Posts by Collection

Advanced Engineering Intern - Powertrain

Isuzu Technical Center of America

• Assisted in the development (MiL/HiL) of the temperature control of the after-treatment system (ATS) of a commercial diesel engine using Model Predictive Control (MPC) to reduce NOx emissions and optimize fuel injection
• Improved MPC controller results by modifying problem setup in Simulink and modifying the problem formulation for the real-time custom solver tool FORCESPRO, bringing the results closer to the baseline
• Set up CAN communication protocol between ECU and PC, and developed and flashed MPC control algorithm on ECU for HiL testing

Machine Learning Intern

Lateetud

• Performed a comparative study of 5 different deep learning, machine learning, and image processing methods for the automatic identification of 11 different US tax forms
• Integrated EAST, a deep neural network architecture, with Tesseract OCR for 95%+ accuracy

Coverage patterns generated by two unicycles pursuing each other

Published in European Control Conference, 2023

Recommended citation: S. S. Purohit and A. Sinha, "Coverage patterns generated by two unicycles pursuing each other," 2023 European Control Conference (ECC), Bucharest, Romania, 2023, pp. 1-6, doi: 10.23919/ECC57647.2023.10178386.
Download Paper

Reachable Sets of Homogeneous Polynomial Dynamical Systems Using Exact Solutions

Published in IEEE Control Systems Letters (Volume: 8), 2024

Recommended citation: S. S. Purohit, C. Chen and R. Vasudevan, "Reachable Sets of Homogeneous Polynomial Dynamical Systems Using Exact Solutions," in IEEE Control Systems Letters, vol. 8, pp. 742-747, 2024, doi: 10.1109/LCSYS.2024.3403468.
Download Paper

Kalman Filtering on Lie Groups

Published: May 01, 2021

• Modeled a fixed, rotating ground robot as a smooth manifold and implemented an observer design on Lie Groups to determine its orientation in the ground frame using Python
• Plotted and compared the behavior of the observer and determined the optimal observer design based on the convergence rate, overshoot, and steady-state errors for the set of parameters
• Observed faster convergence rate and lower overshoot compared to a standard Kalman Filter

Magnetic Shape Sensing of Continuum Robot Segments

Published: July 01, 2022

• Designed experiments to evaluate the performance of an improved embedded magnetic shape-sensing system for the 3D bending of a single-segment, centimeter-scale, continuum robot
• Designed and manufactured a continuum robot segment and an experimental setup in Solidworks and performed data acquisition using Arduino-Uno to obtain magnetic and inertial data as a part of the evaluation experiments
• Reduced errors in bending angle by 12% through calibration and iterative design and development

Multi-agent Pattern Generation using a Range-Only Control Law

Published: August 01, 2022

• Determined and proved conditions necessary to achieve hypotrochoid patterns around a stationary point for two unicycles pursuing each other with equal and unequal velocities using a range-only control strategy
• Devised a switching law to control the minimum and maximum radius of hypotrochoids generated by the pursuit
• Extended the analysis to achieve patterns for multiple (more than two) unicycles with equal velocities

Reachable Sets of Homogeneous Polynomial Dynamical Systems

Published: August 01, 2023

• Devised methods to overapproximate the reachable sets of odeco Homogeneous Polynomial Dynamical Systems (odeco-HPDS) using their exact solutions obtained using tensor algebra
• Formulated a zonotope decomposition method for odeco HPDS with zero and constant control
• Formalized a method to transform certain general (non-odeco) HPDSs to use our methods
• Displayed an improvement in performance over the existing reachability analysis tool, CORA in terms of tightness of overapproximations

Job-Shop Scheduling Using Quantum Annealing

Published: December 01, 2023

• Formulated the Job-Shop Scheduling Problem as a Quadratic Unconstrained Binary Optimization (QUBO) Problem through conversion to a linear program followed by specific binary encoding of continuous variables for quantum annealing solvers
• Determined various methods for constraint reduction such as removing redundancies, formulating and using the dual problem
• Solved the 2 machines-2 jobs, and 3 machines-3 jobs problem on the D-Wave Advantage 4.1 Quantum Computer, using the hybrid and quantum annealers
• Performed hyperparameter tuning for solution improvement and compared the results to simulated annealers.

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.