IRIM Seminar Series

Series Type
Event Series
Associated Organization(s)
Associated Organization(s)

Publication Search Results

Now showing 1 - 10 of 134
  • Item
    Preparing for the Coming Machine Revolution
    ( 2017-11-08) D'Andrea, Raffaello ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machines ; Swiss Federal Institute of Technology (ETH)
    Spanning academics, business, and the arts, Raffaello D’Andrea’s career is built on his ability to bridge theory and practice. At the Swiss Federal Institute of Technology (ETH) in Zurich, his research redefines the capabilities of autonomous systems. D’Andrea is a co-founder of Kiva Systems (acquired by Amazon in 2012, and now operating as Amazon Robotics), a robotics and logistics company that develops and deploys intelligent automated warehouse systems, with over 100,000 autonomous mobile robots deployed in Amazon warehouses alone. D’Andrea was the faculty advisor and system architect of the Cornell Robot Soccer Team, four-time world champions at the international RoboCup competition. With his startup, Verity Studios, he recently created the flying machine design and choreography for Cirque du Soleil’s Paramour on Broadway, Zhang Yimou’s 2047 Apologue, and Metallica’s WorldWired tour. Additionally, D’Andrea is a new media artist with exhibitions at various international venues, including the Venice Biennale, the FRAC Centre and the National Gallery of Canada. Other creations and projects include the Flying Machine Arena, the Distributed Flight Array, the Balancing Cube, Cubli, Flight Assembled Architecture, the Blind Juggler, the Robotic Chair, and RoboEarth. D’Andrea’s TED talks, viewed more than 10 million times, have inspired a generation to pursue engineering, robotics, and computer science.
  • Item
    Physics-Based Manipulation With and Around People
    (Georgia Institute of Technology, 2018-01-31) Srinivasa, Siddhartha ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machines ; University of Washington. School of Computer Science and Engineering
    Robots manipulate with super-human speed and dexterity on factory floors, but they fail even under moderate amounts of clutter or uncertainty. However, human teleoperators perform remarkable acts of manipulation with the same hardware. My research goal is to bridge the gap between what robotic manipulators can do now and what they are capable of doing. What human operators intuitively possess that robots lack are models of interaction between the manipulator and the world that goes beyond pick-and-place maneuvers. I will describe our work on nonprehensile physics-based manipulation that has produced simple but effective models, integrated with proprioception and perception, enabling robots to fearlessly push, pull, and slide objects, and reconfigure clutter that comes in the way of their primary task. Human environments are also filled with humans. Collaborative manipulation is a dance, demanding the sharing of intentions, inferences, and forces between the robot and the human. I will also describe our work on the mathematics of human-robot interaction that has produced a framework for collaboration using Bayesian inference to model the human collaborator, and trajectory optimization to generate fluent collaborative plans. Finally, I will talk about our new initiative for assistive care that focuses on marrying physics, human-robot collaboration, control theory, and rehabilitation engineering to build and deploy caregiving systems.
  • Item
    Dynamic Animation and Robotics Toolkit
    ( 2014-11-12) Liu, Karen ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machine ; Georgia Institute of Technology. School of Interactive Computing
    Designing control algorithms for complex dynamic systems is a challenging and time consuming process. It often requires deriving nonlinear differential equations, formulating optimization problems, and solving numerous small, but tedious, problems, such as inverse kinematics, forward simulation, inverse dynamics, or Jacobian matrix computation. To streamline the process of controller design, we introduced an open-source, cross-platform toolkit, called DART, for rapid development of kinematics and dynamics applications in computer animation and robotics. DART (Dynamic Animation and Robotics Toolkit), one of the default physics engines in Gazebo, provides seamless integration with robotic simulators in the ROS environment. In contrast to many popular physics engines that view the simulator as a black box, DART gives full access to internal kinematic and dynamic quantities, such as the mass matrix, Coriolis and centrifugal forces, and transformation matrices and their derivatives. DART also provides efficient computation of Jacobian matrices for arbitrary body points and coordinate frames. In this talk, I will give an introduction to DART and demonstrate how complicated problems can be implemented using only a few lines of code in DART. I will also show how we use DART to make an Atlas robot walk, a Shadow Hand manipulate objects, a virtual human learn gymnastics, and a variety of aquatic creatures swim in simulated fluid.
  • Item
    Ocean Research and Robotics
    (Georgia Institute of Technology, 2013-01-23) Moroni, Dan ; Rubsamen, Amy ; Georgia Institute of Technology. Center for Robotics and Intelligent Machines
    The presentation will go through the details of the Wave Glider and how Liquid Robotics (LRI) is working to build a platform to enable others to do ocean research without the high costs and risks of conducting ocean operations. Moroni will cover some of the challenges LRI has encountered building a persistent ocean robot and convincing some of the world's largest industries to shift how they do ocean operations. Amy Rubsamen will talk about the PacX mission and the record breaking trans-Atlantic crossing.
  • Item
    Efficient Lifelong Machine Learning
    ( 2015-02-11) Eaton, Eric R. ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machine ; University of Pennsylvania. Dept. of Computer and Information Science
    Lifelong learning is a key characteristic of human intelligence, largely responsible for the variety and complexity of our behavior. This process allows us to rapidly learn new skills by building upon and continually refining our learned knowledge over a lifetime of experience. Incorporating these abilities into machine learning algorithms remains a mostly unsolved problem, but one that is essential for the development of versatile autonomous systems. In this talk, I will present our recent progress in developing algorithms for lifelong machine learning. These algorithms acquire knowledge incrementally over consecutive learning tasks, and then transfer that knowledge to rapidly learn to solve new problems. Our approach is highly efficient, scaling to large numbers of tasks and amounts of data, and provides a variety of theoretical guarantees on performance and convergence. I will show that our lifelong learning system achieves state-of-the-art results in multi-task learning for classification and regression on a variety of domains, including facial expression recognition, landmine detection, and student examination score prediction. I will also describe how lifelong learning can be applied to sequential decision making for robotics, demonstrating accelerated learning for optimal control on several dynamical systems, including an application to quadrotor control. Finally, I will discuss our work toward autonomous cross-domain transfer, enabling knowledge to be automatically transferred between different task domains.
  • Item
    Turning Assistive Machines into Assistive Robots
    ( 2015-01-21) Argall, Brenna D. ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machine ; Northwestern University
    For decades, the potential for automation—in particular, in the form of smart wheelchairs—to aid those with motor or cognitive impairments, has been recognized. It is a paradox that often the more severe a person's motor impairment, the more challenging it is for them to operate the very assistive machines that might enhance their quality of life. A primary aim of my lab is to address this confound by incorporating robotics autonomy and intelligence into assistive machines—turning the machine into a kind of robot and offloading some of the control burden from the user. Robots already synthetically sense, act in and reason about the world, and these technologies can be leveraged to help bridge the gap left by sensory, motor, or cognitive impairments of the users of assistive machines. This talk will provide an overview of some of the ongoing projects in my lab, which strives to advance human ability through robotics autonomy.
  • Item
    Robots for Physical Interaction
    (Georgia Institute of Technology, 2019-01-23) Kim, Sangbae ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machines ; Massachusetts Institute of Technology. Department of Mechanical Engineering
    While robots dominate repetitive works in factories, the design and control of these robots are not suitable for relatively complex tasks that humans do easily. These tasks typically require force sensing and interaction force control. Conventional robots are not built to control force or to be flexible to perform like human arms.
 This talk will discuss how the new design paradigm allows dynamic interactive force control within environments. As an embodiment of such a robot design paradigm, the latest version of the Cheetah robot and force-feedback teleoperation arms will be presented. This new class of robots will play a crucial role in future robot applications such as elderly care, home service, delivery, and services in environments unfavorable for humans.
  • Item
    Weakly Supervised Learning from Images and Video
    ( 2016-09-30) Laptev, Ivan ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machine ; INRIA
    Recent progress in visual recognition goes hand-in-hand with the supervised learning and large-scale training data. While the amount of existing images and videos is huge, their detailed annotation is expensive and often ambiguous. To address these problems, in this talk we will focus on weakly-supervised learning methods using incomplete and noisy supervision for training. In the first part I will discuss recognition from still images and will describe our work on weakly-supervised convolutional networks for recognizing and localizing objects and human actions. The second part of the talk will focus on the learning of human actions from videos. In particular we will consider understanding specific tasks from YouTube instruction videos and corresponding narrations. We will conclude with future challenges and opportunities of visual recognition.
  • Item
    Soft Miniature Mobile Robots
    (Georgia Institute of Technology, 2018-02-28) Sitti, Metin ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machines ; Max Planck Institute for Intelligent Systems
    Soft functional materials could enable physical intelligence for small-scale (from a few millimeters down to a few micrometers overall size) mobile robots by enabling them with unique capabilities such as shape changing and programming, adaptation, and multi-functional and diverse behavior. This talk will explore our recent activities on how to design, manufacture, and control new untethered soft actuators, sensors, robots, and shape-programmable materials at the milli/microscale. First, inflated soft actuators with reversible stable deformations are proposed combining hyperelastic membranes and dielectric elastomer actuators to switch between stable deformations of sealed chambers. Next, new parallel microcracks-based ultrasensitive and highly stretchable soft strain sensors are integrated with gecko-inspired microfiber adhesives for wearable medical devices adhered on the skin. Next, new, untethered milli/microscale swimming robots inspired by spermatozoids and jellyfish are proposed using elastomeric magnetic composite materials. Static and dynamic shapes of such magnetic active soft materials are programmed using a computational design methodology. These soft robots are demonstrated to be able to have seven or more locomotion modalities (undulatory swimming, jellyfish-like swimming, water meniscus climbing, jumping, ground walking, rolling, crawling inside constrained environments, etc.) in a single robot for the first time to be able to move on complex environments, such as inside the human body. Ultrasound-guided navigation of such robots is possible towards medical functions such as local cargo/drug delivery.
  • Item
    Robotic Manipulation: A Broadening View
    (Georgia Institute of Technology, 2019-02-20) Mason, Matthew T. ; Georgia Institute of Technology. Institute for Robotics and Intelligent Machines ; Carnegie Mellon University. School of Computer Science
    Professor Mason discusses a rambling mashup of topics in robotic manipulation, manipulation by apes and humans, early research on manipulation in the blocks world, more recent research motivated by e-commerce and logistics, with a brave attempt to identify cross-cutting principles and future directions.