arm, portable command station and significant reduction in size and weight.
This is a Can size satellite system. By developing this system we can gather knowledge about big satellite. We have developed this Can Satellite by our own mechanism. There were a stack of different circuit board layers for different purpose such as- Communication, Control, Sensor, Power etc. There were different types of Sensors for taking on board data from the atmosphere. A communication module was attached to this so that we can communicate with this satellite remotely through our own build software. For communicating we had built our own antenna & software. As for first time we have lached it from a highraise building but in future we are planning to launch it by a custom made small rocket.
Our prototype is a Briefcase shaped device which is a portable laboratory consists of drill for collecting soil sample, water purifier (Vapor-compression desalination) and various sensors. There is also a display with an on-board computer to show & analyze those data. Our device consists of various Gas detectors like CO, CO2, nitrogen, Argon and Air humidity. Others includes Air temperature, Air pressure sensors. For further analysis of the environment it has Radiation sensor, ambient light sensor. For soil evaluation & detect bio-signature it has Soil moisture +soil temperature probe. Our device also has an emergency situation alarm integrated with environment sensors and emergency oxygen backup.
This research focuses on autonomous traversal of land vehicles through exploring undiscovered tracks and overcoming environmental barriers. Most of the existing systems can only operate and traverse in a distinctive mapped model especially in a known area. However, the proposed system which is trained by Deep Reinforcement Learning can learn by itself to operate autonomously in extreme conditions. The dynamic double deep Q-learning (DDQN) model enables the proposed system not to be confined only to known environments. The ambient environmental obstacles are identified through Faster R-CNN for smooth movement of the autonomous vehicle. The exploration and exploitation strategies of DDQN enables the autonomous agent to learn proper decisions for various dynamic environments and tracks. The proposed model is tested in a gaming environment. It shows the overall effectiveness in traversing of autonomous land vehicles in comparison to the existing models. The goal is to integrate Deep Reinforcement learning and Faster R-CNN to make the system effective to traverse through undiscovered paths by detecting obstacles.
There is a rapid development in technology that is influencing the standard of living immensely in several aspects. In our Country, it is not possible to supply LPG (Liquid Petroleum Gas) through pipes to each and every home as production of LPG is quite less. To make LPG usage easier we have come up with a system of monitoring and booking LPG cylinders in advance through GSM. This will present an effortless and convenient option for people in rural areas and people with busy schedule in cities. One of the problems LPG consumers face is that they do not know the gas level or status of the cylinders which create even more confusion and difficulty for them. However, now with our project using GSM system, customers only need a few taps on their phones to select their options to buy new gas cylinders. This system continuously measures the weight of the cylinder and once it reaches minimum threshold it will automatically send message to the authorized LPG agent so that they can deliver the LPG cylinder in time. Another major problem consumers may face is the unwanted accidents through gas leakage from the cylinders. This project proposes a system that will monitor and detect any gas leakage and alert the consumer about the leakage to avoid such major accidents.
It is well known that level crossing gates are operated manually with the aid of a gate keeper. This human intervention can be prevented by means of automating the gates of railway level crossing. In situations where the train is late for some reason, the gates continue to be closed for longer periods causing dense traffic near the gates. This can be prevented as well with the usage of automation. The proposed device uses infra crimson sensors to come across the arrival and departure of trains on the railway crossing and Arduino to manipulate the opening or closing of gates. The device also uses two IR sensors to locate the advent of the train and a third IR sensor to discover the departure of the train. At the same time, when the appearance of the train is sensed, signs or warnings are provided to the passers-by indicating the arrival of the train at the track. While the second sensor detects the train, the sign turns crimson and the motor operates to shut the gate. The gate remains closed until the train absolutely moves away from the gate. As the departure of the train is detected by the usage of the IR sensor, the visitors sign turns green and the motor operates to open the gate. Therefore, automation of rail gate is a safe step towards proper transportation system.
We are providing a complete solution to enrich the existing lunar data in NASA’s open data portal.Around 3-4 billion years ago huge explosions by the asteroids helped to exchange a lot of materials of outer surface between our home planet and the moon.As the atmosphere of the moon is a vacuum and there is no active biology the rocks of the early stage of the earth which were bought by the explosion are almost unchanged.So if we are able to collect those rocks we can know what was earth like in 3-4 billion years ago.But even after countless lunar missions we are unable to find enough of these rocks.So after conducting a research on past lunar missions we came up with a unique solution.Our proposed solution have both software and hardware prototypes.For autonomous traversal in the lunar soil we are using customized machine learning and computer vision algorithms to analyze the lunar environment and find the proper place to conduct further experiments.We are using thousands of lunar landscape images of NASA’s lunar data repository from NASA open data portal.After reaching the designated place it will scan the surrounding rocks and soils.The rock and geological archive will help to determine whether these soils are present in the latest data repository and if it further analysis needed with the help of custom AI algorithms.It the match each and every rock with the existing database in real time.If it can find what it is looking for the lunar rover will go to the next step in the analysis process the “Science experiment”.For this we are building the second version of our science unit prototype .The first version is already tested in the MDRS(Mars Desert Research Station) successfully.This unit is consist of almost seven differnt soil experiment units : Nitrogen ,Phosphorus ,Potassium ,Amino acid, Biomass ,Water capillary .We have also have separate unit consist of all kinds of gas sensor to detect air samples.If the result is positive the onboard system will automatically collect necessary amount of needed samples and store it in a vacuum chamber to bring it back to the earth.This will solve the drawbacks of existing sample collection process and will save time and cost.We need to understand our neighbour first to reveal many mysteries not just outside our home planet but also of our planet earth.
In this project, we are trying to develop a system to monitor surface water quality data using a small size autonomous hovering boat. In situations where it is difficult to gather, analyze and index water quality data age old systems are used which often requires more manpower and time. This can be prevented as well with the usage of automation. The proposed device uses a miniature boat shaped hovering platform for mobility in water which is guided by compass based autonomous system located on the boat. The device consists of several water quality sensors for onboard data collection. At the same time, two small pumps are being used to collect water samples for laboratory tests and further analysis. Realtime data will be provided to the user through cell phone texts and uploaded on cloud server. The system is integrated in a communication system to autonomously store data on real time servers through internet connectivity. Being lightweight and rechargeable the portable device can be taken to remote places conveniently through GPS. Therefore, autonomous GPS guided water monitoring system can have a good impact in monitoring.
The device will transform emergency respiratory support through its small size, portability, ease of use, versatility and extended battery life. By costing as little as other ventilators and offering noninvasive capabilities, the device is ideally suited, no matter what their location or severity. Our Device is mainly a VCV (Volume Controlled Ventilation) support system. By using it we can control Tidal Volume, Respiratory Rate, I:E Ratio as well as PEEP & FiO2 monitoring. It can be operated in both DC & AC power supply. Gas or electricity supply failure alarms are also included in the system. Infection control is attached by using disposable internal breathing circuits. An Android App system has also been integrated for remote operating systems.
A web based product using machine learning models to predict future stock values based on the historical values.
We are living in the era of smart technology. Life has never been as easy as today ever in the history of mankind. This is a small step to make our daily life more easy. We have been thinking of developing a system which will lead us to automate our room through controlling electronic devices(example: fan, light, computer, etc.) by using voice command and automated synchronization with temperature and light.It will also alert us if it detects smoke in our room through SMS.
This a 2D animation by RS43 batch of Brac University Residential Semester. For making this animation we have used Flash, Adobe After Effects, Adobe Premier Pro softwares. By this animation we have tried to show our cultures which are destroting day by day.
We tried to implement high temperature detecting alarm! While preparing this we used Arduino Uno, PIEZO BUZZER, Temperature Sensor,LCD Display, LED, Potentiometer and Resistors.
We have developed a IP Address Calculator using Assembly Language.