Projects
A Glimpse of my Projects
Machine Learning Enabled Back Exoskeleton
-
Designed a machine learning-enabled back exoskeleton to adapt to the user’s movements, while processing marker data and performing inverse kinematics/dynamics analysis for 10 subjects to estimate flexion moment.
-
Deployed a cutting-edge Temporal Convolutional Neural Network (TCN) model on a Jetson Nano controller by leveraging a 9-axis IMU and motor data integration to predict lumbar flexion moments.
-
Led the Mechanical team to enhance exoskeleton ergonomics by 30%. Engineered an Arduino Teensy velocity controller for closed-loop motor torque control.
Incorporating Multi-Modal Sensing and Navigation for Independent Maze Exploration
-
Devised a sequential neural network model, achieving 91.8% accuracy in detecting traffic signs.
-
Implemented a finite state machine incorporating predictive signs, sensor data, and angular-linear velocity controllers to navigate the maze in under 180 seconds.
AR Virus Spread Simulator
-
Developed and designed the "AR Virus Spread Simulator" application using Unity-3D, a cross-platform game engine, to achieve the core objective of simulating the real-world spread of various viruses through augmented reality (AR), creating an immersive and interactive virus spread simulation.
-
Programmed the application using C# and integrated 3D graphic design software like Blender and 3D Builder to import virus models and other 3D objects into the Unity environment.
-
This application is currently hosted and available for download on Google Play Store under the name "AR Virus Spread Simulator".
Interactive Tic-Tac-Toe
-
Created a gamified object pick-and-place system featuring Tic-Tac-Toe using a KUKA 7DOF Robotic Manipulator (LBR IIWA 14 R820 7-axis).
-
The robot employs a 3rd Path Order Joint Space Trajectory for precise movement within the Tic-Tac-Toe grid.
-
Players participate by taking turns to select a shelf, prompting the robot to identify and place their chosen box onto the designated shelf.
Nurse Bot”, a 2-in-1 Robotic Medicine Delivery System & sanitizing bot
-
Spearheaded the creation of a versatile Nurse Bot proficient in medicine delivery, environmental sanitization (including UV disinfection, and sanitizer spraying), and real-time monitoring via an Android app.
-
Crafted an intuitive app interface for seamless mode transitions, underlining its value in healthcare and sanitation management.
Animating Dynamics of Single-rotor helicopter drones
-
Led the development of an immersive virtual reality animation within V-Realm Builder, incorporating it with SIMULINK to execute dynamic simulations. This involved precise manipulation of state space variables for real-time analysis of the drone's performance.
-
Designed a comprehensive 3D model of a single-rotor helicopter drone using SolidWorks, encompassing all key components and specifications.
Detection of DDoS Attacks using Machine Learning
-
Researched and evaluated machine learning models for the detection of Distributed Denial-of-Service (DDoS) attacks, showcasing the effectiveness of supervised learning methods such as Support Vector Machine (SVM) and Random Forest in achieving over 99% accuracy.
-
Implemented data preprocessing, feature selection, and dimensionality reduction techniques to optimize model performance, demonstrating the ability to reduce the dataset to just two principal components without compromising results.
Inverse Template Matching for Humanoid Robot Vision Systems
-
Pioneered Inverse Template Matching (ITM), an innovative algorithm using area-based subtraction (ABS) to address occlusion challenges in object detection by searching for similarities between input and template images.
-
Validated the ITM algorithm on real-world images with varying occlusion levels, providing comprehensive results and methodologies.
MedBuddy - The Medicine Delivery Robot
-
Engineered a Bluetooth-controlled robot car using Arduino Uno and a smartphone for live feed via MIT App Inventor, featuring a medication delivery tray to enhance patient safety and minimize contact for medical personnel.
-
Implemented a framework with sensors and actuators connected to Arduino Uno, enabling Bluetooth-based data transmission from a mobile app to control the robot through a predefined algorithm
Exploring the Impact of Human Likeness on Trust in Human-Robot Interaction
-
The study explored how human-likeness in robots affects trust in human-robot interactions. Trust increased with greater human-likeness but declined when robots seemed 'uncanny,' echoing the uncanny valley concept.
-
The project's findings impact robot design, highlighting the importance of human-like features for trust. It also stresses the need to account for individual differences in studying trust in human-robot interaction.
Automated Door System Using Arduino For Crowd Management
-
Constructed an Automated Door System utilizing Arduino, PIR sensors for human presence detection, and an intelligent algorithm to control a servo motor for door operation. The system also displays room occupancy on an LCD screen and emphasizes safety measures, by limiting room occupancy in response to COVID-19.
-
Leveraged Arduino's user-friendly platform to create an interactive system, emphasizing benefits for physically challenged individuals and hands-free access scenarios.
