Sneha Ganesh Nayak

Engineering professional with expertise in autonomous systems development, covering motion planning, localization & mapping, computer vision, and machine learning. Skilled in various programming languages, cloud technologies, and simulation software, with significant contributions in robotics and autonomous system development.

Human Detection and Tracker using YOLOv4 and Kernelized Correlation Filter:

Executed human detection algorithm using Yolov4 and non-maximum suppression. Employed OpenCV’s KCF Multi tracker for tracking in C++

Multi-Robot decision-making, navigation & collision avoidance in Warehouses:

Explored decentralized coordination with 100 TurtleBot3 robots while avoiding collisions using ORCA as local and A* with RRT as global planners employing ROS, Stage, C++, and Python.

Connect Four Using Reinforcement Learning:

Programmed a Connect 4 game to have the model train through self-play and reinforcement learning, with performance improvements recorded as a function of no of games using Python

Reinforced Finger Actuation with Haptic Feedback for Piano Lessons:

Prototyped a wearable device with soft finger-reinforcing actuators and haptic feedback as a 5ms vibration mechanism in assisting individuals to play the piano using Arduino

Active Target Tracking with Self-Triggered Communications in Multi-Robot Teams:

Minimized communication between a team of robots in a decentralized target tracking system using a self-triggered communication strategy and Kalman Filter using Python

Implementation of CNN & SVM with PCA/LDA Reduction on FashionMNIST:

Leveraged SVM with linear, polynomial, and RBF kernels and built a CNN architecture with 6 epochs; accuracy by SVM was76%; CNN was 91% using PyTorch & Python

Bayesian and KNN Classifier with PCA/LDA Reduction on FashionMNIST:

Compared optimal Bayesian classifier with K Nearest Neighbors; the highest accuracy accomplished by Bayes was 82.11% and kNN was 86.15%

Lane Detection and Tracking in Self-Driving Cars:

Attained a simple Lane Detection to mimic the Lane Departure Warning System using Hough lines & histogram of Lane pixels in Python3

Lucas-Kanade Template Tracker:

Devised a template tracker to track a car, human, & box on a table using affine transform & warping using OpenCV, Python3

Traffic Sign Detection and Classification:

Executed using MSER and HSV for detection and a multi-class SVM for classification, trained using HOGs of images in Python3