Robotics Software Projects

My software projects combine robotics algorithms, control theory, and machine learning to build intelligent systems. I’ve developed solutions for robot manipulation, localization, motion planning, dynamic modeling, reinforcement learning, deep learning and computer vision using tools like Python, PyTorch, and ROS2 — bridging theory with scalable, real-world applications.

Deep Learning Model for Forward Dynamics of Robot Arm

I developed a deep learning model to learn the forward dynamics of a 3-link robotic arm to imitate the behavior of another arm.

Model Predictive Control and Forward Dynamics Learning

I developed a Model Predictive Controller (MPC) and integrated it with a learned dynamic model

Deep Q-Network for Robotic Arm Control

I developed and implemented a Deep Q-Network (DQN) from scratch to control a 2-link robotic arm to reach specified goal positions. This project involved designing the neural network architecture, implementing the training process, and evaluating model performance on multiple goal-reaching tasks.

Behavioral Cloning for Autonomous Navigation

This project leverages deep learning to enable the robot make decisions based on location or vision inputs

Extended Kalman Filter Implementation for Mobile Robot Localization

I developed and implemented an Extended Kalman Filter (EKF) to estimate the 2D position and orientation of a mobile robot using ROS 2. I integrated sensor data from range and bearing measurements to landmarks, ensuring robust and accurate localization in real time.

Rapidly Exploring Random Tree Based Motion Planning

I implemented the RRT algorithm for sampling-based motion planning and used it to control a UR5 robot arm.

Cartesian Control for Robotic Arms

I developed a ROS2 node to perform Cartesian control and numerical inverse kinematics for the Franka Panda and Universal UR5 robotic arms. The information given was the transforms from the base to the end-effector and an array of transforms from the base to each joint.

Dynamics Engine for Kinematic Chain

I implemented a physics simulator for a 2D kinematic chain with an arbitrary number of links, utilizing the Newton-Euler algorithm for solving dynamics and Euler integration for state updates.