Pouya P. Niaz

Studies

In this section, I have fitted some of the subjects and topics I have either studied in significant detail, taken courses in academia, or worked on projects relating to those fields, granting me knowledge and experience about these areas.

Robotics

I took this course in my master's, and my entire research focus throughout my time at Koc University, and afterwards, has been human-robot interaction and robot planning, which heavily involves coding and working with robots and robotics concepts. I can work with UR5 and KUKA robots, and I am also familiar with ROS. I have also worked with Gazebo and CoppeliaSim alongside it. That is me in the picture using the KUKA LBR iiwa 7 R800 to drill a hole in a marked position on a curved surface, with a custom non-perpendicular drilling angle, while wearing the Microsoft Hololens AR Goggle as an interface.

Machine Learning and Artificial Intelligence

I took a Machine Learning course in my M.Sc. studies, which included methods such as Artificial Neural Networks, K-Nearest Neighbor methods, Deep Learning, Decision Trees, Random Forests, Support Vector Machines, and some Unsupervised Learning methods as well. Feature selection and extraction methods were also included in the curriculum. I used ANN and LSTM models for time series regression and classification in my master's. I am also a co-instructor of the undergraduate-level Machine Learning course at the University of Innsbruck. At the moment, I am training and fine-tuning Large Language Models (LLMs) and Vision-Language Models (VLMs), as well as Computer Vision (CV) models for perception and affordance detection for efficient Task and Motion Planning (TAMP) in contact-rich tasks.

Computer Graphics

I have taken this course in M.Sc. at Koc University. The course used Shader-based OpenGL, which may not be the most popular graphics library nowadays, but it is one of the most basic ones, and learning it would make learning other graphics libraries much easier. I have also worked with the Coin3D library of the OpenInventor library, which is also OpenGL-based. I am right now working on a Unity project to make an application that makes visualizing the movement and control of robotic manipulators much easier and more intuitive.

Hydraulics and Pneumatics

I studied this subject in B.Sc. and engaged in many experiments, as well as making extensive use of it for a Pipe Inspection Robot project at the University of Tehran, in which almost all the movements are actuated hydraulically. The design included cylinders, valves, pumps, hydromotors and air motors, as well as an oil tank and a compressor.

Image Processing

My B.Sc. project involved heavy image processing in OpenCV. My job was to use a 600-FPS camera to calculate the size and frequency of bubbles resulting from static pool boiling of water in a porous medium. I used OpenCV in both MATLAB and Python for this task, and was highly successful at it. The goal of the project was to understand how the forming of bubbles in pool boiling in porous media was affected by the porosity and permeability of the medium itself.

Surface Electromyography

I have worked on surface Electromyography signals (sEMG) obtained from the forearm muscles of humans for intention recognition and muscle fatigue estimation of human operators in physical human-robot interaction. sEMG sensor technologies are advancing at a rapid rate. Wireless EMG products are already available on the market, and EMG equipment is more practical and cheaper than using a human force sensor in a robotic setup.

Modern Control Theory

The control engineering concepts that we studied in B.Sc., almost entirely focused on conventional classical control concepts such as transfer functions, Bode and Nyquist diagrams, root locus methods, and so forth. It was only a year after graduation that I noticed the fundamental importance of state-space analysis of dynamic and control systems, which is why I familiarized myself with it and partook in the Ball-on-beam project at the University of Tehran. I was intrigued by the topics I studied, such as designing, analyzing, and controlling LTI and LTV systems using state-space methods in MATLAB and Simulink.

MEMS/NEMS

I had heard of Micro and Nano-Electromechanical Systems after completing my B.Sc., but I never had the opportunity to study the concept. I found enough time to briefly study the topics of Micro and Nano-sensors, use of Electrostatic forces and Magnetic moments in actuation of these devices, Molding and Plating, Wet and Dry Etching, EDM and other Micro and Nanofabrication technologies used for making them, Vapor Deposition, Beam Lithography Methods, use of Piezoelectricity in MEMS devices, Modern Accelerometers, IMUs, MEMS Gyroscopes, Precision Resonators, Silicon Pressure Sensors, and so forth. I took the Mechanics of Microsensors class and passed it with an A+ in my master's program, where I used COMSOL to analyze a capacitive comb-drive-actuated strain gauge acting as a force sensor.

Fuzzy Control

One of the papers I have co-authored uses Fuzzy control for the adaptable part of a pipe inspection robot. We used trapezoid functions for motor current, robot velocity, and SMA voltages. Fuzzy control can be improved to be adaptive fuzzy control, or learning-based fuzzy control. Fuzzy logic also finds its way into machine learning and artificial intelligence. It has many advantages in comparison to other continuous-valued methods.

Page updated

Google Sites

Report abuse