IAg Logo

FIT ČVUT

INTELLIGENT AGENTS RESEARCHGROUP


Ozobots EVO


The group focuses on various aspects of artificial intelligence with a special regard on intelligent decision making in single and multi-agent scenarios. Interests of members of the group include core symbolic search algorithms for multi-agent path finding (MAPF) and its generalizations towards real-life applications, automated planning for robotic agents, multi-robot planning with focus on coordination, motion planning and related problem solving in robotics by the use of logical reasoning such as propositional satisfiability (SAT), satisfiability modulo theories (SMT) and constraint satisfaction (CSP). The integration of non-local (global) reasoning and consideration of adversarial behaviour represent on-going challenges of the group. The fundamental philosophy of our research is to promote explainability and theoretical guarantees (time/space complexity, completeness, or reasoning proofs) together with every concept being developed.


The research of the group fully supports and is integrated with teaching activities within the bachelor and the master study programs of Artificial Intelligence and Knowledge Engineering at the Faculty of Information Technology (FIT), ČVUT. We provide supervision of all types of projects and theses in some of the themes from the still growing list of topics for inspiration.



Members

Head of the group:

    

Doc. RNDr. Pavel Surynek, Ph.D.

Members:

    

Bc. Zuzana Fílová

     lazy compilation in planning

Bc. Martin Čapek

     DPLL(MAPF)

Bc. Kristián Kuĺka

     game theory in multi-Pacman

Bc. Petr Nešpůrek

     learning of agents in Pac-man

Bc. Klára Dvořáková

     warehouse logistics

Bc. Ondrej Marek

     aerial drone display

Bc. Ján Chudý

     planning for real robots

Bc. Jindřich Kuzma

     traffic simulation

Martin Rameš

     compilation for construction in Minecraft

Jan Czakoj

     machine learning in SAT

Petr Michalíček

     cable-connected robots

Ladislav Miklík

     motion planning for intersections

Lukáš Kameník

     MAPF in multi-Pacman

Silvestr Láník

     localization for multi-robot system

Marek Nevole

     variants of Sudoku

Patrik Schweika

     discrete traffic simulation

Tomáš Valenta

     robotic soccer

Vít Šprachta

     multi-robot construction in Minecraft

Kristýna Janovská

     agent-based modeling in evacuation

Petr Šindlář

     target assignment in multi-Pacman


Research Grants

19-17966S: intALG-MAPFg: Intelligent Algorithms for Generalized Variants of Multi-agent Pathfinding, GAČR - Czech Science Foundation, standard project, 2019-2021.

SGS17/210/OHK3/3T/18: Modern data-mining methods for advanced extraction of information from data, CTU Project, 2017-2019.



Academic Cooperation

Kobe University, Japan (artificial intelligence, robot navigation, distributed CSP), BGU - Ben Gurion University of the Negev, Israel (MAPF, heuristic search), USC - University of Southern California, USA (MAPF, robotics), Waseda University, Japan (graph theory), AIRC/AIST - Artificial Intelligence Resarch Center/National Institute of Advanced Industrial Science and Technology, Tokyo, Japan (robotics, artificial intelligence), NII - National Institute of Informatics, Tokyo, Japan, AIP/Riken, Tokyo, Japan, IBM Resarch, Ireland.