Hi, I'm
Dragoș Paraschiv

The idea came to mind as a next step after the course Control of a flexible robot system. I enjoyed controlling the robot so much, that I decided I wanted to work on all parts of it myself. Since there was no opportunity for me to do that in the University, I made my own robotic arm.

I started my design by researching how other people tackled this problem. After finding a couple of online resources, I was confident I could begin sketching a few prototype ideas. Moving through the design process, I eventually had a few parts ready for printing in Siemens NX.

Of course, there were multiple iterations and refinements along the way. This robot is still being continuusly improved, helping me learn how to transfer the theories I learned in my classes, into a real working robot.

This project was part of the Control of a Flexible Robot System course, where our team of about 8 students was given a real robot arm setup with a conveyor belt and camera system, and had to take it through a full control design cycle ourselves — from modeling to feedback design to final performance testing.

My main focus within the team was on the controller design and building out the Stateflow logic from scratch, since the pick-and-place sequence needed to reliably coordinate the robot's motion, the vacuum gripper, and the camera detection without any of it clashing. I spent a lot of time in Simulink tuning the feedback and feedforward controllers using FRF measurements, then iterating on the Stateflow chart to handle the different states of the task cleanly.

It was a good opportunity to take control theory from lectures and actually see it work on hardware, dealing with real delays, noise, and timing issues that don't show up in a simulation. By the end, we had the robot reliably picking objects off the moving conveyor and placing them at target coordinates.

This project was part of the Mechatronic Design course, where our team of six students was tasked with building a Coin Scanning Atomic Force Microscope (AFM). The goal was to create a safe, tabletop-compatible educational device for classroom demonstrations that could perform a large-scale demonstration of the AFM principle in contact tapping mode to image the 2.5D topography of a 1 Euro coin.

As the Magnetics Designer for the team, my main focus was designing and simulating the custom electromagnetic actuator needed for the precise Z-axis tapping motion. I designed an underhung voice coil actuator (NCC topology) to ensure a uniform force across the full stroke and a linear force-current relationship. The design utilized a 1018 Steel housing for its high permeability, an N45 NdFeB disc magnet, and a 32 AWG copper coil with 240 turns to keep the moving mass low. I spent a significant amount of time in FEMM simulating the magnetic fields, optimizing the geometry to achieve an average magnetic flux density of 0.172 T in the air gap and a Lorentz force of 0.208 N.

It was a great opportunity to apply electromagnetic theory to a real-world mechatronic setup, ensuring the actuator could operate safely within a 0.08 W power budget and avoid magnetic saturation. By the end of the design cycle, this voice coil actuator was fully integrated to drive the cantilever tip at a 43.4 Hz operating frequency, working in tandem with the XY macro stage and laser deflection sensors to reliably track and map the coin's surface.

Written for the Thermodynamics module, this model takes user-defined inlet temperatures, flow rates, and fluid properties, then calculates outlet temperatures and heat transfer rate using both the Log Mean Temperature Difference (LMTD) and effectiveness-NTU (ε-NTU) methods.

The script sweeps across a range of NTU values (0.1 to 5) and plots effectiveness curves for counter-flow and parallel-flow configurations on the same axes — making the performance advantage of counter-flow immediately visible.

A secondary parametric study varied the hot-side flow rate from 0.01 to 0.2 kg/s and tracked how outlet temperature and effectiveness changed, producing a surface plot useful for design selection.

The brief was to design a functional desktop object that could be manufactured from flat sheet stock and assembled without adhesives or fasteners — relying only on press-fit joinery.

I chose a modular desk organiser with interlocking slot joints. The key design constraint was parametric control: every slot, tab, and panel dimension is driven by two master variables — material thickness and grid pitch — so the whole model updates correctly when either changes.

The final version was cut from 3 mm MDF on the university laser cutter. Assembly took under five minutes with no tools. A follow-up iteration explored 4 mm birch plywood, which improved rigidity noticeably.

This simulation was built as a self-directed extension of the Engineering Dynamics module. The goal was to explore where the small-angle approximation (sin θ ≈ θ) breaks down and how drag changes the system's behaviour.

Using MATLAB's ode45 solver, I integrated the full nonlinear equation of motion for a damped pendulum across a range of initial angles (5° to 60°) and drag coefficients. The results are plotted as phase portraits and time-series overlays between the linear and nonlinear models.

An animation was added using MATLAB's built-in animation loop, showing the pendulum bob in real time alongside the live energy plot. At initial angles above ~20°, the approximation error becomes clearly visible.

The team of four was tasked with designing a gripper capable of picking and placing objects of irregular geometry — a cylinder, a hex bolt, a foam cube, a PET bottle, and a flat disc. The constraint was a single-acting pneumatic actuator at 4 bar supply pressure.

My contribution was the mechanical linkage connecting the actuator piston to the two finger bodies. I designed a symmetric toggle mechanism that converts the linear actuator stroke into a parallel jaw motion, maintaining even grip force regardless of object width within a 15–60 mm range.

I also produced the full GD&T drawing set for the finger bodies and pivot pins, specifying tolerances to ensure consistent assembly and repeatable grip force. The prototype was machined from aluminium and assembled in the university workshop.