Hi
I am Pranava Swaroopa
Photo of Pranava Swaroopa
I develop robots that can see, think, and act

About me

M.S. in Automotive Engineering from Clemson Univerisity International Center for Automotive Research

Education

2021 - 2023
Master of Science in Automotive Engineering
Clemson Univerisity - ICAR
2013 - 2017
Bachelors of Technology in Mechanical Engineering
Manipal Institute of Technology, India

Technical Skills

Languages C/C++, Python, JavaScript, MATLAB, Bash
Tools Git, ROS, Docker, Visual Studio
Simulation Env Gazebo, LG-SVL, CoppeliaSim, Project Chrono
Design Software SolidWorks, Siemens NX, AutoCAD, CATIA V5
Operating System Ubuntu, Windows, Raspberry Pi

Experience

2024 Apr - 2024 Sep

Robotics Field Engineer

  • Conducted on-site troubleshooting and diagnosis of network and system issues and coordinated deployments with internal teams.
  • Provided training and technical support to end-users, in operating and maintaining deployed systems.

2023 Jan - 2024 Sep

Research Intern - Clemson University

  • Collaboratively developed an off-road dataset that identifies area of traversability in high-speed autonomous vehicle setting sponsored by VIPR-GS, US Army
  • Led development of robust annotating software using C++, Win32 API and OpenCV and reduced annotation time by 97%
  • Deployed ROS tool-chain to capture high-resolution videos and used Python and FFmpeg to create a comprehensive traversable dataset for off-road terrains.
  • Implemented a video input neural network, enabling precise identification and navigation of unexplored and challenging off-road terrains, such as dense forests and uncharted terrains.

2021 Sep - 2022 Dec

Research Assistant - Clemson University

  • Experienced in ROS deployment for robotic platforms like Clearpath Husky, Jackal, and Turtlebot3 for testing and validation.
  • Demonstrated application of control techniques in simulation environments such as Gazebo, CoppeliaSim, and Project Chrono to develop and validate on wheeled mobile robotic platforms.
  • Conducted HIL and SIL testing for navigation algorithms in mobile robots for robustness and reliability in real-world scenarios.

2020 Jul - 2021 Jun

Research Assistant - Kasturba Medical College

  • Implemented deep learning models to extract patterns and established data analysis protocols.
  • Developed software in MATLAB to analyze data points and visualize the results using matplotlib, decreasing inference and turnaround time by 65%.
  • Established a protocol with bash script and MATLAB to ingest, analyze and extract videos for research.

2017 Aug - 2019 Jun

Assistant Manager - Essar Projects India Ltd.

  • Led a team of 5 engineers as an Engineer consultant for a $6 million project.
  • Orchestrated and directed cross-functional teams comprising design professionals, contractors, and site personnel.

Projects

Off Road Autonomy

Drivable Area Detection under Severe Weather

Autonomous Turtlebot

A* Route Planner

Linux System Monitor

Snake Game

Wed Development

Off Road Autonomy

Highlights

  • PI: Dr Adam Hoover, Holcombe Department of Electrical and Computer Engineering, Clemson Univeristy
  • Video based neural network
  • Dataset generation from Clemson Experimental Forest
  • Annotation and developing CNN pipeline

Tools: Python, C++, OpenCV, TensorFlow, HPC

Drivable Area Detection in Severe Weather

Highlights

  • Team of 3
  • 20 Layer DNN Network
  • Encoder-Decorder configuration
  • 15.6K selected images from BDD100K dataset
  • Pipeline developed that can process a given video and recognize and draw drivable area in it
  • Built with Palmetto HPC - Clemson Univerisity
  • Acheived mIOU = 44.15%

Tools: Python, OpenCV, TensorFlow, HPC, Jupyer Notebook

We wanted to develop a neural network that helps vehicles find traversable/ drivable path where the weather is not ideal. Initial thought was can we develop something that is simple in implementation but shows good accuracy. This was done as part of course credit and which meant we had only a month of time to select dataset, develop a network and tune it. We developed a simple Encoder-Decorder network and selected BDD100K as our dataset. To concentrate our training we selected images tagged with weather conditions like rainy, foggy, snowy, and cloudy. We acheived a mIOU = 44.15%.

Autonomous Turtlebot

Highlights

  • Lead the Team of 4
  • Developed PID controllers to use LIDAR data to keep the robot away from all obstacles and be at the center
  • Using camera images, developed a PID controller to follow a line on the track and used miniYOLO to analyze and respond to a STOP sign.
  • Implemented a cron job to get useful data sent before every test, which resulted in a faster testing schedule.
  • For successfully implementing the project, received the highest grade in the class

Tools: Python, ROS, OpenCV, miniYOLO, Raspberry Pi, Ubuntu

A* Route Planner

Highlights

  • Project built in Ubuntu Linux
  • Uses maps from OpenStreetMap library
  • Implemented core logic using OOP
  • Part of C++ Nanodegree from Udacity

Tools: C++, CMake, OpenStreetMap, Ubuntu

Agrawal Lab Website Project

Highlights

  • Website built and hosted with Wix
  • Made in mind with ease of maintenance
  • Visit at agrawallab.com

Tools: WIX, HTML, CSS, JS

This is the public facing lab website I created when I was working for Agrawal Lab as a Research Assistant. Agrawal Lab is a neuroscience lab focused on Drosophila Melanogaster. I also developed websites for internal use in Microsoft Sharepoint.

Linux System Monitor

Highlights

Tools: C++, CMake, Ubuntu

Snake Game (Improvised)

Highlights

Tools: C++, CMake, Ubuntu

Hello Visiors!