Nilay Varma Profile Picture

Nilay Varma

Robotics & Autonomous Systems Researcher

📧 nilayvarma.india@gmail.com 📱 +1 (217) 904-2232 💼 LinkedIn 📍 Houston, TX

About Me

I am a passionate researcher in Robotics and Autonomous Systems with a strong foundation in Machine Learning, Computer Vision, and Mechanical Engineering. Having recently completed my Master's degree at Boston University, I have extensive experience in developing innovative solutions that bridge the gap between theoretical research and practical applications. My research interests focus on advancing autonomous systems through sophisticated perception algorithms, biomimetic robotics, and intelligent control systems.

7+
Academic Projects
1
Research Assistantship
3
Professional Internships
3.20
Graduate GPA

Research Interests

🤖 Soft Robotics

Biomimetic soft robotic systems, pneumatic actuation, and adaptive materials

🧠 Computer Vision

Neural radiance fields, 3D reconstruction, and visual perception for robotics

🎯 Machine Learning

Deep learning for robotic perception, reinforcement learning, and neural networks

🏥 Medical Robotics

Biomedical optics, prosthetic design, and autonomous medical systems

🚗 Autonomous Systems

Path planning, SLAM, sensor fusion, and autonomous navigation

🔬 Human-Robot Interaction

Natural language processing for robotics and intuitive control interfaces

Research Projects

Enhancing Block-NeRF with Additional Losses
Sep 2024 – Dec 2024
Developed novel loss functions to enhance Block-NeRF scalability for large-scale 3D scene reconstruction. Implemented coarse and fine RGB loss along with coarse and fine transmittance loss using advanced InterPosEmbedding techniques. The enhanced architecture addresses challenges in transient object artifacts, blurred reconstructions, and inconsistent lighting across scene blocks. Achieved 30% improvement in depth consistency and visual fidelity with significant reduction in processing time for large-scale applications.
NeRF PyTorch 3D Reconstruction Computer Vision Neural Networks
Judo Technique Recognition for Enhanced Training
Jan 2024 – May 2024
Developed a comprehensive deep learning system for recognizing Judo techniques to help competitors analyze opponents' strategies. Implemented and compared five different model architectures: GRU with ResNet18 feature extraction, (2+1)D-CNN for spatiotemporal analysis, CNN-LSTM hybrid, pose detection LSTM using MMPose RTMO, and pose tracking LSTM with Kalman filtering. The system processes video clips and classifies three common techniques (Uchi mata, Seoi Nage, Osoto Gari) with the pose detection model achieving 50% accuracy on a dataset of 2,029 pre-labeled video clips.
Deep Learning CNN LSTM Pose Detection Action Recognition MMPose
Data Reduction of LiDAR Depth Maps for Autonomous Vehicles
Sep 2023 – Dec 2023
Implemented and compared three dimensionality reduction techniques (PCA, ICA, autoencoders) on the KITTI dataset to enable autonomous vehicle applications on resource-constrained hardware. Developed advanced neural network frameworks for autoencoders and overlaid LiDAR data onto compressed images for enhanced depth perception. The ICA method achieved the best performance with 0.07% compression ratio while maintaining object recognition capabilities. Achieved 20% increase in processing speed and 15% improvement in compression efficiency, making the technology viable for smaller robotic platforms.
LiDAR Autoencoders KITTI Dataset PCA/ICA Autonomous Vehicles
Generating Robotic-Arm Task-Sequence with Large Language Models
Sep 2023 – Dec 2023
Integrated GPT-4 with ROS Noetic to create an intelligent interface for Sawyer robotic arm control using natural language commands. The system acts as a high-level semantic planner that generates valid task sequences and converts human instructions into precise robotic commands. Implemented comprehensive safety verification to ensure plan validity before execution. Successfully demonstrated pick-and-place operations with coordinate-based navigation, bridging the gap between intuitive human communication and precise robotic control for enhanced human-robot collaboration.
Large Language Models ROS GPT-4 Human-Robot Interaction Natural Language Processing
Leechbot: Biomimetic Soft Robot for Object Manipulation
Jan 2024 – May 2024
Engineered a pneumatically actuated soft robot inspired by leech locomotion, featuring suction-based movement and adaptive grasping capabilities. The design incorporates a three-chambered fiber-reinforced actuator with PneuNets bending actuators for object manipulation. Conducted finite element modeling using Abaqus with Ecoflex 30 material properties to optimize performance. The robot successfully demonstrated locomotion on 40° slopes and effective object grasping, validating the biomimetic approach for navigating challenging terrains and underwater environments with potential applications in medical technology and disaster recovery.
Soft Robotics Pneumatic Actuation Biomimetics Finite Element Analysis Abaqus
Litterbug: Autonomous Litter Collection Robot
Sep 2023 – Dec 2023
Developed a mobile robotic platform for autonomous park cleaning using advanced path planning and computer vision. Integrated POMDP (Partially Observable Markov Decision Process) for high-level planning with wavefront coverage algorithms for local area sweeping. Implemented YOLOv8 litter detection model trained on TACO dataset, achieving 71% accurate localization and operating with track-based locomotion, manipulator arm, RGB camera, LiDAR, and GPS navigation. The system reduces manual cleanup efforts by 30 hours weekly through intelligent path optimization and reliable object detection capabilities.
Autonomous Navigation POMDP YOLO Computer Vision Path Planning LiDAR
Novel 3D-Printed Prosthetic Hand with Dual Actuation
Jan 2024 – May 2024
Designed and fabricated an innovative prosthetic hand combining cable-driven and pneumatic actuation systems for enhanced functionality and cost-effectiveness. The design features a 3D-printed structure with cable-driven clenching motion controlled by stepper motors and pneumatic unclenching using balloon actuators. Utilized SolidWorks for CAD modeling and implemented comprehensive testing using NASA Task Load Index evaluation. The dual actuation approach provides reliable grip mechanisms while maintaining lightweight construction, with successful demonstration of grasping various objects and 20% improvement in user dexterity compared to single-actuation alternatives.
Prosthetics 3D Printing SolidWorks Actuator Systems Biomedical Engineering
Integrated Voronoi-Wavefront Path Planning Algorithm
Academic Project
Developed an innovative hybrid path planning approach combining Voronoi cell decomposition with wavefront algorithms to optimize both safety and efficiency in autonomous navigation. The Voronoi method generates roadmaps that maximize distance from obstacles, while the wavefront algorithm provides optimal cost-based path finding. Successfully integrated Dijkstra's algorithm for graph traversal with grid-based wavefront propagation, demonstrating superior performance in complex environments. The hybrid approach addresses limitations of individual algorithms, providing safe obstacle avoidance while maintaining computational efficiency for real-time robotic applications.
Path Planning Voronoi Diagrams Wavefront Algorithm Dijkstra's Algorithm Autonomous Navigation
Bending and Twisting Behavior of Multilayer Graphene
Feb 2022 – Jun 2022
Conducted comprehensive finite element analysis of graphene's mechanical properties using ANSYS simulation software. Created detailed 3D models of graphene sheets (40cm × 40cm × 30 microns) and performed torsional and bending stress analysis to understand material behavior under various loading conditions. Generated stress-strain curves revealing unique mechanical characteristics: curved relationships for torsional loading and linear behavior with slope changes for bending. The research contributes to understanding graphene's potential applications in nanotechnology and provides foundational data for future material design applications in aerospace and electronics industries.
ANSYS Finite Element Analysis Material Science Graphene Mechanical Testing

Research Experience

Research Assistant – Imaging & Automation
Biomedical Optics Lab, Boston University
May 2024 – Jan 2025
Performed margin assessments of oral cancer tumors using elastic scattering spectroscopy, enhancing diagnosis for 10+ patients. Developed automated scanning system improving mechanical stability and saving 15 hours per month. Collaborated with medical professionals to validate system outcomes ensuring clinical requirement compliance. This research contributes to advancing early cancer detection methods through innovative optical imaging techniques.

Professional Experience

Machine Learning Intern
ROBOTRONIX ENGINEERING TECH PVT. LTD.
Apr 2023 – Aug 2023
Created automated Python scripts for data analysis, reducing processing time by 20% and boosting workflow efficiency. Applied machine learning principles to real-world problems, improving data analysis accuracy by 10% through advanced visualization techniques and statistical models.
IoT & Data Analytics Intern
Divergent Software Labs Pvt. Ltd.
Aug 2020 – Sep 2020
Developed data-driven irrigation system using Python to monitor soil conditions, boosting crop yield by 30%. Reduced water consumption by 15% through successful implementation of sustainable agricultural initiative with integrated microcontrollers and sensors.
Mechanical Design Intern
Hindustan Aeronautics Ltd.
Aug 2021 – Sep 2021
Analyzed aircraft manufacturing data, identifying inefficiencies that contributed to 15% decrease in quality control errors. Enhanced gear efficiency by 5% through Excel VBA analysis of manufacturing data and force calculations on gear teeth.

Technical Skills

Machine Learning & AI

Neural Networks Deep Learning Computer Vision NLP Reinforcement Learning GANs CNN LSTM

Programming & Frameworks

Python MATLAB C++ TensorFlow PyTorch Keras Scikit-learn OpenCV

Robotics & Control

ROS SLAM Path Planning Kinematics Control Systems Arduino Raspberry Pi LiDAR

Engineering Design & Analysis

SolidWorks ANSYS AutoCAD CATIA 3D Printing FEA CAD/CAM Finite Element Analysis

Education

Boston University, College of Engineering - Boston, MA, USA
Master of Science, Robotics & Autonomous Systems
GPA: 3.20/4.0 | Jan 2025

Relevant Coursework: Motion Planning, Perception, Machine Learning, Cyber-Physical Systems, Control Systems

Visvesvaraya Technological University, MS Ramaiah Institute of Technology - Bengaluru, KA, India
Bachelor of Engineering, Mechanical Engineering
GPA: 3.33/4.0 | May 2023

Relevant Coursework: CAD/CAM, Mechanical Design, Manufacturing, Fluid, Heat and Mass Transfer

Certifications

Six Sigma Methodology

Six Sigma White Belt
Council for Six Sigma Certification | June 2025
Certificate: ueVUU3Hmuj
Six Sigma Yellow Belt
Council for Six Sigma Certification | June 2025
Certificate: F6zExW4dRV
Six Sigma Green Belt
Council for Six Sigma Certification | July 2025
Certificate: Dk4AN6Ljit

Programming & Development

Learn Python Programming Masterclass
Udemy | Nov 2022
72.5 hours
Beginning C++ Programming - From Beginner to Beyond
Udemy | Nov 2022
46 hours
C Programming For Beginners - Master the C Language
Udemy | Nov 2022
24 hours
Algorithmic Toolbox
UC San Diego via Coursera | Nov 2022
Verify: CB36VE4BQGRF

Design & Engineering

SOLIDWORKS CAD Design Associate
Dassault Systèmes | July 2025
Certificate: C-5NCHSAMP93

Leadership & Teaching

Master's Level Teaching Assistant
Hand's on Engineering, Boston University
Jan 2024 – Dec 2024
Guided undergraduate students in CAD drawing, laser cutting & 3D printing techniques. Developed 5 detailed lesson plans for circuit-building workshops, resulting in 80% of participants reporting increased confidence in teamwork and technical skills.

Academic Project Papers

Enhancing Block-NeRF with Failure and Additional Losses
Course Project - EC518 Robot Learning, Boston University, 2024
Judo Technique Recognition for Enhanced Training
Course Project - EC523 Deep Learning, Boston University, 2024
Leechbot
Course Project - ME568 Soft Robotics, Boston University, 2024
A Novel 3D Printed Prosthetic Hand
Course Project - ME571 Medical Robotics, Boston University, 2024
Generating Robotic-Arm Task-Sequence with Large Language Models
Course Project - EC545 Cyber-Physical Systems, Boston University, 2023
Litter Bug
Course Project - EK505 Introduction to Robotics, Boston University, 2023
A Study on the Integration of Voronoi Cell Decomposition and Wavefront Algorithm
Course Project - ME570 Robot Motion Planning, Boston University, 2023
Data Reduction of LIDAR Depth-Maps for Self-Driving Cars
Course Project - Learning from Data, Boston University, 2023
Bending and Twisting Behavior of Multilayer Graphene
Junior Year Project - MS Ramaiah Institute of Technology, 2022

Research Statement

My research vision centers on developing intelligent autonomous systems that can seamlessly integrate with human environments and assist in complex tasks. I am particularly interested in advancing soft robotics and biomimetic systems that can adapt to dynamic environments while maintaining safety and efficiency. My work spans multiple domains including medical robotics, where I aim to develop assistive technologies for healthcare applications, and autonomous systems that leverage advanced perception and machine learning techniques.

Having completed my Master's degree in Robotics & Autonomous Systems at Boston University, I am now seeking to pursue doctoral research that focuses on the intersection of soft robotics, computer vision, and machine learning to create next-generation robotic systems that can work alongside humans in complex, unstructured environments. I am committed to translating theoretical advances into practical solutions that can benefit society, particularly in healthcare, manufacturing, and emergency response applications.