resume

MS Robotics Engineering

Relevant Coursework: Deep Learning, Reinforcement Learning, Computer Vision

BE Mechanical Engineering

Graduate Student Researcher

• Waste Pile Rearrangement for Robotic Waste Sorting in Recycling Facilities
  Investigating the use of depth images and point clouds to effectively reconfigure waste piles on a conveyor belt using a robotic manipulator in order to uncover occluded and completely covered objects in the waste stream.

  Skills: Robot Operating System (ROS), Point Cloud Library (PCL), Python, C++

Research Associate

• RoboTRAC - Autonomous Vascular Access Robot for trauma care in resource-limited setting
  • Conducted comparison studies for segmentation of ultrasound images using multiple deep learning frameworks to identify femoral blood vessels, to access them autonomously by performing robotic needle insertions.
  • Also responsible for developing and maintaining ROS software stack of the robot to perform needle insertion experiments.

  Skills: Deep Learning, Robot Operating System (ROS), PyTorch, Python

• Pipe Crawler - Autonomous Gas Pipe Repair
  Developed the entire software stack for control, navigation and localization of pipe crawler robot performing inspection and repair in gas pipes.

  Skills: Robot Operating System (ROS)-2, RViz, Python

Research Intern

• Robust Quadrupedal Locomotion using Reinforcement Learning
  Implemented the Augmented Random Search (ARS) algorithm to train linear feedback policies for quadrupedal locomotion over rough terrain, and implemented on the custom-build quadruped StochLite, resulting in robot climbing over 15 degrees slopes.

  Skills: Reinforcement Learning, PyBullet, Python

• StochLite - Quadrupedal Walking Robot
  Designed and fabricated a small, light-weight, and easy-to-modify quadruped which can be used for software development and can perform basic dexterous tasks like stair-climbing and traverse rough terrain, using 3D printing and off-the-shelf components.

  Skills: SolidWorks, 3D Printing, CAD