I'm
Swati Vishwanath Shirke
Robotics Engineer from WPI university
Robot Control Motion Planning Navigation Engineer, ROS1 and ROS2 Developer, Computer Vision Enthusiast
I am a graduate student researcher in Robotics Engineering at Worcester Polytechnic Institute (WPI), specializing in Robot Control and Motion Planning. I am currently involved with the Aerial Robot Control and Perception (ACP) Lab at WPI. I am working on an Aerial Delivery System using multiple quadrotors. Earlier I completed my research on safety-critical control for autonomous vehicle fleets using multi-agent Deep Reinforcement Learning . I I am proficient in ROS, C++, Python, MATLAB, and Linux, with hands-on experience in robot programming, controls, motion planning, and data structuring. Additionally, I have expertise in BLDC motors, SONAR sensors, and platforms such as Jetson Orin Nano and BeagleBone Black. I worked for five years as a PEGA Application Developer at Vodafone India Services Pvt Ltd, starting as a graduate engineer trainee and progressing to Deputy Manager. I began my journey in robotics at the Cognitive Robotics and Intelligent System Lab (CRISTL) at Vishwakarma Institute of Technology.
Worcester Polytechnic Institute | August 2023 - Present
Coursework:
Robot ControlVishwakarma Institute of Technology (VIT) | July 2014 - May 2017
Coursework: Control System, Optimization Techniques, Pattern Recognition, Electronics Circuit Design
Aerial-Robot Control and Perception (ACP) Lab, Worcester, MA | Aug 2024 - Present
Developing an aerial payload delivery system using multiple quadrotors. Engineering an end-to-end pipeline for trajectory following multi-drone payload transportation system integrating, ROS2, RViz and Jetson Orin Nano.
Under guidance of Prof. Siavash Farzan, Worcester, MA | Jan 2024 - Jul 2024
Built a multi-agent Reinforcement Learning framework for collaborative control of an autonomous vehicle fleet. Accomplished success rates of 97% and 70% for lane merging and intersection respectively.
Vodafone India Services Pvt Ltd, India | Jan 2021 - Jul 2022
Led a team of 6 junior developers and improved work efficiency of the team by 8%: technical design creation; project deliveries; task allocation; effort estimation; bug resolution.
Vodafone India Services Pvt Ltd, India | Nov 2018 - Jan 2021
Automated 12 critical modules of a business case life cycle: system thinking; impact analysis; scalable & maintainable application design, and business report generation.
Vodafone India Services Pvt Ltd, India | Jul 2018 - Oct 2018
Collaborated with senior leaders to establish work processes as part of the pilot batch of the application development team.
Vodafone India Services Pvt Ltd, India | Jul 2017 - Jul 2018
Developed a proof of concept to draw in customer interest and generate business opportunities.
Click the Options from below list to see the description
Devising collision-avoidant trajectory control for three drones using Model Predictive Control (MPC) and configuring custom hardware with Jetson Orin Nano. Achieved trajectory control for a single quadrotor with a payload.
GithubDesigned an MPC-based trajectory control with the Ackerman steering model achieving an RMSE below 0.5 meter.
GithubCrafted hybrid A* motion planner to manoeuvre a differential drive robot and Ackermann steering with kinematic constraints in a simulated environment using kinodynamic simulation. Achieved parking in tight parking spots, achieving average times of 20-35 seconds while avoiding collisions.
GithubDeveloped a multi-agent reinforcement learning framework for autonomous vehicle fleet control, achieving a 97% success rate in collision avoidance, surpassing the baseline performance of an 87% success rate.
Devised a 3 DOF mobile robot integrating C++, Python, ROS1 with Arduino, BeagleBoneBlack, and BLDC motor. Built motor control library for robust PID control in C++ and deployed ROSBRIDGE server for web connectivity
GithubImplemented an A* combinatorial planner and a Probabilistic Roadmap (PRM) sampling-based planner for navigating a firetruck and an arsonist Wumpus through a cluttered environment with burning obstacles. The A* planner averaged 5 seconds, significantly faster than the PRM planner’s 58 seconds.
GithubImplemented an A* combinatorial planner and a Probabilistic Roadmap (PRM) sampling-based planner for navigating a firetruck and an arsonist Wumpus through a cluttered environment with burning obstacles. The A* planner averaged 5 seconds, significantly faster than the PRM planner’s 58 seconds.
Github