Warebots: Multi Agent Path Finding

The project explores the Multi-Agent Pathfinding (MAPF) problem, focusing on one-shot MAPF and its dynamic counterpart, Lifelong MAPF (LMAPF). It delves into the Conflict-Based Search Algorithm (CBS), a two-level approach for resolving conflicts between agents, showcasing its effectiveness in small-scale environments inspired by the League of Robot Runners competition.

LQR based Thrust control of Quadcopter

Implemented an LQR controller, which is linearized depending on the quadrotor’s state, unifies the control of rotational and translational states, handles time-varying system dynamics, control parameters and does direct motor control.

Human Facial Emotion Recognition & Classification

As part of a semester-long machine learning team project, built a CNN model that could recognize and classify human emotions from facial images. The objective was to train the model to identify the emotion of a person from their faces.

Super Visual Lidar Odometry and Mapping

Implemented a robust real-time ROS-based framework for accurate trajectory estimation, 3D Mapping, and Localization by augmenting the feature extraction and matching algorithm with Super-Point descriptor and SuperGlue matching algorithm.