Piyush Sethy

Mechanical engineering student with a strong interest in robotics, research, and machine learning. Experienced in designing and building mechanical and robotic systems, applying machine learning techniques to control and analyze dynamic systems, and working with additive manufacturing technologies. Proven ability to integrate software and hardware solutions in research-driven projects, with experience in bio-inspired robotics, additive manufacturing, and data analysis. Passionate about contributing to cutting-edge engineering and scientific innovation.

Controlling double Pendulum through Reinforcement Learning (https://drive.google.com/file/d/1XqSRatpQPZDgNCDPgtEOpoiqBQmILkXI/view)

• Modelled Open AI's double pendulum system called the Acrobot by using reinforcement learning techniques, Designed and trained a Tabular Q-learning algorithm by discretizing environment states and tuning learning rate and epsilon-greedy exploration to improve reward performance, Improved model performance by implementing a Deep Q-Network in PyTorch to handle continuous system dynamics

Mechanical Engineering and Robotics Portfolio (https://psethy2005.github.io/ePortfolio/)

• Designed and validated multiple mechanical systems through the full engineering lifecycle, including CAD modelling in SolidWorks, structural analysis (FEA), and physical testing, documented in a curated online portfolio, Led mechanical design efforts for autonomous and electromechanical systems, focusing on assemblies, load-bearing structures, and design under strict weight and volume constraints
• Autonomous OTV: Designed and created CAD models of robotic claw and key chassis components; implemented ultrasonic sensing strategies to distinguish material types, pickup, and navigation through obstacles
• Escalator System Design: Developed truss support and guide rail assemblies for an escalator cad model in a team setting; performed FEA on stair components to evaluate structural integrity and load performance