RNDr. Petr Štěpán, Ph.D.

A novel approach for cooperative navigation and guidance of a micro-scale aerial vehicle by an accompanying Unmanned Aerial Vehicle (UAV) using 3D Light Detection and Ranging (LiDAR) relative localization is proposed in this paper. The use of 3D LiDARs represents a reliable way of environment perception and robust UAV self-localization in Global Navigation Satellite System (GNSS)-denied environments. However, 3D LiDARs are relatively heavy and they need to be carried by large UAV platforms. On the contrary, visual cameras are cheap, light-weight, and therefore ideal for small UAVs. However, visual self-localization methods suffer from loss of precision in texture-less environments, scale unobservability during certain maneuvers, and long-term drift with respect to the global frame of reference. Nevertheless, a micro-scale camera-equipped UAV is ideal for complementing a 3D LiDAR-equipped UAV as it can reach places inaccessible to a large UAV platform. To gain the advantages of both navigation approaches, we propose a cooperative navigation and guidance architecture utilizing a large LiDAR-equipped UAV accompanied by a small secondary UAV carrying a significantly lighter monocular camera. The primary UAV is localized by a robust LiDAR Simultaneous Localization and Mapping (SLAM) algorithm, while the secondary UAV utilizes a Visual-Inertial Odometry (VIO) approach with lower precision and reliability. The LiDAR data are used for markerless relative localization between the UAVs to enable precise guidance of the secondary UAV in the frame of reference of the LiDAR SLAM. The performance of the proposed approach has been extensively verified in simulations and real-world experiments with the algorithms running onboard the UAVs with no external localization infrastructure.

This paper presents a family of autonomous Unmanned Aerial Vehicles (UAVs) platforms designed for a diverse range of indoor and outdoor applications. The proposed UAV design is highly modular in terms of used actuators, sensor configurations, and even UAV frames. This allows to achieve, with minimal effort, a proper experimental setup for single, as well as, multi-robot scenarios. Presented platforms are intended to facilitate the transition from simulations, and simplified laboratory experiments, into the deployment of aerial robots into uncertain and hard-to-model real-world conditions. We present mechanical designs, electric configurations, and dynamic models of the UAVs, followed by numerous recommendations and technical details required for building such a fully autonomous UAV system for experimental verification of scientific achievements. To show strength and high variability of the proposed system, we present results of tens of completely different real-robot experiments in various environments using distinct actuator and sensory configurations.

This letter describes a method for autonomous aerial cinematography with distributed lighting by a team of unmanned aerial vehicles (UAVs). Although camera-carrying multi-rotor helicopters have become commonplace in cinematography, their usage is limited to scenarios with sufficient natural light or of lighting provided by static artificial lights. We propose to use a formation of unmanned aerial vehicles as a tool for filming a target under illumination from various directions, which is one of the fundamental techniques of traditional cinematography. We decompose the multi-UAV trajectory optimization problem to tackle non-linear cinematographic aspects and obstacle avoidance at separate stages, which allows us to re-plan in real time and react to changes in dynamic environments. The performance of our method has been evaluated in realistic simulation scenarios and field experiments, where we show how it increases the quality of the shots and that it is capable of planning safe trajectories even in cluttered environments.

This paper presents a novel approach to autonomous extinguishing of indoor fires inside a building by a Micro-scale Unmanned Aerial Vehicle (MAV). In particular, controlling and estimating the MAV state, detection of a building entrance, multi-modal MAV localization during the outdoor-indoor transition, interior motion planning and exploration, fire detection and position estimation, and fire extinguishing are discussed. The performance of these elements, as well as of the entire integrated system, are evaluated in simulations and field tests in various demanding real-world conditions. The system presented here is part of a complex multi-MAV solution that won the Mohamed Bin Zayed International Robotics Challenge 2020 (MBZIRC 2020) competition, and is being used as the core of a fire-fighting Unmanned Aerial System (UAS) industrial platform under development. A video attachment to this paper is available at the website http://mrs.felk.cvut.cz/2020firechallenge-insidefires.

We propose an approach enabling an Unmanned Aerial Vehicle (UAV) to autonomously enter a target building through an open window. We use a fusion of depth camera and 2D Light Detection and Ranging (LiDAR) data for window detection and continuous estimation of its position, orientation, and size. The proposed algorithms are capable of running both with and without available a priori information. The obtained detections are utilized for planning collision-free trajectories through the target window. We use a sensor fusion algorithm for robust altitude estimation from laser rangefinder data while flying over ground with inconsistent elevation. Particular focus is given to the transition between outdoor and indoor environments and vice-versa to achieve the required reliability of UAV state estimation. The proposed approach has been verified in multiple real-world experiments, where the UAV was able to successfully enter and leave the target building both under normal conditions and under decreased visibility conditions in a smoke-filled environment.

This paper addresses the perception, control, and trajectory planning for an aerial platform to identify and land on a moving car at 15 km/hr. The hexacopter unmanned aerial vehicle (UAV), equipped with onboard sensors and a computer, detects the car using a monocular camera and predicts the car future movement using a nonlinear motion model. While following the car, the UAV lands on its roof, and it attaches itself using magnetic legs. The proposed system is fully autonomous from takeoff to landing. Numerous field tests were conducted throughout the year‐long development and preparations for the Mohamed Bin Zayed International Robotics Challenge (MBZIRC) 2017 competition, for which the system was designed. We propose a novel control system in which a model predictive controller is used in real time to generate a reference trajectory for the UAV, which are then tracked by the nonlinear feedback controller. This combination allows to track predictions of the car motion with minimal position error. The evaluation presents three successful autonomous landings during the MBZIRC 2017, where our system achieved the fastest landing among all competing teams.

This article presents computer vision modules of a multi-unmanned aerial vehicle (UAV) system, which scored gold, silver, and bronze medals at the Mohamed bin Zayed International Robotics Challenge (MBZIRC) 2017. This autonomous system, which was running completely on-board and in real-time, had to address two complex tasks in challenging outdoor conditions. In the first task, an autonomous UAV had to find, track, and land on a human-driven car moving at $15$~$km/h$ on a figure-eight-shaped track. During the second task, a group of three UAVs had to find small colored objects in a wide area, pick them up, and deliver them into a specified drop-off zone. The computer vision modules presented here achieved computationally efficient detection, accurate localization, robust velocity estimation, and reliable future position prediction of both the colored objects and the car. These properties had to be achieved in adverse outdoor environments with changing light conditions. Lighting varied from intense direct sunlight with sharp shadows cast over the objects by the UAV itself, to reduced visibility caused by overcast to dust and sand in the air. The results presented in this paper demonstrate good performance of the modules both during testing, which took place in the harsh desert environment of the central area of United Arab Emirates, as well as during the contest, which took place at a racing complex in the urban, near-sea location of Abu Dhabi. The stability and reliability of these modules contributed to the overall result of the contest, where our multi-UAV system outperformed teams from world-leading robotic laboratories in two challenging scenarios.

This paper presents an implementation of a system that is autonomously able to find, follow and land on a car moving at \jed{15}{km/h}. Our solution consists of two parts, the image processing for fast onboard detection of landing platform and the MPC tracker for trajectory planning and control. This approach is fully autonomous using only the onboard computer and onboard sensors with differential GPS. Besides the description of the solution, we also present experimental results obtained at MBZIRC 2017 international competition.

The aim of this paper is to present a system being de- veloped for the Mohamed Bin Zayed International Robotics Challenge 2017 (MBZIRC) by a team of Czech Technical University in Prague, University of Pennsylvania and Univer- sity of Lincoln. The system designed for autonomous landing on a moving vehicle and autonomous collecting of color objects by a team of unmanned aerial vehicles - helicopters (UAVs) will be described with latest results achieved in real- scale outdoor scenarios. Both of these challenges require flying in high speed and strongly rely on vision control feedback and therefore the proposed system and designed approaches of autonomous flying with visual servoing should be within interest of participants of the Vision-based High Speed Autonomous Navigation of UAVs workshop. In our workshop presentation, we would like to introduce descrip- tion and results of computer vision approaches composed from a set of state-of-the-art techniques in a unique way to recognize reliably the landing pattern on the moving vehi- cle and color objects randomly distributed in a 100x100m workspace.

The field of reconfigurable swarms of modular robots has achieved a current status of performance that allows applications in diverse fields that are characterized by human support (e.g. exploratory and rescue tasks) or even in human-less environments. The main goal of the EC project REPLICATOR [1] is the development and deployment of a heterogeneous swarm of modular robots that are able to switch autonomously from a swarm of robots, into different organism forms, to reconfigure hese forms, and finally to revert to the original swarm mode [2]. To achieve these goals three different types of robot modules have been developed and an extensive suite of embodied distributed cognition methods implemented [3]. Hereby the methodological key aspects address principles of self-organization. In order to tackle our ambitious approach a Grand Challenge has been proposed of autonomous operation of 100 robots for 100 days (100 days, 100 robots). Moreover, a framework coined the SOS-cycle (SOS: Swarm-Organism-Swarm) is developed. It controls the transitions between internal phases that enable the whole system to alternate between different modes mentioned above. This paper describes the vision of the Grand Challenge and the implementation and the results of the different phases of the SOS-cycle.

The chapter introduces possibilities and principal approaches to knowledge gathering, preprocessing and keeping in autonomous mobile robots' artificial organisms. These may comprise "classical AI'' concepts as well as "new AI principles'', whereas both approaches themselves may bring up either major advantages, or suffer from certain drawbacks.

The paper deals with a problem of exploring an unknown environment by a team of mobile robots. Widely used approach in the robotic community is frontier based exploration. We combine this technique with a multi-agent architecture A-globe, which allows solving the exploration problem with limited communication accessibility and with changing number of participating robots. Frontier based approach heavily depends on path planning algorithm. A novel method combining A* search with harmonic potential fields is also presented. Finally, the Iterative Closest Point localization algorithm has been improved in several ways, in order to increase its speed and robustness. The whole exploration framework has been implemented and tested in both simulated and real environments.

The paper describes the SyRoTek ("System for robotic e-learning") project in general, its overall design, main hardware and software components and status of the project after one year of solution. The aim of the SyRoTek project is to research, design, and develop novel methods and approaches for building a multi-robot system for distance learning. The foreseen system will allow its remote users to get acquainted with algorithms from areas of modern mobile and collective robotics, artificial intelligence, control, and many other related domains. Advanced users will be able to develop own algorithms and monitor behavior of these algorithms on-line during real experiments. The proposed system reduces a development process and allows a wide spectrum of both individuals and institutions to work with real robotic equipment.

This paper presents architecture of system G-Bots for robotic soccer. Apart from our open architecture description, we focus on new approaches applied in reasoning and planning system components.

The main contribution of this paper is to present a sensor fusion approach to scene environment mapping as part of a Sensor Data Fusion (SDF) architecture. This approach involves combined sonar array with stereo vision readings. Sonar readings are interpreted using probability density functions to the occupied and empty regions. Scale Invariant Feature Transform (SIFT) feature descriptors are interpreted using gaussian probabilistic error models. The use of occupancy grids is proposed for representing the sensor readings. The Bayesian estimation approach is applied to update the sonar array and the SIFT descriptors' uncertainty grids. The sensor fusion yields a significant reduction in the uncertainty of the occupancy grid compared to the individual sensor readings.

This paper describes a technique for exploration of an unknown graph-like environment by a group of robots with the goal to construct a proper topological map. The key idea of the approach stands in simultaneous navigation through the environment using reactive planning similar to ``edge ant walk''. The exploration process is executed independently for each robot and these can exchange their individual maps exclusively whenever meeting each other at the same place. After the map exchange robots merge the maps using information on the edge type i.e. control strategy used to move along the edge. Neither markers to modify the environment nor any knowledge of robot's position in absolute metric coordinate system are required in this approach. Moreover, the approach allows to use only local short-range communication media between robots.

Accurate models of the environment are a crucial requirement for autonomous mobile robots. This paper first introduces a novel method for building an occupancy grid from a monocular color camera, describes a method for fusion of camera data with data from a rangefinder, and presents a new method for measuring the quality of the occupancy grid based on the quality of the path created by the grid.

Příspěvek popisuje novou metodu vytváření mřížky obsazenosti z monokulární kamery s automatickou kalibrací

The article presents the main achievement of the project about cooperation and coordination of multiple robots. At present students are using developed robot platforms for student's projects and some students are preparing his diploma thesis for the theme of coordination and cooperation of team of autonomous robots.

This paper describes advantages of application of artificial intelligence for telerobotics application. The methods of artificial intelligence enables mapping and self-navigation. In safe critical application the telerobotic approach is necessary because it is impossible to prove safeties of navigation methods. The combination of telerobotics and artificial intelligence methods can improve the control of such telerobotics systems. The second advantage of this approach is that the transfer of preprocessed data to teleoperator is easy and the amount of data is reduced.

Příspěvek se zabývá nasazením formálních metod při vývoji a testování aplikací s vysokými nároky na bezpečnost a spolehlivost. Nejprve je prezentován přehled metod model-checking pro důkaz správnosti návrhu systému. Na jejich základě je dále je navržen algoritmus pro automatizované generování testovacích sekvencí, který je popsán spolu s diskusí nad jeho výhodami i nevýhodami.

The process of how to acquire knowledge about the operating environment is one of the most challenging problems that autonomous mobile robots must solve. The quality of the model depends on a number and the kind of sensors used and upon the precision the robot recovers its position. This paper introduces two crucial components of the map building procedure: the position localization based on data gathered from laser rangefinders and the deadreckoning system together with a novel method for map-building through datafusion from a monocular camera and a laser range-finder.

This paper describes new method for building occupancy grid from single camera, automatic calibration of this method and approach how to use this internal representation for cleaning unknown environment. The paper brings also an overview of path-planning methods targeted on complete coverage of an operating space. Two algorithms for planning coverage based on occupancy grid are outlined and their fundamental properties are compared. Attached experiments verify and compare the newly designed and the existing methods as well as illustrate the final performance of the methods in real environments.

Prirozeny jazyk jako nastroj pro rizeni robota