Perception and navigation for mobile robots, with a focus on space robotics.
MSc student in Robotics & Autonomous Systems at Poznań University of Technology · previously Robotics Intern at ESA Spaceship Poland
I'm finishing an MSc in Robotics & Autonomous Systems at Poznań University of Technology. I work mostly on mobile robots: how they read the terrain in front of them and decide where they can actually go.
Most of that work has been hands-on. At ESA Spaceship Poland I spent six months teaching a Clearpath Jackal to judge which terrain it could safely drive on, and took it out to lunar-analogue grounds to see whether it actually could. At the Mobile Robots Laboratory I benchmarked NeRF and Gaussian Splatting for robotic perception, which turned into a paper.
A lot of what I've learned didn't come from coursework. At CybAiR, our student club, I was building a Mars rover for competitions, at first as part of the power and electrical systems team, later leading the Drone and Science teams and sitting on the club's board. Through ESRA I now do a version of the same thing across Europe.
Day to day that means ROS2, Gazebo, Python and C++, point clouds and fairly often a soldering iron. Space robotics is where I want to keep going.
Used the open-source elevation_mapping stack to build a Gridmap-based traversability estimation pipeline for a Clearpath Jackal robot, configured to run on its onboard computer: LiDAR and TF setup, point cloud filtering, and the traversability kernel retuned for analog lunar surfaces. Evaluated in field tests at the OTRK Kąkolewo space robotics facility.
Evaluated neural rendering methods (NeRF, Gaussian Splatting) for robotic perception and built an automated pipeline to benchmark them. The work became the PP RAI 2026 paper listed below.
Designed and assembled industrial control cabinets from electrical documentation and integrated a computer vision system for automated quality control on the production line.
Managing partnerships and outreach, connecting student robotics clubs across Europe and supporting cross-club initiatives.
Managing projects and teams inside the club and representing it externally. Co-organized the Warsaw Robotics Hackathon with GenR and I lead the club's European outreach through ESRA.
Estimating traversability cost for mobile robots from partially occluded LiDAR point clouds — reasoning about terrain the sensor can only partly see.
Model Predictive Control for flying a racing drone through a randomized sequence of gates as fast as possible. Built end to end in 48 hours.
PX4 drone that searches an area for probes, detects them and stores their world positions, then lands on an ArUco tag. Built against the ERC 2026 drone challenge rules. Full simulation stack in PX4 SITL and Gazebo, with ROS2 nodes for detection (ArUco + YOLO) and control, all Dockerized. Demo video in the repo.
RRT path planning on occupancy maps with gradient-based sampling guided by a neural network. ROS2, TensorFlow, Python.
Ongoing student research group project: an autonomous platform for soil quality testing and sample collection. I was responsible for the power and electrical integration systems.
Team project: a wearable "digital dog tag" for firefighters combining indoor 3D localization (floor-level, GPS-denied environments), vital-signs monitoring and critical alerts on a commander's interface. Wearable modules linked over LoRa to a GPS-RTK base station.
Trained a YOLOv8 model on custom fallen-tree data to detect road-blocking trees from dashcam footage in real time. Presented at the National Electronics Conference, Darłowo 2025, and published in Przegląd Elektrotechniczny (above).
Hackathons:
PythonC++ROS2DockerCUDA LinuxGitKUKA (KRL)ABB (RAPID)PCB design
Finishing my MSc and looking for what comes next, ideally perception or autonomy work on mobile robots. If that's what your team does, I'd be glad to talk.