Academic Work

Indoor Wi-Fi Localization: (Nov’21 - Apr'22)

As part of my B.Tech. thesis project at IIT Madras, I collaborated with Sindhu Gunturi under the guidance of Prof. Ayon Chakraborty, in the Sensing and Networked Systems Engineering (SENSE) Group.

We focused on developing calibration-free techniques for indoor localization using Wi-Fi signals. Our work involved designing and executing a series of experiments to explore how Channel State Information (CSI) varies in indoor environments, influenced by the position and movement of Wi-Fi nodes and surrounding objects.

I led the data collection process, logging CSI across different configurations and environmental conditions to generate datasets suitable for training machine learning models. This project deepened my understanding of wireless sensing, signal processing, and the practical applications of CSI in real-world indoor positioning systems.

PathExplorer Robot:           (Sep’21 - Nov'21)

This is a project which we did as a part of course work (Smart Sensing for IOT), instructed by Prof. Ayon Chakraborty

We implemented a path follower robot, in which a user draws an arbitrary path as a hand drawn input, and the car moves in the path as given by user. The car also checks for obstacles in its path and if it is encounters an obstacle, it stops and then it also notifies the APP in which path it encountered the obstacle. So in this way we can visualize the status of the journey the car had made.

Apart from the above functionality, we can manually takeover the control of the car at any given point of time.

Members of Project: Nithin Uppalapati, Srinivas Mareddi, Bharath Simha

1. All videos related to project can be found here

2. A detailed report of this project

3. Link to GitHub Repo

4. Electrical Schematic

CFI (IITM) - Project Radiscent:     (Sep’20-Aug'21)

A Robust Solution to improve the efficiency in lifestyle by modernising the electric switch, this is a students project.

We developed a retrofittable device that could be mounted over existing switchboards, enabling automated control through voice commands, a mobile application, or a dedicated remote, eliminating the need for rewiring or infrastructure changes.

Our interdisciplinary team consisted of 7 students from various departments. I was primarily involved in circuit design, embedded systems development, and prototyping.

The project reached the design and prototyping phase, laying the groundwork for a robust and user-friendly smart home solution targeting seamless automation in Indian households.

RADAR Demonstration:       (Oct’19)

We built a functional 2D RADAR system. The objective was to showcase real-time object detection and tracking using a Plan Position Indicator (PPI) display, as part of our poster presentation.

We developed a working prototype by programming an Arduino UNO and integrating it with a Sharp IR sensor mounted on a rotating servo motor. The setup performed periodic sweeps of the environment, detecting objects within a range of 4.0 cm to 30.0 cm, and displayed their angular positions in a 2D radar-style visualization.

This mini-project demonstrated our ability to bridge hardware and signal processing concepts in a hands-on, embedded systems context.

1. Link to Abstract for this Project

2. Link to Poster

3. Link to GitHub Repo

Solving Travelling Salesmen Problem:             (Oct’21)

• As part of my exploration into heuristic search algorithms, I studied the Travelling Salesman Problem (TSP)—a classic challenge in combinatorial optimization. The objective was to determine the shortest possible route that visits a set of cities exactly once and returns to the starting point, with cities represented as 2D coordinates.

• To facilitate experimentation, I developed a Python-based tool that allows users to visually place ‘n’ cities on a window, dynamically generating custom datasets for algorithm testing. I then implemented and analyzed the performance of multiple heuristic algorithms for solving the TSP:

• This study deepened my understanding of search-based AI techniques and their effectiveness in solving real-world, NP-hard problems like TSP.

• Simulated Annealing (SA)

• Genetic Algorithm (GA)

• List-Based Simulated Annealing (LBSA)

1. Link to Problem Statement

2. Link to Report

3. Link to GitHub Repo

Building Search Engine and Evaluation:                 (Feb’21- May'21)

As part of an in-depth study in Natural Language Processing during the Spring 2021 semester at IIT Madras, I collaborated with a teammate to design and build a functional Information Retrieval (IR) system from the ground up in Python.

The project focused on understanding the core principles behind search engine design and evaluation. We implemented a vector space model for document retrieval and tested its performance using the standard Cranfield Dataset. To enhance retrieval accuracy and relevance, we explored and integrated:

• Latent Semantic Analysis (LSA) for dimensionality reduction and improved semantic understanding

• Query Expansion, allowing user-controlled reformulation for broader or more precise search results

• Spell Check integration to handle noisy or misspelled queries

We compared the performance of our baseline vector space model with the LSA-enhanced version to analyze improvements in retrieval quality.

1. Link to Report

2. Link to GitHub Repo