On the position of Computer Vision Research Engineer, I am investigating possible improvements of Hilti's construction solutions with respect to the computer vision and artificial intelligence aspects.

As an R&D Software Developer Intern, I implemented a calibration procedure for the camera-mount system. I was responsible for interfacing with available sensors, designing the system architecture, parallelizing the tests and integrating everything in Python. The internship's goal was also to learn about the clean software development, with a focus on design patterns and test automation tools.

6-month experience of working as a Junior Robotics Researcher in a start-up focused on developing a robotic hand with artificial hydraulic muscles. The future goal is to build the whole humanoid robot. My resposibilities were to implement Python software solutions, take care of the vision and create a simulation of the hand with MuJoCo physics engine. Check out the project's YT and website!

As a Test Engineer Intern in the Jabil factory, I took care of the testing machines, their maintenance, improvement and fixture inspection. I developed some testing programs in C#, optimized two ABB robots' operation and prepared different analysis in Minitab. All interns participated in workshops on soldering, AVR programming, TIA Portal and Lean Six Sigma with Kaizen.

In this collaboration with Microsoft Mixed Reality and AI Lab, and Computer Vision and Geometry Group at ETH, I worked on my "Master thesis. The aim was to make a Large Language Model reason about spacial properties of the scenes. To this end, I developed a RAG-based system, which by leveraging various modalities (such as images, texts, navigation meshes and SQL databases) was able to answer spatial questions. Additionally, I created a dataset of 1'000 spatial questions with answers, along with a benchmarking system, to assess the accuracy of the solution. The findings, along with the published paper and codebase, can be found on the project's webpage.

In this paper, we showed a comparative analysis of the underwater robot design steps using the example of the PWr Diving Crew project, where a progress from simplified Remotely Operated Vehicles (ROVs) to Autonomous Underwater Vehicles (AUVs) with enhanced environment perception and capability of autonomous task performance has been made. Robots' characteristics and the altered approach to the problem in successive vehicle generations were described and, based on the experience gained in the process, the conclusions on the structure optimization were drawn.

We explored the benefits of using a simulated environment for running software tests, training neural networks and testing the AUV's performance in arbitrary scenarios. We developed a customized Unity3D framework, allowing for setting up environments simulating various elements specific to underwater operation including buoyancy and caustics. This framework supports communication with AUVs systems (collecting observations through simulated sensors and controlling simulated actuators), collecting datasets for offline training and randomizing environment's elements to test the system's robustness.

There exist numerous devices that utilize laser triangulation and similar technologies like light detection and ranging, however they are characterised by high complexity and therefore price. We presented the design of a device intended to be mounted on Autonomous Underwater Vehicles and Remotely Operated Vehicles. Its principle of operation was simplified, allowing for relatively cheap creation, yet holding the feature of being applicable in the robots. This research was complemented with tests in above- and underwater environments, and surrounding elements dimensions and distance measurement.

During EPFL Excellence in Engineering program, I performed my research at CREATE lab. The objective of the project was to use the collaborative robot to perform experiments aimed at utilizing different optimization methods to search the system's parameter space for a resulting product with the best indices. The case-study was powdered cappuccino whose foam after preparation was evaluated with computer vision tools. The usage of computer vision driven feedback in closed loop control for hot drink foam quality improvement was also investigated. The results of this research were the basis for my BSc thesis, recognized by "The Young Innovative" national competition and awarded as faculty's "Best BSc thesis".

At LAMOR (University of Zagreb), we investigated the usage of User Datagram Protocol (UDP) for two-way communication between an Android mobile device and Robot Operating System (ROS). It enables interpreting user's desired movement choices into corresponding datagrams, which are then sent to the robot via wireless network. As a case study, we implemented a solution for controlling the Turtlebot3 differential drive mobile robot within ROS in both simulation and reallife environment. We complemented the paper with modular opensource implementations of our Android application and a ROS package featuring the Turtlebot3 controller. Check out our paper and the presentation video.

As Summer@EPFL was cancelled due to COVID, I remotely participated in research performed at LCAV laboratory by Frederike Duembgen. The objective was to create a ROS2-powered stack to do audio processing for sound-based indoor localization on robots (specifically tested with Crazyflie drone 2.1). My work was about designing a simulation pipeline which allows for the simulation of the acoustic response of a room as a robot equipped with microphones and a speaker moves within it. I worked out a way to integrate the pyroomacoustics room acoustic simulation framework with Robot Operating System 2 - I created a ROS2 package being a part of the research software audioROS.

The scientific circle's aim is to develop robots designed to work in the underwater environment. We focus on Autonomous Underwater Vehicles and Remotely Operated Vehicles engineering which can be used for exploration or lifesaving tasks. I started as a member of the Electronic Team, after a year I joined the Software Team. I was responsible for programming STM microcontrollers (e.g. UART or PWM management), soldering, PCB etching and other maintenance workshop. On the last year I became a Project Manager and my responsibilities were more research-realted. More on the project can be read in the Projects section.

As a member of my faculty's student council I was a volunteer, support, coorganiser and main organiser of a variety of events such as open days, engineers' ball, faculty's rally, billard tournament etc. I gained some experience in the council's PR unit, managing Facebook fanpage, Instagram account and the website. I was chosen to be my year's representant - I mediated between students and professors, managed resources, during the corona-semester additionally helped the teachers with making elearning more efficient and student-friendly. For a whole year I was also a student representant in the Curriculum Committee.

I was a part of the Aquatronik society and for a year and a half I worked in a team developing a mechatronic yacht. We wanted to construct a regatta training unit - a yacht packed with electronic devices and sensors to collect data while sailing. On this basis, conclusions on which techniques a sailer needs to improve and what errors are made would be drawn. We applied tensometers to measure the cap shrouds' tension and developed a tiller deviation sensor. My task was e.g. to design an anemometer - prepare a 3D model, print it and test a circuit based on the Hall sensor. Under this link you can see my first prototype of the anemometer.