I have always been inclined towards research, since that fuels my curiosity towards new thing and exploring some new areas of science.
My B.Tech Major Project : Design and Development of Controller Board for Automation of Home Cooler using Microcontroller
Abstract: Home automation has always been a key region of interest for the researchers. Many home appliances have already been automated. In this paper, the authors propose a model for controller board to automate the home air cooler for implementing the control algorithm using the microcontroller. The surrounding temperature is the input to the algorithm while the speed of the motor and state of the pump is the output. The developed control board has been implemented in a home air cooler to test the working of the developed system and control algorithm. The outputs have been compared with the expected results and it has been found that the developed system works quit well.
My Masters Thesis : An Intelligent Model for Indian Soil Classification using various Machine Learning Techniques
Abstract: Soil classification is useful in site investigation process as it helps to assess the general suitability of the site for acquire the physical and mechanical properties of soil for adequate and economic design and helps to determine the suitability of materials for construction. In this work we developed and tested various machine learning models for soil classification using the classifier learner app available in MATLAB. The soil samples in the form of soil sample image were collected and divided into seven classes’ i.e. clay, peat and sand being the main classes and Clayey Peat, Clayey Sand, Humus Clay, Sandy Clay and Silty Sand being the mixture classes. The machine learning models were tested on the obtained classed to select the best model on the basis of accuracy comparison, after which a coded model was developed with the aim of creating a model for easy on-site easy soil classification. The advantage of soil classification is evident from the fact that if an engineer attempts to save the cost with a low budge investigation, then it may cause additional expenditure later if previously undiscoverable unfavorable ground conditions are encountered later. As a first step for developing this model, images of soil were acquired then preprocessed using 2D-DWT to decompose the acquired images. This is followed extracting relevant features and selecting the strongest feature. Finally the selected features are given to ANN, k-NN and SVM classifier for classification of soil image data. The experimental results indicate decent classification performance of the proposed technique for soil image data. After selecting the best classifier, a Graphical User Interface (GUI) was developed in MATLAB for easy and real time classification of the soil images.
JRF Research Work : Detection and Extinguishing systems for Forest Fire using Sensor Networks and Ground robots
Abstract : Forest fires are a menace to the environment and disturb the ecological balance in the environment. There is a need to detect forest fire in early stages, so as to take appropriate preventive measures. Wireless Sensor Networks have proved to be useful for the early detection of Forest fires. There are many methods proposed for early detection of forest fires, from which monitoring the environmental conditions and gasses have proven to be most useful.In this paper, we discuss the early developments of one such wireless sensor network systems for early and efficient detection of forest fire.
Chandan Taluja 