This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
start [2020/11/23 08:52] IIS Webadmin |
start [2022/06/09 06:25] IIS Webadmin |
||
---|---|---|---|
Line 11: | Line 11: | ||
===== Working With Us ===== | ===== Working With Us ===== | ||
+ | * We are hiring [[jobs|two doctoral students]] in machine learning for computer vision or robotics. | ||
* Check our thesis topics for [[theses/start|Bachelor and Master students]]. | * Check our thesis topics for [[theses/start|Bachelor and Master students]]. | ||
- | * [[https://ellis.eu/|ELLIS]] PhD Program: [[https://ellis.eu/de/news/ellis-phd-program-call-for-applications|Call for Applications]] | + | <html><!-- |
+ | * [[https://ellis.eu/|ELLIS]] PhD Program: [[https://ellis.eu/news/ellis-phd-program-call-for-applications-deadline-november-15-2021|Call for Applications]] | ||
+ | --></html> | ||
<html><!-- | <html><!-- | ||
Line 29: | Line 32: | ||
===== News ===== | ===== News ===== | ||
- | <HTML><table><tr valign="top"><td>2020-11-20</td><td>Simon Haller-Seeber and Patrick Lamprecht present a show <i>Explainable AI: A sneak peek into the Black-Box</i> at <a href="https://youtube.com/watch?v=4NVfPwdLnCg&t=17m43s">Science Slam</a>, online.</td></tr><tr valign="top"><td>2020-06-22</td><td>Justus Piater gives an invited talk <i>Conditional Neural Movement Primitives</i> at <a href="http://www.gdr-isis.fr/index.php?page=reunion&idreunion=424">GdR | + | <HTML><table><tr valign="top"><td>2022-06-02</td><td>Erwan Renaudo contributes a talk <i>Deep Learning for Fast Segmentation of E-waste Devices’ |
- | ISIS Réunion Apprentissage et Robotique</a>, online. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem1', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem1Abstract" style="display:none">Conditional Neural Movement Primitives (CNMP) constitute | + | Inner Parts in a Recycling Scenario</i> at <a href="https://icprai2022.sciencesconf.org/">ICPRAI 2022</a>, Paris/online. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem0', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem0Abstract" style="display:none">Recycling obsolete electronic devices (E-waste) is a |
- | a novel framework for robot programming by demonstration based on | + | dangerous task for human workers. Automated E-waste recycling |
- | Conditional Neural Processes (CNP). Like Bayesian methods such as | + | is an area of great interest but challenging for current robotic |
- | Gaussian Processes (GP), CNP learn how target distributions depend | + | applications. We focus on the problem of segmenting inner parts |
- | on data, and can be conditioned on specific data points to infer | + | of E-waste devices into manipulable elements. First, we extend |
- | new target distributions at test time. Unlike GP that are | + | a dataset of hard-drive disk (HDD) components with labelled |
- | expensive to train and scale poorly to high dimensions, CNP are | + | occluded and non-occluded points of view of the parts, in order |
- | neural networks and are trained by gradient descent. CNMP | + | to increase the diversity and the quality of the learning data |
- | leverage CNP to represent motion trajectories that can be | + | with different angles. We then perform an extensive evaluation |
- | conditioned, at test time, on task paramters such as goal | + | with three different state-of-the-art models, namely CenterMask, |
- | locations, via points, and/or force readings. Moreover, CNMP are | + | BlendMask and SOLOv2 (including variants) and two types of |
- | conditioned on sensor readings during execution, resulting in | + | metrics: the average precision as well as the frame rate. Our |
- | robust, reactive behavior. This talk will present an overview of | + | results show that instance segmentation using state-of-the-art |
- | how CNMP work and how they can be used in various robot | + | deep learning methods can precisely detect complex shapes along |
- | applications.</blockquote></td></tr><tr valign="top"><td>2020-06-03</td><td>Justus Piater appears in the media: Wie der Roboter denken lernt.</td></tr><tr valign="top"><td>2020-01-29</td><td>Justus Piater gives an invited talk <i>Digital Science</i> at <a href="https://www.graduateacademy.uni-heidelberg.de/karriere/veranstaltungsreihen.html">Vortragsreihe | + | with their boundaries, as well as being suited for fast tracking of |
- | „Primers for Predocs – Strategien für eine erfolgreiche | + | parts in a robotic recycling system. (Rojas et al. 2022) |
- | Promotion“</a>, Universität Heidelberg. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem4', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem4Abstract" style="display:none">Massive availability of data and computing power are | + | </blockquote></td></tr><tr valign="top"><td>2022-05-20</td><td>Justus Piater appears in the media: <a href="https://radiothek.orf.at/podcasts/tir/derwoche/derwoche-maschinenzukunft">ORF |
- | promoting data-driven methods in all areas of science and | + | Tirol Podcast (in German)</a>. (Künstliche Intelligenz (KI) ist bereits heute ein täglicher Begleiter, doch vom universellen Haushaltsroboter sind wir noch weit entfernt. Automatisierte Systeme werden zwar Jobs übernehmen, Maschinen werden aber noch lange nicht die Weltherrschaft an sich reißen, falls überhaupt. Zur Langen Nacht der Forschung spricht Martin Kubin im #derWoche „Maschinenzukunft“ mit Prof. Justus Piater vom Institut für Informatik über Chancen und Gefahren von KI. Er weiß auch, warum KI etwa die Justiz unterstützen kann, aber ein KI-Richter niemals objektiv urteilen könnte.)</td></tr><tr valign="top"><td>2022-05-17</td><td>Erwan Renaudo gives an invited talk <i>A Brief tour of autonomous robots' "Cognition"</i> at <a href="https://www.philosophie.uni-konstanz.de/ag-mueller/aktuelles/meldungsdetails-ag-mueller/2022/5/17/event/46426-Creating-Agency-and-Cogn/tx_cal_phpicalendar/">Creating Agency and Cognition in Automated Systems: What can we learn from the Octopus? International Hybrid Workshop</a>, Innsbruck. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem2', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem2Abstract" style="display:none">In the design of "Cognitive Robots" software architectures, |
- | technology. I will describe how the University of Innsbruck | + | researchers had to face many organizational problems due to |
- | supports this via its new Digital Science Center, and will give a | + | the particular nature of robots. Not only should they manage |
- | flavor of machine learning for data analysis.</blockquote></td></tr><tr valign="top"><td>2020-01-20</td><td>Joanna Chimiak-Opoka, Carina König, and Justus Piater appear in the media: <a href="https://www.uibk.ac.at/newsroom/ergaenzung-digital-science-erfolgreich-gestartet.html.de">Ergänzung Digital Science erfolgreich gestartet – UIBK Newsroom</a>.</td></tr><tr valign="top"><td>2020-01-03</td><td>Justus Piater gives an invited talk <i>Künstliche Intelligenz: Grundlagen, Erfolge, | + | to achieve complex tasks with long-terms goals, but they must |
- | Herausforderungen</i> at 47. Tagung des Innsbrucker Kreises von MoraltheologInnen | + | also be reactive to their surroundings to preserve their environment. |
- | und SozialethikerInnen, Innsbruck.</td></tr><tr valign="top"><td>2019-12-19</td><td>Justus Piater appears in the media: <a href="https://tvthek.orf.at/profile/Tirol-heute/70023/Tirol-heute/14035670/Alexa-Siri-Co-Spione-im-eigenen-Haus/14610743">TV interview by ORF 2 Tirol Heute RedHaus (in German)</a>.</td></tr><tr valign="top"><td>2019-12-12</td><td>Justus Piater gives an invited lecture <i>Too Smart to Be Trusted – Do I Even Want to | + | We will go over several layered control paradigms developed over |
- | Understand My Robot?</i> at <a href="http://www.trustrobots.eu/">TrustRobots Lecture series | + | the history of robotics and see how these can relate to the |
- | Trust in Robots</a>, TU Vienna.</td></tr><tr valign="top"><td>2019-12-05</td><td>IIS guest <span style="font-weight:bold">Heiko Neumann</span>, University of Ulm, gives an invited colloquium <i>Biologically inspired visual-auditory processing – from | + | knowledge about the Octopus' brain organization, and discuss |
- | brain-like computation to neuromorphic algorithms</i> at <a href="https://www.uibk.ac.at/informatik/forschung/lunchtime-seminar/index.html.en">IFI Lunchtime Seminar</a>. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem9', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem9Abstract" style="display:none">A fundamental task of sensory processing is to detect | + | how the Octopus itself can inspire cognitive robotic research. |
- | and integrate feature items to group them into perceptual units | + | </blockquote></td></tr><tr valign="top"><td>2022-05-11</td><td>Justus Piater gives an invited talk <i>Robots That Learn Like Humans?</i> at <a href="https://pintofscience.at/events/innsbruck">Pint of |
- | segregating them from other objects and the background. A | + | Science</a>, Innsbruck. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem3', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem3Abstract" style="display:none">Wouldn't it be great to have a helper robot that we can |
- | framework is discussed which explains how perceptual grouping at | + | task with our annoying chores? How would we teach the robot to |
- | early as well as higher-level cognitive stages may be implemented | + | perform these tasks? Can we build such robots by equipping them |
- | in cortex. Different grouping mechanisms are implemented which are | + | with artificial intelligence? We will discuss how human and |
- | attuned to basic features and feature combinations and mainly | + | machine learning differ and how this impacts what skills, |
- | evaluated along the forward sweep of stimulus processing. However, | + | including for communication, our helper robot will be able to |
- | due to limitations of local feature detection mechanisms and | + | learn.</blockquote></td></tr><tr valign="top"><td>2022-04-28</td><td>Simon Haller-Seeber contributes a talk <i>STAIR Learning Lab</i> at <a href="http://rie.science">Robotics in Education 2022</a>, online.</td></tr><tr valign="top"><td>2022-02-19</td><td>Justus Piater gives an invited talk <i>Learning as a Creative Process in Humans and |
- | inherent ambiguities, top-down feedback is required to deliver | + | Machines</i> at <a href="https://www.philosophie.uni-konstanz.de/ag-mueller/online-workshop-on-agency-life-and-creativity/">Workshop |
- | contextual information helping to disambiguate initial | + | on “Agency, Life, and Creativity”</a>, online. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem5', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem5Abstract" style="display:none">Humans constantly engage in creative activity such as |
- | measurements. Feedback of contextual information is demonstrated | + | finding explanations, improving technology, or solving novel |
- | to improve object recognition performance, stabilize learning of | + | problems in everyday life. Even imitating other humans, e.g. while |
- | object categories, and integrate multi-sensory representations. | + | learning a new skill, is creative in that the imitator has to |
- | + | infer the conceptual structure underlying the observed | |
- | The canonical principles of neural computation define a set of | + | manifestation and to regenerate the latter from the former. In |
- | core operations to implement above-mentioned mechanisms of | + | contrast, current AI-enabled agents are unable to generate and |
- | perceptual and cognitive inference. These operations can be | + | draw upon such conceptual structures. Their learning is mostly |
- | mapped, in a simplified form, onto neuromorphic platforms to | + | driven by statistics, and any ability to produce novelty is |
- | emulate brain-like computation. It is demonstrated that an | + | largely limited to trial and error. I will try to characterize |
- | architecture composed of canonical circuit mechanisms can be | + | this fundamental difference between learning in human and |
- | mapped onto neuromorphic chip technology facilitating low-energy | + | artificial agents, and will discuss possible technological |
- | non-von Neumann computation.</blockquote></td></tr><tr valign="top"><td>2019-11-26</td><td>IIS guest <span style="font-weight:bold">Tamim Asfour</span>, Karlsruhe Institute of Technology, gives an invited keynote <i>Engineering Humanoids with Motion Intelligence</i> at <a href="https://www.uibk.ac.at/informatik/studium/inday-students/index.html.en">inday students</a>. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem10', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem10Abstract" style="display:none">Humanoid robotics plays a central role in robotics | + | approaches that might help close this gap.</blockquote></td></tr><tr valign="top"><td>2021-11-24</td><td>Justus Piater is a panelist at a public discussion on <a href="https://www.linkedin.com/posts/elsa-innsbruck_letzte-woche-fand-zum-elsa-day-am-24112021-activity-6871397038093869057-16-7">Privacy |
- | research as well as in understanding intelligence. Engineering | + | in the Digital Age</a>, online. (Organized by ELSA Innsbruck)</td></tr><tr valign="top"><td>2021-10-22</td><td>Justus Piater gives an invited talk <i>Robotik und KI in der Medizin – nur |
- | humanoid robots that are able to learn from humans and | + | Bedarfserweckung?</i> at <a href="https://medical-update-hall.com/veranstaltungen/">Medical |
- | sensorimotor experience, to predict the consequences of actions | + | Update Hall 2021 – Fit für das neue Jahrzehnt</a>, UMIT, Hall in Tirol.</td></tr><tr valign="top"><td>2021-08-26</td><td>Justus Piater contributes a talk <i>Digital Science at the University of Innsbruck</i> at Aurora Research Conference Digital Society & Global Citizenship, Vrije Universiteit Amsterdam. <span class="actions"><a href="javascript:void(0)" onclick="showHide('newsitem8', 'Abstract')">[Abstract]</a></span><blockquote id="newsitem8Abstract" style="display:none">Growing data processing capacities and progress in |
- | and exploit the interaction with the world to extend their | + | analytical methods and artificial intelligence are motivating |
- | cognitive horizon remains a research grand challenge. Currently, | + | entire branches of science to raise new questions and to develop |
- | we are experiencing AI systems with superhuman performance in | + | new approaches. This digital transformation requires progressive, |
- | games, image and speech processing. However, the generation of | + | interdisciplinary synergies between the computational sciences and |
- | robot behaviors with human-like motion intelligence and | + | all other scientific disciplines. The Digital Science Center |
- | performance has yet to be achieved. In this talk, I will present | + | (DiSC) at the University of Innsbruck meets this challenge by |
- | recent progress towards engineering 24/7 humanoid robots that link | + | integrating and promoting digitalisation of scientific research. |
- | perception and action to generate intelligent behavior. I will | + | In this talk I will present some examples of interdisciplinary |
- | show the ARMAR humanoid robots performing complex grasping and | + | research by DiSC staff researchers who are experts in specific |
- | manipulation tasks in kitchen and industrial environments, | + | aspects of digital science.</blockquote></td></tr><tr valign="top"><td>2021-08-24</td><td>Justus Piater teaches a tutorial <i>Picking and Placing</i> at <a href="https://www.uibk.ac.at/informatik/forschung/gmar-robotics-school-2021.html">2021 GMAR Summer School</a>, Villa Blanka, Innsbruck.</td></tr></table></HTML> |
- | learning actions from human observation and experience as well as | + | |
- | reasoning about object-action relations.</blockquote></td></tr></table></HTML> | + | |
[[news|Older News]] | [[news|Older News]] |