About Me
I am an Erasmus Mundus Joint Masters Scholar in Intelligent Field Robotic Systems at University of Girona, Spain and the University of Zagreb, Croatia. My research interests span a diverse range of cutting-edge disciplines, including robotics, computer vision, and machine/deep learning.
I am passionate about exploring the intersection of these fields to push the boundaries of technology and create innovative solutions that address complex real world challenges.
Additionally, I have proven leadership qualities demonstrated in global communities including Google Developer Student Clubs, TEDx, Google Developer Groups Live, U.S. Embassy programs, and 10Pearls.
Education
Erasmus Mundus Joint Masters in Intelligent Field Robotic Systems | University of Girona & University of Zagreb
Semester I & II in Girona: Autonomous Systems, Machine Learning, Multiview Geometry, Probabilistic Robotics (Kalman Filtering), and Robot Manipulation, Localization (SLAM), Planning, Perception (Computer Vision), and Intervention.
Semester III in Zagreb: Aerial Robotics, Multi-Robot Systems, Human-Robot Interaction, Robotic Sensing, Perception, & Actuation, Deep Learning, and Ethics & Technology.
B.E. in Software Engineering | Mehran University of Engineering and Technology
Agent Based Intelligent Systems, Data Science & Analytics, Simulation & Modeling, Cloud Computing, Statistics and Probablity
CGPA 3.96 / 4.00 - Silver Medal Distinction & First Position
Experiences
Master Thesis/Intern | Saxion Smart Mechatronics and Robotics Research Group
My research work on “Vision-based Tracking and Following of a Moving Target Using a UAV”, focused on challenges mainly in RGB modality for robust person detection and tracking, dealing with occlusions and identity switches, and real-time prediction in dynamic environments. The work integrates a hybrid system combining filtering-based tracking (Kalman Filter), vision-based deep learning models (YOLO), and adaptive control strategies. Overall, this research seeks to leverage situational-aware techniques to enable intelligent UAV deployment in security-critical scenarios detecting potential suspicious targets and track them in real-time with minimal response time.
ROSCon 2024 Diversity Scholar | Open Robotics, Denmark
I secured a diversity scholarship to attend the ROSCon 2024 in Denmark where I had the opportunity to network with companies and ROS contributors globally. I specifically got extensive hands-on experience by attending the workshops named “Open source, open hardware hand-held mobile mapping system for large scale surveys” which gave exposure to essential processes such as LIDAR odometry and multi-session refinement for large-scale mapping and “ros2_control” where we learned about controller chaining, fallback controllers, and async controllers.
Robotics Intern | Paltech Robotics GmbH
Worked on testing and comparing two new ultrasonic sensors i.e. Bosch and Valeo for the collision avoidance task to include the safety braking feature (setting thresholds to slow down or stop the robot with ROS2) which involved performing multiple field tests of different high grass.
My Projects
Sim2Real: Controlling a Swarm of Crazyflies using Reynolds Rules and Consensus Protocol
This project implements swarm control for Crazyflies UAVs using Reynolds Rules for flocking and a Consensus Protocol for coordinated movement. It integrates rendezvous and formation control in ROS2 and Gazebo, enabling agents to converge and maintain geometric formations. Tested in both simulation and real-world environments, the system demonstrates adaptability and scalability.
Stereo Visual-Odometry (VO) on the KITTI Dataset
Implementation of Stereo VO pipeline in Python on the KITTI dataset. It processes stereo image data using SIFT, feature matching using BFMatcher, triangulation of points, to estimate the motion of a camera (w.r.t its starting position) in 3D space using the approach of minimizing the 3D to 2D reprojection error with PnP and RANSAC.
Deep Learning
In this Deep Learning course lab work, PyTorch implementations included working on logistic regression and gradient descent, implementing fully connected models on the MNIST dataset, Convolutional models for image classification tasks on MNIST and CIFAR, Recurrent models for analysis of sentiment classification with the Stanford Sentiment Treebank (SST) dataset followed by detailed implementations on metric embeddings.
Human Detection and Tracking
This project focuses on human detection and tracking using the state-of-the-art YOLOv9 object detection model and the DeepSORT multi-object tracking algorithm. The methodology integrates Kalman filtering for motion prediction and deep learning-based appearance matching. The system is tested under various conditions, addressing challenges such as occlusions, identity switches, and tracking interruptions.
Frontier Based Exploration Using Kobuki Turtlebot
Using RGB-D camera mounted on a Kobuki Turtlebot, the project integrates advanced path planning techniques, combining the RRT* algorithm with Dubins path to map unknown environments with the primary objective of enabling the Turtlebot to autonomously explore unknown environments by identifying and navigating to frontiers. A hybrid control system, which merges PID control with principles from the Pure Pursuit Controller is used. Validated both in simulation and real world.
Monocular Visual Odometry (M-VO) for an Autonomous Underwater Vehicle (AUV)
M-VO for an AUV through an integrated approach combining extended Kalman filter (EKF) based navigation. The methodology employs SIFT feature detection and FLANN matching to (offline / post) process images from a ROSBag. A key contribution of this work is the incorporation of EKF to provide a refined estimation of the vehicle´s motion and trajectory.
Pose Based SLAM using the Extended Kalman Filter (EKF) on a Kobuki Turtlebot
Implemented a Pose-Based P- EKF SLAM system integrating IMU and 2D LiDAR data on a Kobuki Turtlebot. The algorithm maintains robot pose history for map building and localization, using ICP for scan matching and robust state updates. Validated in both simulated (Stonefish) and real-world environments, PEKFSLAM demonstrated superior accuracy and stability compared to conventional EKF-based SLAM approaches.
Kinematic Control System for a Mobile Manipulator, based on the Task-Priority Redundancy Resolution Algorithm
Designed and implemented a kinematic control system for a mobile manipulator (Kobuki Turtlebot 2 with a 4-DOF uArm Swift Pro), using a task-priority redundancy resolution algorithm. Developed in ROS and tested in the Stonefish simulator, the system performed complex pick-and-place tasks, including ArUco marker-based navigation.
Stereo Visual Odometry (VO) for Grizzly Robotic Utility Vehicle
Developed VO pipeline from stereo camera calibration, feature extraction, and matching using SURF features and utilizing bucketing strategies and circular matching for accurate apparent motion computation and effective noise/outlier rejection, Structure from motion (2D-to-2D, 3D-to-2D, and 3D-to-3D) for triangulation and refinement using bundle adjustment. The final VO trajectory was also extensively compared with GPS-generated ground truth data.
