Multi-robot Systems - Martin Saska, FEE CTU in Prague

Name: Multi-robot Systems - Martin Saska, FEE CTU in Prague
Address: Karlovo Náměstí 13, Praha, 12135, CZ
Telephone: +420224357634

Domů
Česko
Praha
Multi-robot Systems - Martin Saska, FEE CTU in Prague

Zavolat Trasa Web Kontaktovat společnost

Official page of the Multi-robot Systems (MRS) group at Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague.

20/10/2026

A drone that understands human speech? Our research is bringing autonomous flight closer to real-world use.
Developed by Fly4Future and the MRS Group, this fully autonomous drone can understand voice commands in natural language and navigate complex, unknown environments — without GPS or a human pilot.
🏆 The system reached the finals of Germany’s SPRIND Fully Autonomous Flight 2.0 competition and won one of the four challenges: finding a person in an unknown environment.
With applications ranging from search and rescue to logistics and delivery, the technology is now heading towards intensive testing, certification and commercial deployment.
The next step? Making autonomous drones faster, more capable and able to work together in teams to find missing or injured people.
🔗Learn more about the technology and its future applications here https://www.ceskenoviny.cz/zpravy/autonomni-dron-vyvinuty-fly4future-a-cvut-muze-pomahat-hasicum-i-zachranarum/2887725

16/10/2026

Our colleague Tomáš Musil presented his work on MonoSpheres, an approach to autonomous UAV exploration of large-scale 3D environments using only a single monocular camera and IMU. The method combines perception-aware mapping and planning to enable safe exploration of unknown indoor and outdoor environments without any LiDAR or GNSS.

📄Paper: https://ieeexplore.ieee.org/document/11517436

👾Code: https://github.com/ctu-mrs/monospheres

🎥Video: https://www.youtube.com/watch?v=hxYtRvwQrKc

Thanks to Tomáš for sharing his work with the faculty!

03/10/2026

🚁 From research to real-world autonomous drones

How can drones understand human instructions, navigate complex environments and work together as a fully autonomous swarm?
In a new interview with Forbes, Věra and Martin Saska talk about the development of autonomous flying robots at Fly4Future and the technologies behind them — from onboard AI and natural-language commands to fully autonomous drone swarms.

The interview also explores how these technologies can be used in real-world applications, including search and rescue, logistics and industrial warehouses, where autonomous drones can coordinate tasks and adapt to their environment without central human control.

The article also highlights the close connection between robotics research and practical applications, showing how technologies developed in research can make their way into real-world autonomous systems.
🔗 Read the full interview in Forbes:
https://forbes.cz/jako-hejno-ptaku-manzele-saskovi-stavi-plne-autonomni-roje-dronu/ Forbes Česko

30/09/2026

🚀 MRS researchers at IROS 2026 in Pittsburgh!

Yesterday, our researchers from the Multi-Robot Systems Group (MRS CTU) presented their latest research at IROS 2026 – the IEEE/RSJ International Conference on Intelligent Robots and Systems, one of the leading international conferences in robotics.

Michal Pliska, Pavel Pochobradský and Manuel Boldrer presented their papers focusing on autonomous UAVs, agile flight, multi-robot coordination, state estimation, and collision avoidance:

📌 “Distributed Lloyd-based algorithm for uncertainty-aware multi-robot under-canopy flocking”
Authors: Manuel Boldrer, Vít Krátký, Viktor Walter Martin Saska
A distributed approach enabling multiple drones to safely flock through complex environments such as dense forests, using only onboard sensing and computation. The method handles uncertainty, maintains safe distances between robots, and was validated in both simulations and real-world experiments with up to four drones.
🔗 Paper: https://arxiv.org/abs/2504.18840
🎬 Video: https://mrs.fel.cvut.cz/rbl
💻 Github: https://github.com/ctu-mrs/rbl

📌 “Geometric Model Predictive Path Integral for Agile UAV Control with Online Collision Avoidance”
Authors: Pavel Pochobradský, Ondřej Procházka, Robert Pěnička, Vojtěch Vonásek, Martin Saska
GMPPI combines accurate geometric UAV control with online obstacle avoidance using depth images directly in the control loop. The approach enables highly agile flight while maintaining accurate trajectory tracking and was demonstrated in real-world flights at speeds of up to 17 m/s.
🔗 Paper (IEEE): https://ieeexplore.ieee.org/abstract/document/11417433
🎬 Video: https://youtu.be/HEo4MQNX6xc
💻 Code (GitHub, open-source): https://github.com/ctu-mrs/gmppi

📌 “Towards Agile Vision-Based Multi-UAV Flight: Revisiting State Estimation”
Authors: Michal Pliska, Matouš Vrba, Ondřej Víta, Martin Jiroušek, Viktor Walter, Martin Saska
This work investigates how onboard vision can improve the estimation of neighboring UAVs during agile flight. By incorporating information about the UAVs’ orientation and thrust direction, the proposed approach significantly reduces velocity and acceleration estimation errors and enables more responsive multi-UAV tracking.
🔗 Paper: https://arxiv.org/pdf/2609.39611
🎬 Video: https://youtu.be/sR79rxmZ-Vk
💻 GitHub: https://mrs.fel.cvut.cz/agile-uav-estimation

👏 Congratulations to Manuel, Pavel, and Michal for representing MRS CTU at IROS 2026!

10/09/2026

🚁 No GPS. No pilot. Just tell the drone what to do.
What if you could simply tell a drone what you need — in natural language — and it would plan and execute the mission completely on its own?
That’s exactly what the F4F & CTU SkyGraph team demonstrated in the final of Germany’s SPRIND Fully Autonomous Flight 2.0 competition.

Competing against seven European technology teams, the Czech team developed a system that can:
🔹 understand instructions given in everyday language
🔹 plan a mission autonomously
🔹 navigate without GPS
🔹 understand and map its surroundings
🔹 find people and objects and adapt its actions to the situation

And the results? 🏆
The team delivered the best solution in the “find and help an injured person” challenge — and performed best among all finalists during the final demonstrations.

The technology combines four cameras, a 3D LiDAR, and onboard computing with advanced AI and autonomous planning. More importantly, it brings aerial robotics one step closer to a future where operating a complex robotic system doesn’t require a robotics expert.

👉 Read the full story: https://fel.cvut.cz/en/what-s-on/news/85031-a-drone-that-follows-verbal-commands-succeeds-in-germany-s-sprind-competition-developed-by-the-fly4future-and-ctu-fee-team

17/08/2026

IEEE RAS Summer School on Multi‑Robot Systems
📅 July 29 – August 4, 2026
📍 CTU in Prague, Czech Republic

One week, 250+ students, inspiring lectures, hands-on challenges, and plenty of 🤖 robotics! From learning from leading researchers to working together on real-world problems, this year’s Summer School brought together a truly international community🌍 passionate about multi-robot systems.

What we did:
📚 Intensive lectures from world‑class researchers
👥Topic-oriented practicals and networking in small groups
💻Programming sessions
🛠️ Hands‑on tutorials & team‑based projects
💡 Students presentations
🎯 Real‑world swarm‑robotics applications in action

Featured speaker
👨‍🏫 Prof. Oussama Khatib
👨‍🏫 Prof. Chuchu Fan
👨‍🏫 Prof. Marija Popovic
👨‍🏫 Jorge Peña Queralta
👨‍🏫 Fabrizio Schiano
👨‍🏫 Prof. Itzik Klein
👨‍🏫 Prof. Martin Saska

🌐 Over 250 students from leading universities gathered in Prague, diving deep into multi‑robot systems and sharing ideas across borders. Optional weekend trips gave participants an extra chance to explore Czech culture with visits to Karlštejn Castle🏰, the historic town of Kutná Hora⛪, or the spa city of Karlovy Vary🏺.

🙏 A huge thank you to everyone who made this year’s Summer School so special — our amazing participants, inspiring speakers, and dedicated organizers! 🙏
It was a week full of new ideas, new connections, and unforgettable experiences. We’re already looking forward to welcoming you all back in Prague next year! 🤖

13/07/2026

🌏 After ICRA 2026, our journey continues at Robotics: Science and Systems (RSS) 2026 in Sydney, one of the world's leading conferences dedicated to cutting-edge robotics research.

Throughout the conference and workshops, our colleague Filip Novák will represent the Multi-Robot Systems Group while also helping to organize the conference as a volunteer.

📅On Friday, July 17 (14:00–18:00), Filip will present the workshop paper "Perception and Planning Framework for Agile UAV Landing on Marine Vessels in Rough Seas" at the Planning and Control with Imperfect Sensors and Perception workshop. The work has been selected for both a spotlight talk and a poster presentation.
The paper addresses one of the key challenges in aerial robotics for marine environments: enabling UAVs to safely and reliably land on moving marine vessels despite rough sea conditions, where perception uncertainty and dynamic motion make autonomous landing particularly demanding.

📄Workshop paper: Perception and Planning Framework for Agile UAV Landing on Marine Vessels in Rough Seas (https://mrs.fel.cvut.cz/data/papers/2026_rss_workshop_novak.pdf)

✏️Authors: Filip Novák, Ondřej Procházka, Tomáš Báča, Parakh M. Gupta, Robert Pěnička, and Martin Saska.

If you're attending RSS 2026, we'd be delighted to meet you during the workshop and discuss our latest research.🚀

08/07/2026

Can robot swarms remain stable even when all robots lose localization?🤔

Dr. Jiří Horyna, a former PhD student and current researcher of our Multi-Robot Systems Group, has won the 1st Prize in the Joseph Fourier Prize for Computer Sciences 2026 of the Embassy of France in the Czech Republic🏅. The award, organized by the Embassy of France in the Czech Republic in partnership with DataSentics, recognizes his PhD research on decentralized state estimation and control for autonomous UAV swarms.

His work demonstrated that neighboring robots can collectively compensate for localization failures through mutual observations, enabling coordinated swarm operation even under partial or complete localization loss. The research provides new foundations for resilient multi-robot autonomy with potential applications in industrial inspection, emergency response, and security-critical environments.
Besides the financial award, Jiří also received a scholarship for a one-month research internship in France 🇫🇷.

This year's success means that our group has successfully defended last year's victory in the Computer Science category (Pavel Petráček). Together with second-place awards in 2021 (Robert Pěnička) and 2022 (Tomáš Báča), it reflects the consistently high quality of research conducted within the Multi-Robot Systems Group and its recognition by the scientific community.

📸Photos credit: Lukáš Bíba

30/06/2026

🚁 Autonomous drone exploration of large outdoor spaces using nothing but a single monocular camera, no LiDAR, no depth sensor, not even a GPU? Yes, it is possible!

Our new paper, MonoSpheres, was accepted to IEEE Robotics and Automation Letters (RA-L) and is fully open access. To the best of our knowledge, it is the first method to achieve 3D monocular exploration in real-world unstructured outdoor environments. 🎉

How does the proposed system work? Most exploration methods rely on heavy, expensive LiDARs or depth cameras. We use only a lightweight monocular camera 📷 and an IMU. Because motion-based depth estimates from a single moving camera are often sparse and uncertain, we design both the mapping and the planning around this limitation: the map oversamples free space in texture-poor areas so the UAV can safely cross open fields, while carefully tracking obstacle uncertainty so it doesn't wrongly delete a real obstacle from its map which could otherwise lead to crashes. In flight, the UAV always faces in its direction of motion and rapidly replans the instant new obstacles appear. The map is also represented by a graph of spheres instead of traditional occupancy grids, which allows rapid safety-aware planning.

We validated the full pipeline fully onboard in the real world, exploring an abandoned farmhouse 🏚️ and a cluttered orchard 🌳, and across 4 diverse simulated worlds, complete with ablation studies. The full implementation is open-source.

📄 Paper: https://ieeexplore.ieee.org/document/11517436
🎥 Video: https://youtu.be/hxYtRvwQrKc
💻 Code: https://github.com/ctu-mrs/monospheres

25/06/2026

Martin Saska se zúčastnil odborné diskuse "Technologická transformace obrany"
V panelu zazněla důležitá témata – od transformace nákupního procesu armády, přes nutnost budování strategického know-how v oblasti dronů a umělé inteligence, až po potřebu užší spolupráce mezi akademickou sférou, armádou a průmyslem.
Diskutovalo se o tom, jak klíčové je, aby armáda disponovala vlastním know-how Cílem je vybudovat strategické know-how v České republice v oblasti dronů a umělé inteligence, abychom měli k dispozici ta nejpokročilejší řešení, která pomohou ochránit náš stát. Vyvinout produkty, které tomu napomohou, je teď i hlavní cíl našeho spin-outu Fly4Future.
Děkujeme všem panelistům a moderátorům za podnětnou diskuzi! 👏

Adresa

Karlovo Náměstí 13
Praha
12135

Telefon

+420224357634

Web

http://mrs.fel.cvut.cz/, https://martinsaska.cz/

Upozornění

Buďte informováni jako první: pošleme vám e-mail, když Multi-robot Systems - Martin Saska, FEE CTU in Prague zveřejní novinky a akce. Vaši e-mailovou adresu nepoužijeme k žádnému jinému účelu a odběr můžete kdykoli zrušit.