Autonomy Engineering Intern

• May 2023 - Present

Multi-view Stereo and Plane Sweep for 3D Point Cloud Reconstruction

• Nov 2022 - Dec 2022

· Obtained rectified views from left and right images correspondences. Computed disparity map and depth using 2-view stereo and plane sweep algorithm. · Obtained the 3D point clouds from the depth computed.

Pick and Place with Panda Arm Motion Planning

• Oct 2022 - Dec 2022

· Using inverse kinematics to determine joint angles needed for effective path planning. Both static and dynamic blocks were picked up and stacked. · ROS is used as framework and gazebo for simulation.

NYC Taxi Cab Tips Prediction

• Sep 2022 - Dec 2022

· Used the NYC public data to analyse the factors that effected tipping using multiple models. Combined multiple datasets such as weather and holidays to see correlation. · Hypertuned multiple models for good RMSE results.

Flight Operations Analysis Intern

• Jul 2021 - Sep 2021

• Worked on automatic Flight classification into Multiple groups for more efficient flight operations analysis. • The multiple classes considered were test, delivery, training, and normal flights. • Worked on automatic Flight classification into Multiple groups for more efficient flight operations analysis. • The multiple classes considered were test, delivery, training, and normal flights.

Autonomous Multi-Drone Path Planning for Large Structure Inspections

• Apr 2021 - Aug 2021

Advisor: Dr. Debasish Ghose, Guidance | Control and Decision Systems Laboratory, IISc Bangalore · One amongst 60 students selected from across India as a candidate for the Indian National Academy of Engineering Mentorship Program 2021-22. · Developed an in-house controller and a 3D coverage path planning algorithm using Lissajous curves without explicit geometry. It can easily be implemented at nominal computation costs. Advisor: Dr. Debasish Ghose, Guidance | Control and Decision Systems Laboratory, IISc Bangalore · One amongst 60 students selected from across India as a candidate for the Indian National Academy of Engineering Mentorship Program 2021-22. · Developed an in-house controller and a 3D coverage path planning algorithm using Lissajous curves without explicit geometry. It can easily be implemented at nominal computation costs.

Derivative-free Adaptive Spacecraft Attitude Control

• Mar 2021 - Aug 2021

Advisor: Dr. Dipak Kumar Giri | Space Dynamics and Flight Control Laboratory · Developed a purely adaptive control system with derivative-free weight update laws on a spacecraft attitude system represented using MRPs. · Working on applications of machine learning and evolutionary optimization techniques on adaptive laws and its update rates as a part of my undergraduate research project. Advisor: Dr. Dipak Kumar Giri | Space Dynamics and Flight Control Laboratory · Developed a purely adaptive control system with derivative-free weight update laws on a spacecraft attitude system represented using MRPs. · Working on applications of machine learning and evolutionary optimization techniques on adaptive laws and its update rates as a part of my undergraduate research project.

Control System Design using Bond Graph Representation

• Mar 2021 - Jun 2021

Advisor: Dr. N. Selvaganesan | Indian Control Conference 2021 · Developed a power-based graphical representation of quadcopter system and its controller using bond graph approach. Reduced model is derived and is used to obtain feedback. · The entire closed-loop MIMO system is represented using 20-sim software and simulations were performed to meet satisfactory stable responses under tracking and disturbance conditions. Advisor: Dr. N. Selvaganesan | Indian Control Conference 2021 · Developed a power-based graphical representation of quadcopter system and its controller using bond graph approach. Reduced model is derived and is used to obtain feedback. · The entire closed-loop MIMO system is represented using 20-sim software and simulations were performed to meet satisfactory stable responses under tracking and disturbance conditions.

Controls Development Engineer

• Sep 2020 - Feb 2021

• Automating the process of tuning PID controllers in a control system simultaneously. • Find the optimal gains using evolutionary optimization techniques such as the Genetic Algorithm. • Using PX4 simulator for the simulation environment with Quadcopter simulation using JMAVsim or connections to Pixhawk. • MAVLink connections through UDP ports were used in MATLAB for optimization and parameter setting. • Automating the process of tuning PID controllers in a control system simultaneously. • Find the optimal gains using evolutionary optimization techniques such as the Genetic Algorithm. • Using PX4 simulator for the simulation environment with Quadcopter simulation using JMAVsim or connections to Pixhawk. • MAVLink connections through UDP ports were used in MATLAB for optimization and parameter setting.

Wingtip Vortices Analysis

• Nov 2019 - Dec 2020

PROF NAVROSE AEROSPACE DEPT, IIT KANPUR • Verification of Batchelor and Lamb-Oseen Models at low Reynolds using curve fitting of simulation data for Re=1000 using MATLAB curve fitting tools and VisIT flow visualization. • Performed the above activity by analysis of the : 1D curvefits of streamwise velocity and vorticity along multiple lines with different slopes on Lamb-oseen model. 2D curvefits of streamwise velocity, vorticity, and azimuthal velocity at different streamwise locations on Batchelor model. • Analysis of instability in wingtip vortices and capture the formation of vortices. • Perform an extensive analysis for flows with various Reynolds number for both single wing and split-wing configurations. • Use Nek5000 to make simulations for relevant analysis and understanding the computational conditions Show less

Deep Learning Internship

• May 2020 - Sep 2020

• Classify satellite images into classes of military targets using PyTorch and Sci-Kit learn libraries. • Worked with a dataset of planes ( 25000+ images) to do binary classification and a dataset of different classes of ships ( 8000+) to do multiclass classification. • Implementation of Kalman Filter over Neural Networks to perform target tracking on passive Radar signal-bearing angles. Using the bearing angles and timestamps, we predict the target trajectory. • Made the Dataset for training and testing using Robotics System Toolkit, MATLAB. Show less