Robotic Perception Group

Czech Technical University in Prague
Czech Institute of Informatics, Robotics, and Cybernetics
Robotics and Machine Perception Department

Seminars

Tue, Feb 18, 2025 (13:00 JP:B-633) Karla Štěpánová: Single-Grasp Deformable Object Discrimination: The Effect of Gripper Morphology, Sensing Modalities, and Action Parameters. An IEEE Transactions on Robotics, 2024 (link).

The regular ROP seminar time slot for the winter semester 2024/2025 is Tuesday 13:00 — 14:00.

The seminar organizer is Martin Matoušek.

Past seminars

  • Tue, Dec 10, 2024 (13:00 JP:B-633) Karla Štěpánová: Bridging Language, Vision and Action: Multimodal VAEs in Robotic Manipulation Tasks. An IROS 2024 conference paper (link).
  • Tue, Nov 26, 2024 (13:00 JP:B-633) Ondřej Holešovský: MovingCables: Moving Cable Segmentation Method and Dataset. A paper published in IEEE Robotics and Automation Letters (Vol: 9, Issue: 8, August 2024) (link)
  • Tue, Nov 19, 2024 (13:00 JP:B-633) Jan Kristof Behrens: CoBOS: Constraint-Based Online Scheduler for Human-Robot Collaboration. An IROS 2024 conference paper (link).
  • Tue, Nov 12, 2024 (13:30 JP:B-633) Pavel Studenovsky: Optimal trajectory for robotic spraying in a pipeline. The work was motivated by the robot produced by the Czech company Jetty Robot, (www.jettyrobot.com). The author helped the company as an advisor in the past. The presented work would allow higher-quality spraying inside bends of pipelines in both circular and square cross sections than the Jetty Robot solution. I will formulate the task, introduce the novel idea, sketch its derivation, and mention the advantages and disadvantages of the method.
  • Tue, Oct 22, 2024 (13:00 JP:B-633) Michal Vavrečka: Two papers from Artificial Neural Networks and Machine Learning (ICANN 2024) conference.
    Paper 1: Anastasia Ostapenko and Michal Vavrecka – Action recognition system integrating motion and object detection (link).
    Paper 2: Michal Vavrecka, Jonas Kriz, Nikita Sokovnin and Gabriela Sejnova – Modular Reinforcement Learning In Long-Horizon Manipulation Tasks (link).
  • Tue, Jan 9, 2024 (13:00 JP:B-633) Alena Smutná: Investigating Semi-Generalized Camera Pose Estimation – the research and life at TUAT in Japan. First I would like to talk about the research work I did during my stay at Tokyo University of Agriculture and Technology (TUAT) in Japan. I continued with the topic from my bachelor thesis, which is Investigation of visual localization methods based on the semi-generalized camera pose estimation. I tried to investigate if the accuracy of particular pose estimation methods is dependent on the geometry of the captured scene and used cameras. The discovered dependency could lead to changes in camera sampling strategies, which could help overall accuracy of the pose estimation. Afterwards, I would like to share a little bit of my experience of staying in Japan, the cultural differences and similarities and the life of a research student at TUAT.
  • Mon, Sept 4, 2023 (13:30 JP:B-633) Richárd Lengyel: Our ex-intern: Report on his further Master’s study and current project/product – An energy management software. In the Spring of 2018, I interned at CVUT CIIRC under the supervision of Prof. Václav Hlaváč, supported by Erasmus+, where I examined knowledge representation for autonomous cars using Bayesian Logic. Since then, I graduated in the Autonomous systems specialization of the EIT Digital Masterschool, studying my first year at TU Berlin and my second year at Aalto University, Helsinki. In the summer of 2020, I started working at our family-run company in Budapest, primarily an engineering office offering consultancy on building energetics. During the work with different organizations, it became clear that the lack of energy consumption data records often set a limit to the possible analysis. This inspired the software Panda, which we developed in the last three years and is the primary topic of today’s presentation. With the software, we gather the customers’ energy consumption data from different providers and try to show wasteful energy usage points in their portfolios. The entry into the Hungarian market was successful, and today, we have more than 40 customers: companies, local municipalities, universities, and hospitals. In the seminar, the main functionalities of the software will be explained and shown in a live demo using real consumption data. Further information about the software: https://www.pannonmuhely.hu/panda/en.php
  • Tue May 2, 2023 (13:00 JP:B-633) Karla Štěpánová: GACR proposal: Multimodal interactive dialogue for adaptive robotics. I will present the main objectives of my GACR proposal which was prepared in a close collaboration with Ivana Kruijff-Korbayova (DFKI Saarbrucken). The project aims to develop a multimodal interactive dialog system for human-robot interaction that will enable the robot to select the type of interaction with the human and environment, allowing him to effectively fulfill the given task and quickly adapt to a new environment.
  • Tue Mar 7, 2023 (13:00 JP:B-633) Miroslav Uller: Introduction to containerization and Docker tool.
  • Tue Feb 14, 2023 (13:00 JP:B-633) Michal Vavrečka: MyGym: Modular Toolkit for Visuomotor Robotic Task. We introduce myGym, a toolkit suitable for fast prototyping of neural networks in the area of robotic manipulation and navigation. Our toolbox is fully modular, enabling users to train their algorithms on different robots, environments, and tasks. We also include pretrained neural network modules for the real-time vision that allows training visuomotor tasks with sim2real transfer. The visual modules can be easily retrained using the dataset generation pipeline with domain augmentation and randomization. Moreover, myGym provides automatic evaluation methods and baselines that help the user to directly compare their trained model with the state-of-the-art algorithms. We additionally present a novel metric, called learnability, to compare the general learning capability of algorithms in different settings, where the complexity of the environment, robot, and the task is systematically manipulated. The learnability score tracks differences between the performance of algorithms in increasingly challenging setup conditions, and thus allows the user to compare different models in a more systematic fashion. The code is accessible at https://github.com/incognite-lab/myGym. See https://www.computer.org/csdl/proceedings-article/ictai/2021/089800a279/1zw6e4Y67dK.
  • Tue Dec 13, 2022 (13:00 JP:B-633) Richard Hájek: Discrete optimalization problem solvers in neural networks. Integrating discrete problems, such as logical reasoning,  within deep learning architectures has been a major goal of modern AI systems. This seminar will focus on recent approaches to model several discrete problems transparently in neural networks. This seminar will discuss modeling MAXSAT, Integer linear programming, TSP and some others.
  • Thu Nov 10, 2022 (13:00 JP:B-633) Karla Štěpánová: Talk and pictures from Japan and IROS 2022. : I spent 10 days in Japan – at IROS 2022 conference, some sightseeing, visiting TUAT (Tokyo University of agriculture and technology) in Tokyo and Ritsumeikan university near Kyoto where I visited 10 different robotics labs and also gave a talk about our research. I liked it a lot. I will show some general pictures from Japan and from the robotics labs of the visited universities. Then I will mention some papers presented at IROS that I found interesting.
  • Tue Oct 4, 2022 (13:00 JP:B-633) Petr Vanc: Context-aware gesture control of the robot. This seminar will focus on using hand gestures to control a robot and how context features, such as the type of manipulated objects, might be incorporated to build confidence. We will focus on our recent work on inferring the human intent from the observations of the scene and the hand movements. First, we will show the context of the overall Mirracle project, where gestures are one of the utilized modalities, next to the language and video data. Second, we will show how individual gestures are detected from hand movements. Third, we will describe how the mapping between the gestures and the intent can be learned in a data-driven approach. Finally, we will show how we generate a sequence of robot actions to fulfill the inferred intent. We are interested in discussing how the overall system might be evaluated as a whole and how individual parts might be improved. (ICRA recent submission)
  • Tue Sept 20, 2022 (15:00) Hijiri Akahane, TUAT Tokyo student: Single-rod brachiation robot. Brachiation is an aerial movement by alternately grasping and releasing branches and swinging the body. Brachiation robots are one of the ways to realize energy-efficient aerial mobile robots. There are many studies of brachiation done by multiple pendulum-like robots that mimic monkeys. They have disadvantages such as the unstable trajectory of the hand by chaos and complicated structure. So, we propose brachiation with a single-rod robot to solve those problems. It has the disadvantage that it’s difficult to change the moving distance per one brachiation, but this can be solved by including an aerial phase. In my research, the robot can increase its mechanical energy by center-of-gravity moving. I realized excitation by moving the center of gravity, continuous brachiation, and brachiation including the aerial phase by the actual machine. I`m now trying to do the model predictive control using the steepest descent method.
  • Thu Jun 23, 2022 (13:00) Kazuki Sekine, TUAT Tokyo student: A Single Motor Driving and Steering Mechanism for a Transformable Bicycle. In this research, I want to develop a self-driving bicycle steered and driven by a single motor that steers and drives by a single motor. First, I will attempt to transform the bicycle into a stable shape with a low center of gravity suitable for autonomous driving. I will use the mechanism of the folding bike to place wheels parallel to stabilize the robot. Then, I will rotate each wheel by a single motor using bevel gears and a one-way clutch and drive the robot using the differential drive, the major driving principle for two-wheeled robots.
  • Fri Jun 18, 2021 (13:00) [B670 and MS Teams] Jan Behrens and Karla Štěpánová: Embodied Reasoning for Discovering Object Properties via Manipulation (ICRA paper). In this paper we present an integrated system that includes a reasoning from visual and natural language inputs, action and motion planning, executing tasks by a robotic arm, manipulating objects and discovering their properties. The vision to action module recognises the scene with objects and their attributes and analyses enquiries formulated in natural language. It performs multi-modal reasoning and generates a sequence of simple actions that can be executed by the embodied agent. The scene model and action sequence are sent to the planning and execution module that generates motion plan with collision avoidance, simulates the actions as well as executes them by the embodied agent. We extensively use simulated data to train various components of the system which make it more robust to changes in the real environment thus generalise better. We focus on the tabletop scene with objects that can be grasped by our embodied agent, which is 7DoF manipulator with a two-finger gripper. We evaluate the agent on 60 representative queries repeated 3 times (e.g., ‘Check what is on the other side of the soda can’) concerning different objects and tasks in the scene. We perform experiments in simulated and real environment and report the success rate for various components of the system. Our system achieves up to 80.6% success rate on challenging scenes and queries. We also analyse and discuss the challenges that such intelligent embodied system faces.
  • Tue June 14, 2022 (13:00) Václav Hlaváč: Reading group – paper H. Kautz: The Third AI Summer. AI Magazine, Vol. 43, No 1, March 2022, pp. 93-104. Abstract: This article summarizes the author’s Robert S. Englemore Memorial Lecture presented at the Thirty-Fourth AAAI Conference on Artificial Intelligence on February 10, 2020. It explores recurring themes in the history of AI, real and imagined dangers from AI, and the future of the field.
  • Fri May 7, 2021 (13:00) on-line [MS Teams] Marko Sahan: Current Natural Language Processing, an overview. The Computational Linguistics subfield called Natural Language Processing has done a considerable leap in text classification, part-of-speech tagging, language modeling, machine translation, text mining, etc., over the past several decades. The research area has switched from language-based statistical approaches to language-independent deep neural networks with sometimes almost human-level context understanding. The essential purpose of the lecture is to go through the main parts of the NLP pipeline. We will start with the evolution of text encoding (bag-of-words, word2vec, sentence embeddings, etc.), followed by details of various NLP use cases and the state-the-art algorithms in language modeling and so-called transformers. The lecture aims to cover multiple modern approaches with high-level explanations of the theory and its algorithms.
  • Thu Apr 1, 2021 (10:00) Blahoslav Dolejší: Mastering our CNC devices. (presentation) The aim is to inform about ROP efforts to open our 3D manufacturing devices: (1) Průša 3D printers, two principles; (2) 40 W laser cutter/engraver; (3) 3D milling machine. It is one step towards the idea of open labs. The contribution might be three folds: (i) We follow the unified pipeline using open software; (ii) We invite other CIIRC groups to add their devices under the unified methodology/pipeline (e.g., fancy 3D printers in Testbed). We can improve the pipeline using incentives from you; (iii) We like to offer 3D design/manufacturing skills and capacities as a service.
  • Fri Mar 12, 2021 (13:00) on-line [MS Team link] Ondřej Holešovský – Experimental Comparison between Event and Global Shutter Cameras. Independent pixels of event cameras generate asynchronous events in response to local log intensity changes, in contrast to traditional frame cameras sampling image frames at regular intervals. In order to better understand the strengths and weaknesses of event cameras, we experimentally compared event and frame cameras on several high-speed computer vision tasks: observing flying bullets, tracking black dots and recognizing ArUco markers printed on a white rotating disk. Our results include pixel contrast sensitivity and latency, position estimation errors and sampling/detection rates. We will also share our experience with publishing the methodology and results in an open-access journal Sensors (MDPI publisher). Our paper is available at https://doi.org/10.3390/s21041137.
  • Tue September 29, 2020  (13:15) on-line [Join Zoom Meeting, Meeting ID: 878 0059 9814, Passcode: 246802] Matouš Vrba (Ph.D. intern at Multi-robot Systems Group, FEL) – Marker-less detection and localization of autonomous aerial vehicles. Abstract: Onboard marker-less detection of small flying objects is a topic that has lately been gaining increased attention. When compared to conventional marker-based solutions, it has advantages especially for the relative localization of Micro Aerial  Vehicles (MAVs), where removing the need for carrying markers frees up some of the limited payload capacity. In addition, such an approach also enables the detection and relative localization of non-cooperating aerial vehicles enabling the realization of autonomous aerial interception. However, a practical realization of marker-less detection methods and their deployment onboard MAVs with restrictions on computational resources presents novel challenges that need to be addressed. We present our current research in this area and some suggestions for its future development.
  • Tue September 8, 2020  (14:30) in JP:B-633: Karel Horák (BTU Brno, intern at CIIRC): Introduction to Karel Horák’s past work – The speaker will be our intern until the end of this year, coming to us every odd week. He will work with Vašek Hlaváč and Jan Kristof Behrens on high-level representation/reasoning in robotics, mainly for industrial manipulation tasks. The talk aims at telling us what Karel Horák worked on before in applied computer vision (traffic inspection, industrial inspection, biomedical applications). The demos will be shown too.
  • Tue November 19, 2019 (13:15): Gabriela Sejnova – Vision-Language Mapping: from Visual Question Answering to Multimodal Variational Autoencoders – The talk will first summarize our previous results in the Visual Question Answering task which were presented on the IROS conference two weeks ago. The second part of the talk will be about variational autoencoders focused on multimodal mapping and their possible implementation in a human-robot collaborative task.
  • Tue October 1, 2019 (13:15) in JP:B-633 Azumi Ueno (master student TUAT Tokyo, 3 months internship with V. Hlavac since Sep 16): The human-mimetic robot for reproducing human-to-human physical contact. Abstract: My research purpose is the development of the robot reproducing human-to-human physical contact. My focus is “how humans feel humanity with touch”. So, my prototype artificial hand/arm has human-like shape, fleshiness, temperature, elastic joints, and so on. Additionally, it has finger flection ability and wrist abduction as an active touch function. To realize above, I made the robot using urethane resin, silicone rubber, living hinge joint, and so on.
  • Wed September 25, 2019 (16:00) Igor Farkas, JP:B-670 – Bio-inspired task learning in cognitive robotics – I will briefly introduce our Cognition and Neural Computation group whose research in general focuses on using neural networks and reinforcement learning to solving various tasks mostly related to cognitive robotics. I will report about two recent works: a computational model of enactive visuospatial mental imagery, and a connectionist model of self-organized emergence of perspective in/variant mirror neurons in a robotic system.
  • Tue September 24, 2019 (13:15) David Estevez Fernandez – Towards a humanoid robot household companion – In this seminar we will present the main research lines of the RoboticsLab at University Charles III of Madrid, including our work towards the development of a humanoid robot household companion capable of doing laundry-related tasks such as unfolding, hanging or ironing clothes.
  • Tue August 13, 2019 (13:15) Felix von Drigalski, OMRON SINIC X: Sensing and planning in robotic assembly and manipulation – a reality (Abstract and speaker biography; Host Vaclav Hlavac)
  • Tue May 7, 2019 (13:15): RMP department meeting in JP:B-670
  • Tue May 7, 2019 (13:45): V. Smutny – Manipulator Robotics – Introduction, Practical Issues – The seminar will introduce participants into basic practical issues when using robot as a tool, not as an object of research. The seminar will cover basic terminology, forward and inverse kinematics of serial manipulators, some accuracy aspects, path and trajectory planning and executing.Robotics course pages
  • Tue April 30, 2019: V. Petrik: One-year postdoc at Aalto University – In the seminar, I will present the work I have done during my 1yr postdoc stay in Intelligent Robotics group lead by Ville Kyrki. The presentation will start with a brief description of AISpider project which I was hired for. It will continue with the description of static stability of fabric strip folding and with the application of reinforcement learning to fabric strip folding.
  • Tue April 23, 2019: V. Smutny, M. Matousek, P. Krsek: Cameras, principle, practical issues, calibration – 2nd part– The seminar is intended for people using or intending to use cameras although they are not familiar with the computer vision, photography, etc. Practical issues like choosing the camera and lens, manipulating them etc. will be discussed by Vladimir Smutny. Mathematical model, understanding camera parameters and their calibration will be presented by Martin Matousek.
  • Tue April 16, 2019: Kristyna Malinovska, Elena Sikudova, Julia Skovierova, Radoslav Skoviera, Miroslav Uller: UP-Drive WP6: Scenario Understanding – In the scope of project UP-Drive we evaluate and predict the motion and intention to cross the street for pedestrians. We use several approaches including Kalman filter, Bayesian networks and other ML methods. Our prediction is solely based on the location of the pedestrians over time.
  • Tue April 9, 2019: RMP department meeting in JP:B-670
  • Tue March 26, 2019: Ondrej Holesovsky: Towards event cameras improving interactive robotic manipulation-Event cameras provide low latency and high temporal resolution visual sensing. Therefore we wish to utilize them for robotic manipulation. The seminar is going to cover the research topic in three parts: 1) Introduction to event cameras. 2) Our RA-L/IROS 2019 submission, Real-Time Optical Flow for Event Cameras Sped up by One-Dimensional Block-Matching. 3) Motion segmentation: problem definition, use cases and methods. The miniseminar slides are available here.
  • Tue March 19, 2019: V. Smutny, M. Matousek, P. Krsek: Cameras, principle, practical issues, calibration– The seminar is intended for people using or intending to use cameras although they are not familiar with the computer vision, photography, etc. Practical issues like choosing the camera and lens, manipulating them etc. will be discussed by Vladimir Smutny. Mathematical model, understanding camera parameters and their calibration will be presented by Martin Matousek. * Tue March 12, 2019: RMP department meeting in JP:B-670
  • Tue March 5, 2019: Karla Stepanova, Gabriela Sejnova, Radoslav Skoviera: How to extract important information from language data – In this seminar, we will show our recent results in the area of language processing (both for Czech and English). We will show our results in two showcases. First, we will show how an abstract task description can be created from linguistic input. This abstract plan can be later used for robot task and motion planning. The second showcase is the automated creation of testing scenarios for production lines. We would like to show the methods and approaches we tested as well as their limitations. We also hope that this could motivate a discussion about alternative approaches. Presentation: Prezentace 1,Prezentace 2
  • Tue Feb 5, 2019: RMP department meeting in JP:B-670
  • Tue Jan 15, 2019 Matúš Tuna, FMFI UK Bratislava: Few-shot learning in artificial neural networks: Artificial neural networks require many training examples for learning a certain task. Few-shot learning approach deals with the problem of training artificial neural networks with only a few examples. In this lecture we will present an overview of various approaches to the few-shot learning problem and briefly consider extending this approach to problems beyond image classification.
  • Fri Jan 11, 2019 10:15 JP:B-633 Guillem Alenya, Universitat Polytechnica de Catalunya“ Challenges in robotics for human environments and specially for textile manipulation
  • Fri Jan 11, 2019 9:30 JP:B-633 Kimistoshi Yamazaki: Vision-Based Cloth Manipulation by Autonomous Robots
  • Thu Jan 10, 2019 14:00 JP:B-670 Vladimir Petrik, Ph.D. defense
  • Tue Jan 8, 2019: Vladimir Smutny: Classification of concrete blocks defects from images: The overview and lessons learned from industrial inspection feasibility study.
  • Tue Dec 18, 2018: Yaroslava Lochman: Quantitative Analysis of Rectification Accuracy on Real Images and Symmetry Detections from a Single Image Using CNNs: The first part of the talk will present the first dataset and protocol for measuring rectification accuracy of imaged scene planes from a single image. Rectification results presented in the state of the art are either qualitative or evaluated on synthetic scenes. The second part of the talk will present ongoing research for the dense segmentation of symmetries from a single image using CNNs.
  • Tue Dec 18, 2018: Jiri Sedlar: Estimating 3D Motion and Forces of Person-Object Interactions from Monocular Video: We introduce a new method for automatic reconstruction of the 3D motion of a person interacting with an object from a single RGB video. Our method estimates 3D poses of the person and the object, contact positions and forces, and torques
  • Tue Dec 11, 2018: Jan Behrens: A Constraint-Programming Approach to Task and Motion Scheduling for Industrial Dual-Arm Robots: Modern lightweight dual-arm robots bring the physical capabilities to quickly take over tasks at typical industrial workplaces designed for workers. In times of mass-customization, low setup times including the instructing/specifying of new tasks are crucial to stay competitive. Programming dual-arm robots is for itself a very complex task, because the execution time (makespan) shall be minimized while robot-robot collisions must be avoided. The descriptive nature of Constraint Satisfaction Problems (CSP) together with the flexibility of Constraint Programming (CP) allows the formulation of diverse problems. Our constraint programming approach to simultaneous task allocation and motion scheduling for industrial manipulation and assembly tasks leverages this expressiveness to formulate this hybrid planning problem in a coherent formalism. The proposed approach covers dual-arm and even multi-arm robots as well as connected machines. The key concept is Ordered Visiting Constraints, a descriptive and extendable model to specify such tasks with their spatiotemporal requirements and task-specific combinatorial or ordering constraints. Our solver integrates such task models and robot motion models into constraint optimization problems and solves them efficiently using various heuristics to produce makespan-optimized robot programs. The proposed task description model is robot independent and thus can easily be deployed to other robotic platforms. This also greatly reduces the complexity in the specification and thus allows to employ Natural Language (NL) and demonstration as input. We show that we achieve solutions of higher quality than other approaches. The flexibility and portability of our proposed model is validated through several experiments on different simulated robot platforms. For large manipulation tasks with 200 objects, our solver implemented using Google’s Operations Research tools and ROS requires less than a minute to compute usable plans.
  • Tue Dec 4, 2018: RMP department meeting in JP:B-670
  • Tue November 20, 2018, Megumi Miyashita: Robot arm control with force compliant and reinforcement learning: I will report the progress of my project in CTU. The goal is that a robot arm covers trouser with the object by using force compliant control and reinforcement learning. I will talk about the ｔhe state so far and future plans.
  • Tue November 6, 2018, Kristina Malinovska: Towards more biologically plausible error-driven learning: Since the standard error backpropagation algorithm for supervised learning in artificial neural networks was shown biologically implausible, alternative models of training that use only local activation variables have been proposed. In this talk I will present a novel Universal Bidirectional Activation-based Learning algorithm for standard perceptron networks based on contrastive Hebbian-anti-Hebbian principle with special parameters allowing the model to perform a spectrum of tasks including heteroassociation, denoising, and classification. I will briefly introduce the background, explain the model, and report on the algorithm’s performance in selected tasks.
  • Tue October 30, 2018 Meeting of RMP CIIRC department in JP:B-670.
  • actuated by the human limbs.
  • Tue October 16, 2018, Jan Stria: Convolutional Neural Networks for Processing of 3D Point Clouds: I will introduce the idea of applying convolutional neural networks on the unstructured clouds of 3D points. Our own contribution of generalizing convolution operation over a local neighborhood of point will be described. I will tell you about the application of the method on the classification of hanging garments which is a topic of our recently published IROS paper. I will also mention the extensions that have been made since the paper submission.
  • Mon October 8, 2018, Petr Svarny: How to get good with Git{lab}, Slides here: https://sites.google.com/site/svarnypetr/programming
  • Mon September 24, 2018, Mircea Cimpoi: From Indoor Localization to Indoor Navigation: I will present our CVPR18 spotlight, InLoc (https://arxiv.org/abs/1803.10368), cover our NIPS18 spotlight Neighbourhood Consensus Networks and the work with UCU interns this summer on MINOS (https://arxiv.org/pdf/1712.03931.pdf) environmnent, and briefly present the Gibson environment, introduced at CVPR18: https://storage.googleapis.com/gibson_material/Gibson_CVPR2018.pdf. Slides available at: https://drive.google.com/open?id=1iLPR6FYGC_9iqTUrlGBy3CkBgkMhQZB9. You need to be logged in to CIIRC in order to access them. Please do not redistribute.
  • Mon September 17, 2018, Michael Tesař Neural representation of dynamic scenes
  • Mon September 10, 2018, Gabriela Šejnová Overview of Visual question answering architectures and their possible modifications
  • Mon September 3, 2018 Meeting of RMP CIIRC department in JP:B-670.
  • Mon June 25, 2018 Daniel Seifert Grab software for a high-speed camera
  • Mon June 11, 2018 Gabriela Sejnova, Michael Tesar, Michal Vavrecka: Visual question answering (including demo with Pepper robot)
  • Mon June 4, 2018 Matěj Hoffmann & Petr Švarný: Taking part in (and winning) the KUKA Innovation Award and CoBots at Hanover Fair 2018 : We will describe the successful participation of the IIT Co-Aware team in the KUKA Innovation Award 2018: Real-World Interaction Challenge. Then we will share what we learned at the Hanover Fair in particular in the area of collaborative robots (a major theme of the whole Fair). Finally, we will present our plans on how to use the KUKA LBR iiwa robots @ CIIRC. KUKA Award links: https://www.kuka.com/en-de/technologies/research-and-innovation/kuka-innovation-award, video: https://youtu.be/XVGfBgOhaqw
  • Mon May 14, 2018 Ondrej Holesovsky: Impedance control of redundant manipulators : Several ROP people have been trying to design and implement a suitable controller for our force-compliant KUKA robots for some time. Impedance control is one possible candidate for the sought controller. Relevant ideas of two papers will be presented, namely “N. Hogan, Impedance Control: An Approach to Manipulation, 1985” and “C. Ott et. al., Prioritized multi-task compliance control of redundant manipulators, 2015”. The latter paper proposes a compliance controller, which was developed for the DLR Justin robot, tested on a 7-DOF arm and is likely integrated also in the KUKA IIWA robot controllers we have in our lab. slides
  • Wed May 9, 2018 Meeting of RMP CIIRC department in JP:B-670. (Moved to 18.6.)
  • Mon April 23, 2018 Miniseminar discussing research plans in two years horizon (2nd part).
  • Mon April 16, 2018 Miniseminar discussing research plans in two years horizon (as agreed at the March 16, 2018 reading group). All researchers including Ph.D. students are expected to prepare the entry in Google Doc document.
  • Mon April 9, 2018 Meeting of RMP CIIRC department in JP:B-670.
  • Tue April 3, 2018 in JP:B-633 at 10:30. Meeting was cancelled because the speaker could not come. It will be given in the furture. Vojtěch Kabelka, student of the 1st year at master level ETH Zürich, Department of Mechanical and Process Engineering study program Robotics, Systems & Control (with focus on intelligent robotics). The aim is to share with us a student view after finishing one semester of the study. He studied FEL ČVUT for his bachelor degree.
  • Fri March 16, 2018 11 AM, Reading group, Grand Challenges of Science Robotics: every seminar participant reads this article before the seminar. VH: If you did not read the paper, do not come to the seminar. There will be given the intro about the article content in 10 minutes by Michal Vavrecka together with his own position to the article in another 10 minutes. The rest will be discussion. article for the reading group * Mon March 12, 2018 Radoslav Skoviera: highlights from the meeting Cognitively inspired perception based explainable AI in London on Feb 8, 2018. slides
  • Mon March 5, 2018 Meeting of RMP CIIRC department in JP:B-670.
  • Mon Feb 19, 2018 Vladimír Petrik: KUKA robot: The mini-seminar is divided into two parts: the quick overview of Capek testbed and the programming tutorial. The overview explains the hardware connection of the individual testbed parts and provides the basics information about the software infrastructure. The initial draft of presentation covering the first part of the mini-seminar is attached. In the second part of the seminar, we will compile the tools required to control the robots. We will create the code for arm controlling in C++, and we will test it in simulation and on real HW. You will need the ROS Kinetic with MoveIt installed on your laptop in order to follow the tutorial part of the seminar. Presentation for miniseminar
  • Mon Feb 12, 2018 Meeting of RMP CIIRC department in JP:B-670.
  • Mon Dec 11, 2017 Karla Štěpánová: EuCog 2017 “Learning: Beyond Deep Neural Networks” Highlights from the EuCog meeting 2017 which spawn a debate on the topic of machine learning technique in “embodied” (i.e. robotic) cognitive systems. A special focus of this conference is on limitations of deep learning in the area of cognitive robotics. The limitations in terms of time, energy, and computing power, characteristic of autonomous robotic systems that can support humans in their daily life and in hazardous environments, limit the application of the computations-, data-, and energy-“hungry” DNNs in this field..
  • Mon Dec 4, 2017 Meeting of RMP CIIRC department in JP:B-670.
  • Mon Nov 27, 2017 Václav Hlaváč: Basic Skills to be implemented on Collaborative Robots. We have two force-compliant KUKA arms. There is a repertoire of basic skills, on which the robot-human collaboration relies. Such robots are around for several years. Other research groups demonstrated needed skills already. I will review these skills and illustrate them by videos. Other ROP people have some skills and applications in mind too. The talk should initiate a discussion/coordination leading to the list of skills we will implement.
  • Mon Nov 20, 2017 Radoslav Škoviera: ROS Intro. This hands-on seminar will give a brief introduction to the basics of working with ROS (robot operating system) – a framework which provides libraries and tools to help software developers create robot applications. The seminar is intended only for those who have never worked with ROS before and is not compulsory. Please take your laptop with you, if you would like to join in with some practical examples. Your laptop should run Linux Ubuntu 14.04 LTS, either natively (better), inside a virtual machine, or as a bootable image, with preinstalled ROS: http://wiki.ros.org/ROS/Installation (at least version Indigo), Python (2.7 or 3.0), and catkin_tools (http://catkin-tools.readthedocs.io/en/latest/installing.html). Be aware that this mini-seminar will be longer than the usual 1 hour. Slides: ros_intro.pdf
  • Mon Nov 6, 2017 Meeting of RMP CIIRC department in JP:B-670.
  • Mon Oct 23, 2017 No miniseminar
  • Mon Oct 16, 2017 Martin Pecka, Jan Stria and Vlado Petrík: Co se urodilo na IROSU
  • Mon Oct 2, 2017 Meeting of RMP CIIRC department in JP:B-670.
  • Mon Sept 25, 2017 Gábina Šejnová: Neuronal noise and its influence on signal and information transfer in axons and Michael Tesař: Brain electrophysiology in the context of developmental robotics
  • Mon 2017-09-11 V. Hlaváč: Meeting of RMP CIIRC department Moved one week ahead from the regular first Monday of a month because of the V. Hlavac’s business trip to Setkani kateder in Banska Stiavnica on Sept 4, 2017. The meeting will be in a bigger room JP:B-670.
  • Mon 2017-08-24, extraordinary at 10:30 in JP:B-633 Lenka Mudrová: Task scheduling and merging in space and time, Ph.D. student na U of Birmingham, about her Ph.D. research. Abstract: Every day, robots are being deployed in more challenging environments, where they are required to perform complex tasks. In order to achieve these tasks, robots rely on intelligent deliberation algorithms. In my thesis, I study two deliberation approaches – task scheduling and task planning. I extend these approaches in order to deal with temporal and spatial constraints imposed by the environment.
  • Thu 2017-07-27 13h Valentýn Číhala: Představení robota KUKA iiwa
  • Mon 2017-07-24 J. Škovierová: ACM Chapters Computer Science in Cars Symposium CSCS 2017 and BMW Summer School Report
  • Mon 2017-07-10 V. Hlaváč: DANTE, a submitted H2020 proposal following CloPeMa, RadioRoSo projects. DANTE acronym stands for Dual-arm cognitive robotic manipulation with soft objects. This € 3.12 mil. proposal coordinated by V. Hlaváčwas submitted on April 25, 2017. First, I will briefly describe the overall proposal. Second, I will explain the tasks I foresee for our ČVUT team in DANTE, which also corresponds to our (my) vision for the research program in dual-arm manipulation for the next future.
  • Mon 2017-07-03 No miniseminar.
  • Mon 2017-06-26 V. Hlaváč: Meeting of RMP CIIRC department (https://internal.ciirc.cvut.cz/wiki/research_depts/rmp_dept/rmp_meetings/m2017_06_26) *
  • Mon 2017-06-19 No miniseminar.
  • Mon 2017-06-12 K. Štěpánová: Hierarchical probabilistic model of language acquisition (PhD. defense training)
  • Mon 2017-06-05 Students, 30 min each: Rehearsal before bachelor and master defenses. Extraordinary time 9:00-15:00. Time schedule is in Google Sheet.
  • (extraordinary miniseminar) Mon 2017-05-30 13:00 JP:B-633 Michal Vavrečka: Japanese humanoid robotics (experience from my research visit): I will present recent projects in humanoid robotics at Japanese universities seen during my research visit. I will briefly describe software and hardware parts of these projects.
  • Mon 2017-05-29 V. Hlaváč: CoHoRT, a submitted H2020 proposal following TRADR project. CoHoRT acronym stands for Collaboration in Human-Robot Teams for Long-Term Deployment in Situation Assessment Tasks. This € 4 mil. proposal with 10 partners was submitted on April 25, 2017. First, I will briefly describe the overall proposal. Second, I will explain the tasks I foresee for our ČVUT team in CoHoRT, which also corresponds to our (my) vision for research program for the next future. We will be responsible for WP2: Active perception and environment understanding. (Presentation )
  • Mon 2017-05-22 D. Seifert: Radiopharmacy 4.0 Fundamentals of radiopharmacy and radiopharmaceuticals. Future of radiopharmacetuticals production
  • Mon 2017-04-10 M. Pecka and V. Kubelka: Comparison of tracked robot simulation models & Localization using HTC Vive
  • Mon 2017-03-27 R. Škoviera: Fire detection for TRADR <font 10pt/inherit;;#0000FF;;white>The simple fire detector that will be presented is used in the TRADR project. It uses thermal camera to detect possible fires. The key issue is grouping and localizing multiple fire detections.</font>
  • Mon 2017-03-20 No miniseminar organized
  • Mon 2017-03-13 J. Škovierová: UP-Drive: Scenario UnderstandingThe UP-Drive (automated Urban Parking and Driving) project’s endeavor is to create a car capable of self-driving in an unconstrained urban environment with speeds up to 30 km/h. The main purpose of the scenario understanding module is to extract contextual information about other traffic participants, relevant to the particular traffic situation, from the environment. Currently, we are focusing on intention estimation of other traffic participants using Bayesian networks.
  • Mon 2017-02-20 M. Vavrečka: TransferNet network for semantic segmentation Abstract: I will present the analysis of Transfernet – deep network for semantic segmentation with attentional module. This network allows transfer learning (pixel-wise annotated source dataset helps to learn novel classes from image-level class labels). Mon 2017-02-13 K. Zimmermann, L. Wagner et al. Lab task with TurtleBot in Autonomous Robotics class.
  • Mon 2017-02-06 No seminar organized. Many people will be at CVWW 2017 in Austria.
  • Mon 2017-01-30 No seminar organized.
  • Mon 2017-01-23 Marko Prelevikj: (a visiting student): Texture-based segmentation with CNNs. Abstract: CNNs are a very powerful tool for processing images, and there are many applications. One of them is extracting image features from segments and using them to classify objects in images. This presentation will explain how the CNNs can be used to extract features from images. Another focus of the presentation will be transfer of knowledge, what it is, and how it is applied in this particular case. Work has connection to the project TRADR and is mentored by R. Škoviera.
  • Mon 2017-01-16 No minisemiar organized.
  • Mon 2017-01-09 Karla Štěpánová: Estimating mixture of Gaussian models.