Burak Brooke Karabillioglu — Mechanical Engineer

Selected projects

P.01 · CAPSTONEAutonomous Pool Skimmer Robot

P.02 · ROBOTICSKUKA KR 4 Programming & Vision Pick-and-Place

P.03 · FEATruss & Frame FEM with Custom MATLAB Solvers

P.04 · ADDITIVE MFGCAD-to-STL Tessellation Study

P.05 · MECHATRONICSAutonomous Robotic Wheelchair

P.06 · MACHININGCNC Milling & Metal Cutting

P.01 — CAPSTONE · 2025–2026 · TEAM OF 5

Autonomous Pool Skimmer Robot

End-to-end mechanical design, simulation, fabrication & testing

A compact, solar-assisted surface robot that autonomously skims floating debris from residential pools. Over two terms the team took it from concept sketches and a QFD through CFD hull optimization, FEA of every printed part, full engineering drawings, and a waterproofed prototype tested in water.

The hull was refined over seven CAD versions; each was run through ANSYS CFD at 0.5 m/s with an 0.08 m waterline, and the final hull reached a drag coefficient of 0.121 — about 19% lower than the original concept. All PLA components (hull, upper housing, paddle wheels, motor mounts, basket) were checked in SolidWorks Simulation using slicer-derived 15%-infill material properties.

SolidWorksANSYS CFDSolidWorks SimulationMATLABFDM printingArduinoGD&T drawings

PropulsionTwin paddle wheels on 6 V geared DC motors, differential steering
Sensing3× HC-SR04 ultrasonic, MPU-6050 IMU, HC-06 Bluetooth
PowerTriple 18650 pack with solar-assisted charging, sealed compartment
BuildFDM-printed PLA, acrylic-sealed; ≈ CAD $306 in materials
ModellingMATLAB paddle-thrust, buoyancy and odometry simulations

0.121Final hull drag coefficient

7Hull versions simulated in CFD

0.53 m/sTop measured speed in water

2 hrsIn water with a dry enclosure (< 1% mass gain)

Completed pool skimmer prototype with solar panels — **FIG 1.1**Completed prototype — solar panels on the sealed lid

SolidWorks exploded assembly — **FIG 1.2**Exploded assembly — SolidWorks

Hull drag coefficient by CAD version

ANSYS CFD · 0.5 m/s · 0.08 m waterline

Seven hull body models compared — **FIG 1.3**Hull models 0–6 compared in CFD

ANSYS CFD of the chassis — **FIG 1.4**CFD flow around the hull — ANSYS

FEA of hull body — **FIG 1.5**FEA of the PLA hull body

FEA of debris basket frame — **FIG 1.6**FEA of the debris basket frame

Exploded assembly drawing with BOM — **FIG 1.7**Assembly drawing & BOM

Hull body engineering drawing — **FIG 1.8**Hull body drawing

Paddle wheel engineering drawing — **FIG 1.9**Paddle wheel drawing

Hull in slicer with supports — **FIG 1.10**Hull print prep — slicer

Hull under static load — **FIG 1.11**Buoyancy load test

Electronics layout inside the enclosure — **FIG 1.12**Internal layout

Debris collection test in water — **FIG 1.13**Debris collection test

Forward velocity vs motor power

Measured over 2.75 m in water — near-linear response

Testing also showed ultrasonic ranging within 2–6% error over 10–60 cm, and reliable obstacle avoidance once the detection threshold was set at 22 cm or more.

P.02 — ROBOTICS & AUTOMATION · FALL 2025

KUKA KR 4 Programming & Vision-Guided Pick-and-Place

Industrial robot programming in KRL on a KUKA KR 4 R600 AGILUS

A series of lab projects programming a six-axis industrial arm. One program drives a pen through planned Cartesian waypoints using inverse-displacement calls to draw initials, with pen-up/pen-down heights and halts on unreachable poses. Another integrates the VISOR 2D vision system: the camera locates randomly placed blocks, and the robot uses the returned pose to approach from a safe height, grip, and place each part into fixed pallet positions with spline and linear motions.

KUKA KRLRobot visionInverse kinematicsPath planningWorkcell safety

RobotKUKA KR 4 R600 AGILUS, six-axis
VisionVISOR 2D camera, calibrated to the robot frame
MotionSpline (SPL) and linear (SLIN) moves, 250 mm safe approach height
TeamPair project, shared workcell

KUKA robot in workcell — **FIG 2.1**KR 4 in the workcell

Planned drawing path with coordinates — **FIG 2.2**Planned path & coordinates

Initials drawn by the robot — **FIG 2.3**Result drawn by the robot

KUKA pendant showing vision calibration — **FIG 2.4**Vision calibration on the pendant

P.03 — FINITE ELEMENT ANALYSIS · FALL 2023

Truss & Frame FEM with Custom MATLAB Solvers

Multi-element beam & frame modelling

Built and validated multi-bar and beam 3D space frames two ways: in SolidWorks Simulation, and with MATLAB solvers written from scratch — assembling global stiffness matrices to solve for nodal displacements and reactions, refining meshes with the H-method, and cross-checking results between the two.

MATLABSolidWorks SimulationStiffness methodH-method refinement

Max displacement9.09 mm (SolidWorks beam model)
Max stress75.7 MPa upper-bound axial + bending vs 710 MPa yield
SolversCustom MATLAB truss solver and H-method refinement solver

SolidWorks displacement plot — **FIG 3.1**SolidWorks displacement study

SolidWorks stress plot — **FIG 3.2**SolidWorks stress study

MATLAB H-method plot — **FIG 3.3**Custom MATLAB H-method displacement solver

MATLAB truss plots — **FIG 3.4**Custom MATLAB truss displacement solver

P.04 — ADDITIVE MANUFACTURING · WINTER 2024

CAD-to-STL Tessellation Study

How mesh resolution settings affect print-ready files

Modelled three parts of increasing complexity — a primitive pyramid, a three-part hinge mechanism sized for a 25 cm³ print envelope, and a freeform seashell — with full engineering drawings. Each was exported to STL at ten deviation/angle settings, and file size, triangle count, node count and generation time were compared to understand tessellation accuracy versus file weight.

As a bonus, I wrote a MATLAB parser that reads binary STL files byte-by-byte (normals and vertices), rebuilds the mesh, and checks the stored facet normals against recalculated ones.

SolidWorksSTL / tessellationMATLABEngineering drawings

FindingFile size scales with triangle count alone — facet size limits only matter through it
RangePyramid: 6 → 1,752 triangles across ten exports
CodeBinary STL reader + normal-vector verification in MATLAB

Hinge CAD model — **FIG 4.1**Hinge mechanism

Hinge engineering drawing — **FIG 4.2**Hinge part drawing

Freeform seashell CAD — **FIG 4.3**Freeform seashell

STL export resolution settings — **FIG 4.4**STL deviation & angle settings

P.05 — MECHATRONICS · WINTER 2023 · TEAM OF 5

Autonomous Robotic Wheelchair

LEGO EV3 prototype programmed in Python

A scaled autonomous wheelchair that had to navigate a maze without touching the walls, find three points of interest, and climb a ramp. After two prototypes proved too wide and unstable, the design moved to a compact vertical layout with a lowered ultrasonic sensor, a centred gyro, and a seat with a locking seatbelt.

The control code (Pybricks MicroPython) uses a gyro-based proportional controller to hold a straight line, a servo-swept ultrasonic sensor to check for open corridors, and a light sensor to stop at each point of interest.

PythonP-controlSensor integrationIterative prototyping

SensorsGyro, ultrasonic on a rotating mount, light sensor
ControlProportional gyro correction for straight-line driving
IterationsThree design versions before the final build

Wheelchair robot with labelled components — **FIG 5.1**Final design, labelled

Front view of final design — **FIG 5.2**Front view

Multiview of design — **FIG 5.3**Multiview

P.06 — MATERIAL REMOVAL PROCESSES · WINTER 2024

CNC Milling & Metal Cutting

Hands-on machining, CAM and tooling knowledge

Shop time milling aluminium parts on CNC equipment, alongside Fusion 360 CAM toolpath work. To go deeper on cutting tools, I completed the Sandvik Coromant Academy Metal Cutting Technology programme and ten Mitsubishi Materials machining courses covering drilling, turning, face and end milling, tool grades and workpiece materials.

CNC millingFusion 360 CAMCutting dataTool selection

Burak Brooke
Karabillioglu

Selected projects

Autonomous Pool Skimmer Robot

Hull drag coefficient by CAD version

Forward velocity vs motor power

KUKA KR 4 Programming & Vision-Guided Pick-and-Place

Truss & Frame FEM with Custom MATLAB Solvers

CAD-to-STL Tessellation Study

Autonomous Robotic Wheelchair

CNC Milling & Metal Cutting

Skills & credentials

Design & CAD

Analysis & code

Hands-on

Sandvik Coromant Academy

Mitsubishi Materials

Professional

Let's talk