HemanthJoseph

ML-Projects-with-Deployment

An end-to-end Machine Learning project from writing a Jupyter notebook to check the viability of the solution, to breaking down the same into modular code, creating a Flask web app integrated with a HTML template to make a website interface, and deploying on AWS and Azure.

Design-and-Simulation-of-a-mobile-robot

Design a robot car using Solidworks. Convert it to a URDF (Universal Robot Description Format). Apply inertial physics to it, add a LIDAR sensor, controllers for motion control. Create a ROS package and run it in Gazebo.

Dijkstra-based-Contact-Routing-Graphs

The principles and techniques for routing among ground elements and spacecraft in near-Earth orbit and deep-space are enacted in the Contact Graph Routing (CGR) framework. CGR blends a set of non-trivial algorithm adaptations, space operations concepts, time-dynamic scheduling and graph models like Dijkstra to build a Delay Tolerant Networks, DTN.

Dijkstra-Path-Planning-Mobile-Robot

Using Dijkstra Path Planning algorithm to solve the shortest path problem for a mobile robot.

Kinematics-and-Dynamics-on-Kuka-IIWA-Robot-Arm

Calculated DH parameters and evaluated the Forward Kinematics, Inverse Kinematics and Dynamics on the Kuka iiwa Robot Arm

MEDUSA-An-Urban-Search-And-Assistance-Drone

A novel product, a drone, to help with Emergency Services.

Path-Planning-a-Turtlebot-using-A-Star-Algorithm-in-ROS

Using A* algorithm to generate a path for a non-holonomic robot like the Turtlebot. Use the path generated to move the robot along that path in a Gazebo environment.

Stereovison-Depth-Mapping

Using the OpenCV computer vision library and NumPy to use two images of the same scene together and estimate the depth of the objects in it through Epipolar geometry.

Basic-Machine-Learning-for-Breast-Cancer-Detection

Using simple ML algorithms like Logistic Regression, Decision Trees, Random Forests from the Scikit learn library to classify breast cancer data

Enhancing-Rehabilitation-Outcomes-with-a-Virtual-Reality-Integrated-Robotic-System-for-Arm-Leg-Gait

ENPM640: Rehabilitation Robotics

ENPM661-A_Star

Histopathological-Cancer-Detection-Using-Deep-Learning

Building a Deep Learning model based on Convolutional Neural Networks to help create a model which can classify histopathological slide images of lymph node sections and determine if a certain image has metastatic tissues.

Lane-Detection-with-Turn-Prediction-for-Self-Driving

Using traditional computer vision techniques to detect lanes on the road. Also done here is finding out the curvature of the road which is used to predict turns.

Monocular-Visual-Odometry

Final Project in Perception for Autonomous Robots ENPM 673

urban_search_and_rescue

ROS1 package for search and rescue in C++ using move_base and gmapping package.

Affine-Structure-from-Motion

Applying the the Affine SfM pipeline for 3D scene reconstruction.

AI-for-Robotics---Udacity

A repository of my practice and solutions to the Problem Sets from the Artificial Intelligence for Robotics course taught by Sebastian Thrun on Udacity.

Deep-Learning-from-Scratch

An in-depth understanding of Deep Learning fundamentals only using Python and the NumPy library.

DSA-Python

Generative-Adversarial-Network-on-MNIST-dataset

Creating a straightforward GAN by replicating Ian Goodfellow's original GAN paper and using it to generate fake MNIST data.

HemanthJoseph.github.io

Human-Emotion-Acquisition-And-Recognition-Technology

Using the principles of Product Development and Product Management to design robot architectures for Human Robot Interactions.

Humanoid-Robot-Modelling-for-Pick-and-Place

Image-Stitching

Using the OpenCV computer vision library to stitch images of the same scene together.

Mathematics-for-Control-Theory

MNIST-Classifier-NN-vs-CNN

I used PyTorch to build a multiclass classifier. I made a simple Neural Network and a simple Convolutional Network architecture and compared their results.

Sampling-based-path-planning-using-RRT-Star-algorithm

Using a sampling-based path planning algorithm such as the RRT* algorithm to generate the map and a path to traverse in the map. A variation was explored where randomly sampled new points were generated closer to the goal thus ensuring faster convergence.

SLAM-from-Scratch

Performing SLAM from Scratch using only Python.

Slot-Controller-and-Simple-Impedance-Controller-for-Rehabilitation-Robots

sudoku_repository

code for all sudoku files