Seminars

Seminars From Year 2014

View seminars by year:     2003    2004    2005    2006    2007    2008    2009    2010    2011    2012    2013    2014    2015    All    

All seminars will be held in Wu & Chen Auditorium, Levine Hall (3330 Walnut Street) unless otherwise indicated.

  • Talk: GRASP Special Seminar: Jeremy Gillula, University of California at Berkeley, "Guaranteeing Safe Online Machine Learning via Reachability Analysis"
    Date: Thursday, January 16, 2014 - 12pm to 1pm
    Presenters: Jeremy Gillula
    Alternate Location: Moore 317 (inside Moore 316)
    Reinforcement learning has proven itself to be a powerful technique in robotics, however it has rarely been employed to learn in a hardware-in-the-loop environment due to the fact that spurious training data could cause a robot to take an unsafe (and potentially catastrophic) action. We will present a method for overcoming this limitation known as Guaranteed Safe Online Learning via Reachability (GSOLR), in which the control outputs from the reinforcement learning algorithm are wrapped inside another controller based on reachability analysis that seeks to guarantee safety against worst-case disturbances. After defining the relevant backwards reachability constructs and explaining how they can be calculated, we will formalize the concept of GSOLR and show how it can be used on a real-world target tracking problem, in which an observing quadrotor helicopter must keep a target ground vehicle with unknown (but bounded) dynamics inside its field of view at all times, while simultaneously attempting to build a motion model of the target.  Extensions to GSOLR will then be presented, which allow the safety of the system to automatically become neither too liberal nor too conservative, thus allowing the machine learning algorithm running in parallel the widest possible latitude while still guaranteeing system safety. These extensions will be demonstrated on the task of safely learning an altitude controller for a quadrotor helicopter.  These examples demonstrate the GSOLR framework's robustness to errors in machine learning algorithms, and indicate its potential for allowing high-performance machine learning systems to be used in safety-critical situations in the future.
  • Talk: GRASP Special Seminar: Jnaneshwar Das, University of Southern California, "Data-driven Robotic Sampling for Marine Ecosystem Monitoring"
    Date: Friday, January 17, 2014 - 11am to 12pm
    Presenters: Jnaneshwar Das
    Alternate Location: Levine 307 (3330 Walnut Street)
    Robotic sampling is attractive in many field robotics applications that require persistent collection of physical samples for ex-situ analysis. Examples abound in the earth sciences in studies involving the collection of rock, soil, and water samples for lab analysis. The desirability of samples in these domains can be expressed as a property that cannot be determined in-situ, but can be predicted by covariates measurable in real-time using sensors carried aboard a robot. In our test domain, marine ecosystem monitoring, accurate measurement of plankton abundance requires lab analysis of water samples, but predictions using physical and chemical properties measured in real-time by sensors carried aboard an autonomous underwater vehicle (AUV) can guide sample collection decisions. We present a principled approach to minimize cumulative regret of plankton samples acquired by an AUV over multiple surveys in batches of k water samples per survey. Samples are labeled at the end of each survey, and used to update a probabilistic model that guides sampling in subsequent surveys. The problem is formulated in an online setting: given a predetermined survey duration and a probabilistic model learned from earlier surveys, the AUV makes irrevocable sample collection decisions on a sequential stream of candidates, with no knowledge of the future. Our experimental results are based on extensive retrospective studies emulating 100 campaigns, each composed of 17 surveys. The campaigns were emulated by mining historical field data collected by an AUV operating at depths of up to 100 m over a 40 sq. km area in an 8 day period. These studies establish the efficacy of the approach - beginning with no prior, successive surveys by the AUV result in samples that are progressively higher-abundance in a pre-specified type of plankton. Additionally we carried out a one-day field trial with an AUV operating at depths of up to 30 m over a 1 sq. km area. Beginning with a prior learned from data collected and labeled in an earlier campaign, the AUV field survey resulted in samples with a high-abundance of a pre-specified type of plankton - a potentially toxinogenic alga of interest to marine ecologists. This is the first time such a field experiment has been carried out in its entirety in a data-driven fashion, in effect 'closing the loop' on a significant and relevant ecosystem monitoring problem. Although the experimental context for work is marine ecosystem monitoring, it is well-suited for autonomous and persistent robotic observation of any property that cannot be measured in-situ, but possesses observable covariates, thus opening up the potential for advanced autonomous robotic exploration of unstructured environments that are inaccessible to humans.
  • Talk: Spring 2014 GRASP Seminar: Joelle Pineau, McGill University, "Learning Socially Adaptive Navigation Strategies : Lessons from the SmartWheeler Project"
    Date: Friday, January 31, 2014 - 11am to 12pm
    Presenters: Joelle Pineau
    A key skill for mobile robots is the ability to navigate efficiently through their environment. In the case of social or assistive robots, this involves navigating through human crowds. Typical performance criteria, such as reaching the goal using the shortest path, are not appropriate in such environments, where it is more important for the robot to move in a socially acceptable manner. In this talk I will describe new methods based on imitation and reinforcement learning which we have developed to allow robots to achieve socially adaptive path planning in human environments. Performance of these methods will be illustrated using a smart power wheelchair developed in our group, called the SmartWheeler.
  • Talk: Spring 2014 GRASP Seminar: Aaron Dollar, Yale University, "Reengineering the Hand: "Mechanical Intelligence" in Robotic Manipulation"
    Date: Friday, February 7, 2014 - 11am to 12pm
    Presenters: Aaron Dollar
    Despite decades of research, current robotic systems are unable to reliably grasp and manipulate a wide range of unstructured objects in human environments. The somewhat traditional approach of attempting to copy the immense mechanical complexity of the human hand in a stiff "robotic" mechanism, and the subsequently required levels of sensing and control, has not yet been successful. Alternatively, with careful attention to the design of the mechanics of hands, including adaptive underacted transmissions and carefully tuned compliance, we have been able to achieve a level of dexterity and reliability as yet unseen in the robotics community. I will describe ongoing efforts to further develop grasping and dexterous manipulation capabilities in engineered systems as well as our work in studying human hand function to guide some of the efforts.
  • Talk: Spring 2015 GRASP Seminar: Soon-Jo Chung, University of Illinois at Urbana-Champaign, "Revolutionary Aerial Drones: Control and Estimation Challenges of Developing Robotic Falcons to Prevent Airport Bird Strikes"
    Date: Thursday, February 13, 2014 - 11am to 12pm
    Presenters: Soon-Jo Chung
    The rapid and ubiquitous proliferation of reliable rotorcraft platforms such as quadcopters has resulted in a boom in aerial robotics. However, rotorcraft have issues of safety, high noise levels, and low efficiency for forward flight. The objective of this NSF CAREER project, motivated by the problem of keeping airfields clear of disruptive avian flocks, is to develop control and sensing strategies for bird-like flapping robots that can be deployed in swarms to fend off "antagonists." This talk gives an overview of technical challenges in developing a bio-inspired aerial robot platform from the dynamics and controls standpoint. We study the stability of coupled nonlinear oscillators by using contraction analysis to prove that flapping flight dynamics without traditional aerodynamic control surfaces can be effectively controlled by a reduced set of central pattern generator (CPG) parameters that generate complex 3D oscillatory motions of two main wings. New motion planning and flight control strategies are used to demonstrate agile, high-speed flight in a forest and perform perched landings on a human hand. This talk also presents a PDE boundary control formulation of controlling flexible wings described by PDEs and whose output is given by a spatial integral of weighted functions of the state. For wing bending, this talk discusses a novel control scheme based on a dyadic perturbation observer (DPO). A new design approach to optimal nonlinear estimation is discussed with emphasis on its application to vision-based Simultaneous Localization and Mapping (SLAM). The observer gain synthesis algorithm, called linear matrix inequality state-dependent algebraic Riccati equation (LMI-SDARE) guarantees stochastic incremental stability for a set of Itô stochastic nonlinear systems.
  • Talk: Spring 2014 GRASP Seminar: Al Rizzi, Boston Dynamics, "Legged Robotics at Boston Dynamics"
    Date: Friday, February 14, 2014 - 11am to 12pm
    Presenters: Alfred Rizzi
    Only about half the Earth's landmass is accessible to wheeled and  tracked vehicles, yet people and animals can go almost everywhere on foot. Our goal is to develop novel locomotion systems that can go anywhere people and animals go. The systems we build combine dynamic control systems, actuated mechanisms and sensing to travel on terrain that is too rocky, sandy, muddy, snowy, wet or steep for existing conventional vehicles. This presentation will discuss progress at Boston Dynamics in building such systems, including WildCat, LS3, Atlas, RHex, PETMAN and others.
  • Talk: MEAM / GRASP Seminar: Matthew Turpin, University of Pennsylvania, "Scalable Trajectory Computation for Large Teams of Interchangeable Robots Applied to Quadrotor MAVs"
    Date: Wednesday, February 19, 2014 - 1pm to 2pm
    Presenters: Matthew Turpin
    Alternate Location: Levine 307 (3330 Walnut Street)
    Large teams of robots have been implemented to great success in Kiva's automated warehouses as well as UPenn's and KMel Robotics' swarms of quadrotors. In settings such as these, robots must plan paths which avoid collisions with other robots and obstacles in the environment. Unfortunately, trajectory planning for large teams of robots generally suffers from either the curse of dimensionality or lack of completeness. I will demonstrate that relaxing the assumption of labeling each robot and specifying a  fixed assignment of robots to destinations in the trajectory generation problem yields a number of computational and performance benefits. My algorithm to solve this Concurrent Assignment and Planning of Trajectories (CAPT) problem has bounded computational complexity of O(N^3), preserves completeness properties of a user specified single agent motion planner, and tends to minimize effort exerted by any one robot. This algorithm generates solutions to variants of the CAPT problem in settings ranging from kinematic robots in an obstacle free environment to teams of robots with 4th order dynamics in a cluttered environment. Finally,  I will show experimental results of the algorithm applied on teams of second order aquatic vehicles as well as on quadrotor micro aerial vehicles. I will also outline how time consuming aspects of this approach can be parallelized and discuss possible decentralized implementations.
  • Talk: CANCELED: Spring 2014 GRASP Seminar: Kris Hauser, Indiana University, "Motion Planning for Real World Robots"
    Date: Friday, February 21, 2014 - 11am to 12pm
    Presenters: Kris Hauser
    Motion planning -- the problem of computing physical actions to complete a specified task -- has inspired some of the most theoretically rigorous and beautiful results in robotics research.  But as robots proliferate in real-world applications like household service, driverless cars, warehouse automation, minimally-invasive surgery, search-and-rescue, and unmanned aerial vehicles, the classical theory appears to have fallen behind the pace of practice.  At odds with the "clean" assumptions of theory, the reality is that robots must handle large amounts of noisy sensor data, uncertainty, underspecified models, nonlinear and hysteretic dynamic effects, exotic objective functions and constraints, and real-time demands.  This talk will describe efforts to bring theory up to speed, in the context of three projects: 1) ladder climbing in the DARPA Robotics Challenge; 2) intelligent user interfaces for human-operated robots; and 3) navigation amongst many moving obstacles.  I will present new planning algorithms and architectures whose performance is backed both by theoretical guarantees and empirical evaluation.
  • Talk: Spring 2014 GRASP Seminar: Ryan Eustice, University of Michigan, "SLAM in the Wild: Robust and Persistent Visual SLAM for Autonomous Underwater Hull Inspection"
    Date: Friday, February 28, 2014 - 11am to 12pm
    Presenters: Ryan Eustice
    The field of simultaneous localization and mapping (SLAM) has made tremendous progress in the last couple of decades, to the point where we have mature-enough methods and algorithms to explore applications on interesting scales both spatially and temporally.  In this talk we discuss some of our current efforts in deploying large-scale, long-term SLAM systems in real-world field applications, and in particular, our current work in autonomous underwater ship hull inspection.  We will discuss our developments in modeling the visual saliency of underwater imagery for pose-graph SLAM, how this saliency measure can be used within an active SLAM planning paradigm, and our development of generic linear constraints---a principled framework for pose-graph reduction, which is important for controlling multi-session SLAM graph complexity.
  • Talk: Spring 2014 GRASP Seminar: Leila Takayama, Google[x], "Designing for the Seemingly Nonsensical Ways People See, Treat, and Use Robots"
    Date: Friday, March 7, 2014 - 11am to 12pm
    Presenters: Leila Takayama
    As robots are entering our everyday lives, it is becoming increasingly important to understand how untrained people will interact with robots. Fortunately, untrained people already interact with a variety of robotic agents (withdrawing cash from ATMs, driving cars with anti-lock brakes) so we are not completely starting from scratch. In the moment of those interactions with robotic agents, people behave in ways that do not necessarily align with the rational belief that robots are just plain machines. Through a combination of controlled experiments and field studies, this talk will examine the ways that untrained people interact with robotic agents, including (1) how we interact with personal robots, and (2) how we interact through telepresence robots. Drawing from theories of human-computer interaction and this type of empirical research, we provide implications for both theory and the design of interactive robotic systems.
  • Talk: GRASP Special Seminar: Sergio Pequito, Carnegie Mellon University, "A Framework for Structural Input/Output and Control Configuration Selection of Large-Scale Systems"
    Date: Thursday, March 13, 2014 - 11am to 12pm
    Presenters: Sérgio Pequito
    Alternate Location: Levine 307 (3330 Walnut Street)
    The structure control system design consists mainly of two steps: input/output (I/O) selection and control configuration (CC) selection. The first one is devoted to the problem of computing how many actuators/sensors are needed and where should be placed in the plant to obtain some desired property. Control configuration is related to the decentralized control problem and is dedicated to the task of selecting which outputs (sensors) should be available for feedback and to which inputs (actuators) in order to achieve a predefined goal. The choice of inputs and outputs affects the performance, complexity and costs of the control system. Due to the combinatorial nature of the selection problem, an efficient and systematic method is required to complement the designer intuition, experience and physical insight. Motivated by the above, this presentation addresses the structure control system design taking explicitly into consideration the possible application to large-scale systems. We provide an efficient framework to solve the following major minimization problems: i) selection of the minimum number of manipulated/measured variables to achieve structural controllability/observability of the system, and ii) selection of the minimum number of measured and manipulated variables, and feedback interconnections between them such that the system has no structural fixed modes. Contrary to what would be expected, we showed that it is possible to obtain the global solution of the aforementioned minimization problems in polynomial complexity in the number of the state variables of the system. To this effect, we propose a methodology that is efficient (polynomial complexity) and unified in the sense that it solves simultaneously the I/O and the CC selection problems. This is done by exploiting the implications of the I/O selection in the solution to the CC problem. An example illustrate the main features of the proposed procedure.
  • Talk: GRASP Special Seminar (Joint Talk): Daigo Muramatsu & Ikuhisa Mitsugami, Osaka University)
    Date: Thursday, March 13, 2014 - 2pm to 3pm
    Presenters: Daigo Muramatsu & Ikuhisa Mitsugami
    Alternate Location: Levine 307
    Presenter: Daigo Muramatsu "Cross-view Gait Recognition" Abstract: Gait recognition is a biometric method used to recognize a person from their walking style, which can be acquired from a camera. Unlike many biometric techniques such as fingerprint, iris or face recognition, gait recognition can authenticate a person some distance from the camera, because it has high accuracy even when the resolution of an image sequence is relatively low. However, the accuracy of gait recognition is often degraded by view difference. In this talk, we focus on the view issue of gait recognition and discuss some solution against accuracy degradation caused by the view difference. ---------------------------------------------------------------------- Presenter: Ikuhisa Mitsugami"3-D Measurement and Analysis of Walking Person by Range Sensing" Abstract: Consumer depth sensors, e.g. Microsoft Kinects, are getting more attention because of their low cost and ability to obtain 3-D measurements. We adopt such depth sensors for gait analysis. In this talk, we introduce some of our achievements. One of the recent achievements is full body reconstruction of a walking person. Since the scene is dynamic, we cannot achieve the full body reconstruction only by merging asynchronous range data of Kinects. We thus propose a synchronization method to virtually obtain depth data at the same moment. We also introduce a new gait feature representation based on range observation. It is basically an extension of an existing silhouette-based feature, but shows promising performance in person authentication task.
  • Talk: GRASP Special Seminar: Hyun Soo Park, Carnegie Mellon University, "Understanding a Social Scene from Social Cameras"
    Date: Friday, March 14, 2014 - 1pm to 2pm
    Presenters: Hyun Soo Park
    Alternate Location: Levine 512
    A social camera is a camera carried or worn by a member of a social group, (e.g., a smartphone camera, a hand-held camcorder, or a wearable camera). These cameras are becoming increasingly immersed in our social lives and closely capture our social activities. In this talk, I argue that social cameras are the ideal sensors for social scene understanding, as they inherit social signals such as the gaze behavior of the people carrying them. I will present a computational representation for social scene understanding from social cameras.  In the first part of my talk, I will show how visible social signals, such as body gestures, gaze directions, or facial expression, can be recovered in 3D from social cameras. This work includes 3D trajectory reconstruction and motion capture from body-mounted cameras. The second part of the talk will focus on analysis on the relationship between the social signals using 3D joint attention. This analysis allows us to predict social gaze behaviors.
  • Talk: Spring 2014 GRASP Seminar: Martial Hebert, Carnegie Mellon University, "Challenges in Semantic Perception for Autonomous Systems"
    Date: Friday, March 21, 2014 - 11am to 12pm
    Presenters: Martial Hebert
    Despite considerable progress in all aspects of machine perception, using machine vision in autonomous systems remains a formidable challenge. This is especially true in applications such as robotics, in which even a small error rate in the perception system can have catastrophic consequences for the overall system. This talk will review a few ideas that could be used to start formalizing the issues revolving around the integrating vision systems. They include a systematic approach to the problem of self-assessment of vision algorithm and predicting quality metrics on the inputs to the vision algorithms, ideas on how to manage multiple hypotheses generated from a vision algorithm rather than relying on a single "hard" decision, and methods for using external (non-visual) domain- and task-dependent information. These ideas will be illustrated with examples of recent vision for scene understanding, depth estimation, and object recognition.
  • Talk: Spring 2014 GRASP Seminar: Stefanie Tellex, Brown University, "Natural Language and Robotics"
    Date: Friday, March 28, 2014 - 11am to 12pm
    Presenters: Stefanie Tellex
    Natural language can be a powerful, flexible way for people to interact with robots.  A particular challenge for designers of embodied robots, in contrast to disembodied methods such as  phone-based information systems, is that natural language understanding systems must map between linguistic elements and aspects of the external world, thereby solving the so-called symbol grounding problem.  This talk describes a probabilistic framework for robust interpretation of grounded natural language, called Generalized Grounding Graphs (G^3).  The G^3 framework leverages the structure of language to define a probabilistic graphical model that maps between elements in the language and aspects of the external world.  It can compose learned word meanings to understand novel commands that may have never been seen during training.  Taking a probabilistic approach enables the robot to employ information-theoretic dialog strategies, asking targeted questions to reduce uncertainty about different parts of a natural language command.  By inverting the model, the robot can generated targeted natural language requests for help from a human partner.  This approach points the way toward more general models of grounded language understanding, which will lead to robots capable of building world models from both linguistic and non-linguistic input, following complex grounded natural language commands, and engaging in fluid, flexible dialog with their human partners.
  • Talk: AMCS/GRASP Seminar: Marty Golubitsky, Ohio State University, "Patterns of Synchrony: From Animal Gaits to Binocular Rivalry"
    Date: Friday, March 28, 2014 - 2pm to 3pm
    Presenters: Marty Golubitsky
    Alternate Location: Towne 337
    This talk will discuss previous work on quadrupedal gaits and recent work on a generalized model for binocular rivalry proposed by Hugh Wilson. Both applications show how rigid phase-shift synchrony in periodic solutions of coupled systems of differential equations can help understand high level collective behavior in the nervous system.  For gaits the symmetries predict unexpected gaits and for binocular rivalry the symmetries predict unexpected percepts.
  • Talk: GRASP Special Seminar: Masaki Ogura, Texas Tech University, "Stability Analysis of Switched Linear Systems with Non-Traditional Switching Signals"
    Date: Monday, April 7, 2014 - 2pm to 3pm
    Presenters: Masaki Ogura
    Alternate Location: Levine 307
    The talk presents my recent research on the stability analysis of switched systems, which are a class of dynamical systems whose dynamics can abruptly change. Examples include the control of systems over unreliable networks or with a failure-prone controller. In this talk I will discuss a fundamental property called stability of switched linear systems. I will in particular focus on the case when switching is modeled by non-traditional stochastic processes, in particular, by non-Markovian processes.
  • Talk: Spring 2014 GRASP Seminar: Andrea Thomaz, Georgia Institute of Technology, "Designing Learning Interactions for Robots"
    Date: Friday, April 11, 2014 - 11am to 12pm
    Presenters: Andrea Thomaz
    In this talk I present recent work from the Socially Intelligent Machines Lab at Georgia Tech. One of the focuses of our lab is on Socially Guided Machine Learning, building robot systems that can learn from everyday human teachers. We look at standard Machine Learning interactions and redesign interfaces and algorithms to support the collection of learning input from naive humans. This talk covers results on building computational models of reciprocal social interactions, high-level task goal learning, low-level skill learning, and active learning interactions using several humanoid robot platforms.
  • Talk: Spring 2014 GRASP Seminar: E. Michael Golda, Navy Sea Systems Command Carderock Division, "A Brief Overview of United States Navy Machinery Automation Challenges"
    Date: Friday, April 18, 2014 - 11am to 12pm
    Presenters: E. Michael Golda
    A large naval warship ship is the most complex structure built by man.  The technology trends over the last 70 years have made automation a necessity for controlling the components, systems, and integrated systems of systems that make up a warship.  The presentation will provide a brief introduction of the ship as a system of systems.  The evolution of the Navy’s automation to intelligent agent-based distributed controls will be described.  In addition, opportunities for educational support and joint research with the Navy opportunities will be discussed.
  • Talk: GRASP Special Seminar: Anthony Hoogs, Kitware, Inc., “Persistent, Wide-Area Video Analytics at Big Data Scales”
    Date: Tuesday, April 22, 2014 - 2pm to 3pm
    Presenters: Anthony Hoogs
    Alternate Location: Towne 337
    The scale of video data continues to grow exponentially, including city-scale wide-area aerial video showing hundreds or thousands of simultaneous movers. Extracting the most interesting, salient content from this type of video is of increasing importance as the data volume grows while the vast majority of events are not of interest. However, traditional methods often fail because of low resolution and low frame rates in this domain. At Kitware we have developed methods for detecting events, actions, complex activities, patterns of life and anomalies in large-scale video domains. We detect events, anomalies and complex activities efficiently by detecting and tracking all movers, then characterizing their behavior using event-independent descriptors. Efficient inference is achieved through layered, approximate evaluation as model complexity increases. Functional scene elements such as parking spots are recognized by analyzing behavior within and around them. Behavioral normalcy models are learned, and anomalies are detected using location-dependent and location-independent techniques. The talk will provide an overview of these methods and results on wide-area and ground-level surveillance video.
  • Talk: GRASP Special Seminar: Dr. Andrzej Banaszuk, Dr. Jason Derenick, Dr. Alberto Speranzon, UTRC, "Autonomous and Intelligent Systems at United Technologies Research Center"
    Date: Friday, May 16, 2014 - 11am to 12pm
    Presenters: Dr. Andrzej Banaszuk, Dr. Jason Derenick, Dr. Alberto Speranzon
    Alternate Location: Levine 307
    We will present UTRC’s research initiative in Autonomous and Intelligent Systems with an emphasis on complex human/machine intelligent systems including unmanned rotorcraft. The research, conducted by a diverse team of researchers in robotics, dynamical systems, control, applied mathematics, computer vision, and computer science (in partnership with several leading universities including CMU, MIT, UPenn, and Caltech) includes: •         Real-time algorithms for dynamic collision avoidance in an obstacle-rich environment using probabilistic roadmaps. •         Multi-vehicle missions including efficient search algorithms based on ergodic theory methods. •         Multi-vehicle navigation with imperfect and intermittent sensors in GPS degraded environments. •         Intelligent system design methodology including architectures for autonomy, human-machine systems, and formal verification. In particular, we will provide an overview of a new hierarchical planning framework for mission planning and execution in uncertain and dynamic environments. We consider missions that involve motion planning in large, cluttered environments, trading off mission objectives while satisfying logical/spatial/temporal constraints. Our framework enables the decomposition of the planning problem across different layers, leveraging the difference in spatial and temporal scales of the mission objectives. Of the hierarchical planner we will describe, in some more detail, a novel motion planning algorithm that, starting from a probabilistic roadmap, efficiently constructs an expanded graph used to search for the optimal solution of a multi-objective problem that trades off path length and state estimation accuracy when navigating in a GPS denied environment. Tradeoff between optimality and computational complexity will be discussed as well as open challenge problems. During the second half of this seminar, the focus turns to the problem of state estimation in GPS-denied, but structurally rich environments (e.g., urban canyons).  A homological mapping and state estimation pipeline is formulated that makes use of dual simplicial nerve complexes for metric-free, time-asynchronous map building using only the binary (and visibility-based) detection of signals of opportunity (e.g., MAC addresses of WiFi access points).  Using these constructs as an approximation to environmental free-space, the notion of “homological sensing” is proposed whereby an agent uses local homology computations to implicitly enumerate the number of physical structures (e.g., buildings) within some coarse proximity to its unknown location. Experimental results demonstrating the utility of said approach for coarse localization using a non-parametric Bayesian filter are presented. We will conclude with research problems of interest to UTRC and discuss existing and future career opportunities in the broad area of robotics.
  • Talk: GRASP Special Seminar: Zhengyou Zhang, Microsoft Research, "3D Computer Vision and Multimodal Interaction for Large Touch Displays"
    Date: Monday, May 19, 2014 - 1pm to 2pm
    Presenters: Zhengyou Zhang
    Alternate Location: Levine 307
    We describe a system that augments touch input with visual understanding of the user to improve interaction with a large touch-sensitive display. A commodity color plus depth sensor such as Microsoft Kinect adds the visual modality and enables new interactions beyond touch. Through visual analysis, the system understands where the user is, who the user is, and what the user is doing even before the user touches the display. Such information is used to enhance interaction in multiple ways. For example, a user can use simple gestures to bring up menu items such as color palette and soft keyboard; menu items can be shown where the user is and can follow the user; hovering can show information to the user before the user commits to touch; the user can perform different functions (for example writing and erasing) with different hands; and the user’s preference profile can be maintained, distinct from other users. User studies are conducted and the users very much appreciate the value of these and other enhanced interactions.
  • Talk: GRASP Special Seminar: Pratap Tokekar, University of Minnesota, “Sensing Planning for Robotic Environmental Monitoring: Systems and Algorithms”
    Date: Tuesday, May 27, 2014 - 4pm to 5pm
    Presenters: Pratap Tokekar
    Alternate Location: Levine 307
    Data collected from the environment is becoming increasingly important for studying complex phenomena. Robotic sensing systems will revolutionize this area by enabling access to data gathered at unprecedented spatio-temporal scales. Environmental science applications, in turn, motivate challenging robotics research problems. In this talk, I will present new algorithms for sensing planning problems motivated by these applications, and the design of robotic monitoring systems developed for two real-world applications. In the first part, I will present the design of an autonomous robotic system developed for monitoring invasive fish in lakes. In the second part, we will focus on the algorithmic aspects of the art gallery problem, a classical camera placement problem, under constraints motivated by surveillance and visual monitoring. The standard formulation does not take into account self-occlusions of an object-of-interest within the environment. We will formulate the art gallery problem to guarantee that despite self-occlusions, any convex object present anywhere in the environment, will be seen from all sides. I will present bounds and approximation algorithms for placing cameras under these constraints. In the third part, I will describe a robotic data collection system for precision agriculture. We will study the problem of planning energy-aware trajectories for collaborative aerial and ground robots in the context of this application. Throughout the talk, along with theoretical results, I will present experiments conducted with autonomous boats in Minnesota lakes, wheeled robots operating on frozen lakes, and an aerial robot operating over a corn plot.
  • Talk: MEAM / GRASP Seminar: Jun Seo, University of Pennsylvania, "Grasping and Assembling with Modular Robots"
    Date: Friday, June 20, 2014 - 12pm to 1pm
    Presenters: Jun Seo
    Alternate Location: Towne 337
    In this talk, I will present my research on robotic grasping and assembling: how to enable robots to grasp objects and assemble target structures autonomously. They are fundamental problems in robotic manipulation where we have robots rearrange our environment; they can be applied to a wide range of scenarios such as manufacturing, disaster response, space exploration, and assisted living. Modular robot platforms can facilitate the problems with versatility and robustness. In the first part, I will discuss grasping/fixturing objects with customized effectors and implementing the ideas with modular arm links and end-effectors. In the second part, I will focus on assembly planning for target structures of the common brick wall pattern with rectangular, modular robots, which is a part of collective efforts for realizing a swarm of autonomous boats in the Modular Robotics Lab at Penn.
  • Talk: GRASP Special Seminar: Matthew Mason, Carnegie Mellon University, "What is Manipulation?"
    Date: Tuesday, July 1, 2014 - 12pm to 1pm
    Presenters: Matthew T. Mason
    Alternate Location: Levine 307
    Robotic manipulation has been developing for over fifty years, but the true history of manipulation goes back for millions of years, including teleoperation, tool-wielding humans (and other animals), and even "naked" humans and other animals. When we draw examples from the entire history of manipulation, we see a broad set of problems and techniques, and many interesting research opportunities. This talk addresses the dreary issue of how to define manipulation, interspersed with some potentially less dreary examples of manipulation performance and research, including recent work on "extrinsic dexterity" (which you can find on YouTube).
  • Talk: GRASP Special Seminar: Changhyun Choi, Georgia Institute of Technology, "Visual Object Perception in Unstructured Environments"
    Date: Friday, July 18, 2014 - 1pm to 2pm
    Presenters: Changhyun Choi
    Alternate Location: Levine 307
    As robotic systems move from well-controlled settings to increasingly unstructured environments, they are required to operate in highly dynamic and cluttered scenarios. Finding an object, estimating its pose, and tracking its pose over time within such scenarios are challenging problems. Although various approaches have been developed to tackle these problems, the scope of objects addressed and the robustness of solutions remain limited. In this talk, I will present a robust object perception using visual sensory information, which spans from the traditional monocular camera to the more recently emerged RGB-D sensor, in unstructured environments. I will address four important challenges, such as significant clutter in backgrounds, objects with and without texture, object discontinuous cases during tracking, and real-time constraints, to robust 6-DOF object pose estimation and tracking that current state-of-the-art approaches have, as yet, failed to solve. Various object pose estimation and tracking examples will be shown with several applications in robotics.
  • Talk: GRASP REU Site Oral Presentations - Summer 2014
    Date: Tuesday, August 5, 2014 - 10am to 3pm
    Presenters: GRASP REU Summer 2014 Participants
    Alternate Location: Wu and Chen Auditorium (Levine 101)
    GRASP REU Site Oral PresentationsTuesday, August 5, 2014Wu and Chen Auditorium10:30am - 12:00pm and 1:30pm - 3:00pm  Morning Session: 10:30 a.m. to NoonWelcome by Katherine J. Kuchenbecker and Max Mintz, GRASP REU Site Co-Directors 10:30 a.m.      Eli Soffer Rising Junior in Electrical Engineering at The Cooper Union Advised by Professor Michelle Johnson and mentored by Addwiteey Chrungoo Developing a Graphical User Interface and Improved Trajectory Modelling for Task Oriented Robot Assisted Therapy 10:45 a.m.      Hunter Young Rising Senior in Mechanical Engineering at Oklahoma State University Advised by Professor Michelle Johnson and mentored by Dalton Banks Investigating Under-Actuated Robot Hands for Stroke Rehabilitation 11:00 a.m.     Michelle Neuburger Rising Senior in Computer Science at Bryn Mawr College Advised by Professor Katherine J. Kuchenbecker and mentored by Naomi Fitter Human-Robot Interaction Through Cooperative Hand-Clapping Games 11:15 a.m.      Lan Pham Rising Junior in Mechanical Engineering at Syracuse University Advised by Professor Katherine J. Kuchenbecker and mentored by Rebecca Pierce Refining and Testing a Wearable Tactile-Feedback Device for Telerobotics 11:30 a.m.     Armon Shariati Rising Senior in Computer Engineering at Lehigh University Advised by Professor CJ Taylor and mentored by Anthony Cowley A General Platform for Immersive Virtual Reality and Interactive Manipulation with Haptic Feedback via Motion Capture 11:45 a.m.      Kam Hou U Rising Senior in Computer Engineering at Stony Brook University Advised by Professor Mark Yim and mentored by Tarik Tosun Design of a Control and Communication Framework for the SMORES Robots Afternoon Session: 1:30 a.m. to 3:00p.m.Welcome by Katherine J. Kuchenbecker and Max Mintz, GRASP REU Site Co-Directors 1:30 p.m.      Samer Nashed Rising Senior in Computer Science and Physics at Swarthmore College Advised by Professor Vijay Kumar and mentored by Justin Thomas and Monroe Kennedy Decentralized Cooperative Towing using Implicit Communication 1:45 p.m.      Aaron Effron Rising Senior in Physics at Yale University Advised by Professor Dan Koditschek and mentored by Garrett Wenger Tail Effector Manipulation for Improved Jumping Actuation in a Hexapedal Robot2:00 p.m.      Sarah Cen Rising Junior in Mechanical and Aerospace Engineering at Princeton University Advised by Professor Vijay Kumar and mentored by Sarah Tang and Mickey Whitzer Autonomous Flight and Landing of a Quadrotor on a Moving Ground Vehicle Using April Tag Vision-Based Control 2:15 p.m.      Cody Worth Senior in Computer Science and Mathematics at Lamar University Advised by Dr. Ufuk Topcu and mentored by Dr. Jie FuAutomata Learning for Autonomous, Reactive Systems 2:30 p.m.      Lisa Lee Rising Senior in Mathematics at Princeton University Advised by Professor Eric Eaton and mentored by Dr. Haitham Bou Ammar and Chris Clingerman Robotic Search & Rescue via Online Multi-Task Reinforcement Learning 2:45 p.m.      Peter Wei Rising Senior in Electrical and Computer Engineering at Carnegie Mellon University Advised by Professor Dan Lee and mentored by Chris Clingerman Dynamic Replanning with Manipulability Estimation and Probabilistic Costmaps Many thanks to the advisors, mentors, colleagues, staff, GRASP Lab, and larger Penn community for helping make the third year of the GRASP REU Site such a success. We are especially indebted to Charity Payne and Shafag Idris for their excellent work in running the program this summer. We also appreciate the service of the judges for the presentations. Congratulations to all twelve GRASP REU students!
  • Talk: GRASP Special Seminar: Adam Spiers, Yale University, "Haptics Research in Art and Medicine"
    Date: Monday, August 11, 2014 - 1pm to 2pm
    Presenters: Adam Spiers
    Alternate Location: Levine 307
    In this talk I will cover two quite different areas of research that are linked by the common theme of haptics. First, I will describe some medical haptics research I recently completed at the Bristol Robotics Laboratory (UK). Here we investigated various haptic concepts in relation to manual and tele-operated (robotic) minimally invasive surgery, leading to several observations and technical developments. Some highlights include the development of a dexterous gripper for palpation, a deformable fingertip sensor and multi-modal haptic feedback systems featuring fingertip deformation and skin stretch. The second part of the talk relates to the development of haptic navigation devices for use in cultural spaces. An ongoing collaboration with a London based visually impaired theatre group and the Open University, the first iteration of this work led to the ‘Haptic Lotus’ device in 2010. This handheld device expanded as blind and sighted audience members approached specific areas in an immersive art installation. In this talk I will explain the motivation behind our work and the developments we hope to achieve with our most recent set of funding for a new project, ‘Flatland’.
  • Talk: Fall 2014 GRASP Seminar - GRASP Faculty Research & Welcome
    Date: Friday, September 5, 2014 - 11am to 12pm
    Presenters: GRASP Faculty Members
    Dan Ueda (GRASP Education & Outreach)Dr. Mark Yim (presented by Matt Piccoli) Dr. CJ Taylor Dr. Jianbo Shi (presented by Jihua Huang) Dr. George Pappas (presented by Nick Atanasov) Dr. Daniel Lee (presented by Steve McGill) Dr. Vijay Kumar Dr. Katherine Kuchenbecker Dr. Daniel Koditschek (presented by Dr. Aaron Johnson) Dr. Michelle Johnson Dr. Ali Jadbabaie Dr. Robert Ghrist Dr. Jean Gallier Dr. Eric Eaton Dr. Kostas Daniilidis 
  • Talk: GRASP Special Seminar: Pei-Chun Lin, National Taiwan University, "Model-based Dynamic Locomotion on Legged Robots"
    Date: Tuesday, September 9, 2014 - 12pm to 1pm
    Presenters: Pei-Chun Lin
    Alternate Location: Raisler Lounge (Towne 225)
    Dynamic and agile locomotion performed by legged animals demonstrates a feasible but still very challenging style of locomotion to be exhibited on artificial legged platforms. How animals evolved to their present forms and how they coordinate their high degree-of-freedom (DOF) active/passive joints for dynamic maneuver are still puzzles to be explored. Here, with the goal of exciting dynamic behaviors on empirical robots, we at Bio-inspired Robotics Laboratory, National Taiwan University, would like to address this problem by exploring simple reduced-order dynamic models and linking their behaviors to these of the complex robot platforms.  Two types of robots will be introduced in this talk. One is the leg-wheel transformable robot. It is implemented with a unique leg-wheel transformation mechanism that directly switches the morphology of the driving mechanism between wheels (i.e., a full circle) and 2-DOF leg (i.e., combining two half circles as a leg), so that the same actuation system can be efficiently utilized in both wheeled and legged modes. The robot is also implemented with coupled oscillators, which act as the main mechanism to coordinate four leg-wheel motions for leg-wheel transformation as well as generation and transition of the legged gaits. The robot can be regarded as having “leg-level” dynamics.  The other robot introduced in this talk is the well-known RHex-style robot, where several new reduced-order models are constructed to excite the “gait level” dynamics on the robot. A simple reduced-order model, the rolling spring-loaded inverted pendulum (R-SLIP), is developed and served as the running template of the RHex-style robot. Experimental validation confirms that by merely deploying stable running gaits of the R-SLIP model on the physical robot with a simple open-loop control strategy, the robot can easily initiate its dynamic running behaviors with a flight phase and can move with similar body state profiles to those of the model, in all five speeds that were tested. By using the same methodology, a two-leg sagittal-plane model is constructed that serves as the fundamental multi-leg gait-level template. It successfully excites the bounding and pronking behaviors of the “quadruped version” of the RHex-style robot. In addition to these two models, two three-leg planar models are also constructed to develop leaping behavior and climbing behavior of the RHex-style robot, respectively.  
  • Talk: GRASP Special Seminar: Roy Fox, Hebrew University, "Optimal Selective Attention and Action in Reactive Agents"
    Date: Monday, September 15, 2014 - 2pm to 3pm
    Presenters: Roy Fox
    Alternate Location: Levine 307
    Intelligent agents, interacting with their environment, operate under constraints on what they can observe and how they can act. Unbounded agents can use standard Reinforcement Learning to optimize their inference and control under purely external constraints. Bounded agents, on the other hand, are subject to internal constraints as well. This only allows them to partially notice their observations, and to partially intend their actions, requiring rational selection of attention and action.In this talk we will see how to find the optimal information-constrained policy in reactive (memoryless) agents. We will discuss a number of reasons why internal constraints are often best modeled as bounds on information-theoretic quantities, and why we can focus on reactive agents with hardly any loss of generality. We will link the solution of the constrained problem to that of soft clustering, and present some of its nice properties, such as principled dimensionality reduction.
  • Talk: Fall 2014 GRASP Seminar - Julie Shah, Massachusetts Institute of Technology, "Integrating Robots into Team-Oriented Environments"
    Date: Friday, September 19, 2014 - 11am to 12pm
    Presenters: Julie Shah
    Recent advances in computation, sensing, and hardware enable robotics to perform an increasing percentage of traditionally manual tasks in manufacturing. Yet, often the assembly mechanic cannot be removed entirely from the process. This provides new economic motivation to explore opportunities where human workers and industrial robots may work in close physical collaboration. In this talk, I present the development of new algorithmic techniques for collaborative plan execution that scale to real-world industrial applications.  I also discuss the design of new models for robot planning, which use insights and data derived from the planning and execution strategies employed by successful human teams, to support more seamless robot participation in human work practices. This includes models for human-robot team training, which involves hands-on practice to clarify sequencing and timing of actions, and for team planning, which includes communication to negotiate and clarify allocation and sequencing of work. The aim is to support both the human and robot workers in co-developing a common understanding of task responsibilities and information requirements, to produce more effective human-robot partnerships.
  • Talk: Fall 2014 GRASP Seminar - Steve LaValle, University of Illinois at Urbana-Champaign (on leave at Oculus VR), "Virtual Reality: The Next Generation"
    Date: Friday, September 26, 2014 - 11am to 12pm
    Presenters: Steve LaValle
    Using the latest technology, we can safely hijack your most trusted senses, thereby fooling your brain into believing you are in another world. Virtual reality (VR) has been around for a long time, but due to the recent convergence of sensing, display, and computation technologies, there is an unprecedented opportunity to explore this form of human augmentation with lightweight, low-cost materials and simple software platforms. This is an intense form of human-computer interaction (HCI) that requires re-examining core engineering principles with a direct infusion of perceptual psychology research. Developing systems that optimize classical criteria might lead to overcomplicated solutions that are too slow or costly in practice, and yet could make no perceptible difference to users. Simple adaptation of techniques that were developed for on-screen viewing, such as cinematography and first-person shooter game play, often lead to unpleasent VR experiences due the presentation of unusual stimuli or due to mismatches between the human vestibular system and other senses. With the rapid rise in consumer VR, fundamental research questions are popping up everywhere, slicing across numerous disciplines from engineering to socialogy to film to medicine. This talk will provide some perspective on where we have been and where we might be going next.
  • Talk: Fall 2014 GRASP Seminar - Timothy Bretl, University of Illinois at Urbana-Champaign, "Mechanics, Manipulation, and Perception of an Elastic Rod"
    Date: Friday, October 3, 2014 - 11am to 12pm
    Presenters: Timothy Bretl
    My talk in the language of robotics: "I will show how to establish an appropriate configuration space for robotic manipulation of canonical 'deformable linear objects' like a Kirchhoff elastic rod (e.g., a flexible wire). This result leads to simple algorithms for manipulation and perception that are easy to implement and that work well in practice." My talk in the language of mathematics: "Any framed curve traced by a Kirchhoff elastic rod in static equilibrium can be described as a local solution to an optimal control problem on the special Euclidean group SE(3). By application of Lie-Poisson reduction, I will show that the set of all normal extremals for this problem is a smooth manifold of finite dimension that admits a single global chart. I will also show that the subset of all local optima is open and is locally diffeomorphic to the space of boundary conditions."
  • Talk: GRASP Special Seminar: Marcell Missura, University of Bonn in Germany, "Balanced Bipedal Walking with Capture Steps"
    Date: Monday, October 6, 2014 - 2pm to 3pm
    Presenters: Marcell Missura
    Alternate Location: Levine 307
    Bipedal Walking is a balance critical full body motion that demands fast controller response in order to maintain stability with the right steps at the right time. Aside from the complexity of the walking motion itself, sensor and actuation noise on real hardware combined with latency in the sensomotoric feedback loop make the computation of well placed footsteps a particularly challenging task. In this presentation, an inverted pendulum based method will be introduced that produces an omnidirectional walk with a good amount of stability for various disturbances. Footstep locations and step timing are computed efficiently in closed form. During the presentation you will be able to enjoy videos of a famous soccer robot defending its balance while being tortured in push experiments.
  • Talk: Fall 2014 GRASP Seminar - Torsten Kroeger, Google Inc. & Stanford University, "Highly Reactive Robot Motion Generation and Control"
    Date: Friday, October 17, 2014 - 11am to 12pm
    Presenters: Torsten Kroeger
    Online and instantaneous robot motion generation is an important feature for robot motion control systems to let robots respond instantaneously to unforeseen events. An algorithmic concept that enables instantaneous changes from sensor-guided robot motion control (e.g., force/torque or visual servo control) to trajectory-following motion control, and vice versa, will be presented. The resulting class of on-line trajectory generation algorithms serves as an intermediate layer between low-level motion control and high-level sensor-based motion planning. Online motion generation from arbitrary states is an essential feature for autonomous hybrid switched motion control systems. It enables robotic systems to perform coordinated and deterministic reflex motions. Samples and use-cases - including manipulation and human-robot interaction tasks - will accompany the talk in order to provide a comprehensible insight into this interesting and relevant field of robotics.
  • Talk: Fall 2014 GRASP Seminar - Sonia Chernova, Worcester Polytechnic Institute, "Crowds and Robots: Towards Robots that Learn from Everyday People"
    Date: Friday, October 24, 2014 - 11am to 12pm
    Presenters: Sonia Chernova
    Recent innovations in crowd computing, crowdsourcing and remote access technologies have altered the way in which many traditional artificial intelligence and robotics studies are designed, conducted and evaluated.  Research on shared autonomy, human-robot interaction and robot learning has particularly benefited from the greater access to data and users that such techniques enable, leading to new data-driven techniques and more extensive evaluations.  In this talk, I will present ongoing projects aimed at enabling robots to learn from everyday people, examining how access to thousands of potential robot users can be leveraged to overcome several challenges common in robotics research. 
  • Talk: Fall 2014 GRASP Seminar - Kostas Bekris, Rutgers University, "From Foundational Progress in Motion Planning to Credible Co-Robots"
    Date: Friday, October 31, 2014 - 11am to 12pm
    Presenters: Kostas Bekris
    Motion planning has progressed over the last couple of decades in addressing complex challenges in robotics. An important milestone was the development of practical sampling-based solutions, for which recently the conditions that allow these methods to achieve asymptotic optimality have been identified. Based on the state-of-the-art, this talk will highlight a series of recent foundational contributions by our research group in this area:   a) probabilistic near-optimality bounds for sampling-based planners after a finite amount of computation;   b) small, sparse roadmaps that can quickly return near-optimal paths based on results from graph theory;   c) optimality guarantees for kinodynamic planning, even when a steering function is not available. To solve more complex robotics problems, motion planners need to be integrated with higher-level reasoning. This talk will present a framework for the efficient rearrangement of multiple similar objects using a Baxter robot arm. The framework integrates sampling-based algorithms for manipulation with combinatorial solvers for pebble motion graph problems.  The talk will conclude on how such algorithmic, foundational progress together with technological developments, like cloud computing, new compliant arms and capable hands, bring the hope of credible co-robots in terms of i) Collaboration skills when interacting with people or other robots; ii) Resourcefulness when operating in unstructured human spaces and iii) Dexterity in interacting with the environment.
  • Talk: GRASP / ESE Seminar - Pramod Khargonekar, NSF Director of Engineering, "Challenges and Opportunities in Engineering Research and Education: A View from NSF"
    Date: Friday, November 7, 2014 - 11am to 12pm
    Presenters: Pramod Khargonekar
    The main goal of this talk is to share thoughts and perspectives on key opportunities and challenges in engineering research; education and broadening participation, and innovation ecosystem as seen from my perspective at the National Science Foundation. I will begin with a discussion of the overall context, drivers, and trends that are driving our strategies. I will then give a high level overview of some of the key priorities and initiatives in the Engineering Directorate (ENG) at NSF.  The talk will conclude with a brief discussion of certain overarching issues including broader impacts, grand challenges, and new models for interdisciplinary research.
  • Talk: GRASP Special Seminar: Stephen L. Smith, University of Waterloo, "Balancing Expressivity and Efficiency in Robot Path Planning Languages"
    Date: Tuesday, November 11, 2014 - 12pm to 1pm
    Presenters: Stephen L. Smith
    Alternate Location: Levine 512
    Robot path planners consist of a language for specifying tasks, and a solver for generating paths. There is generally a tradeoff between the expressivity of the language and efficiency of the solver. On one hand, specialized and fast planners have been developed for specific tasks including point-to-point motion and simple combinatorial tasks. On the other hand, multi-purpose planners based on linear temporal logics (LTL) can handle a wide range of robot tasks. This expressivity, however, comes at a cost, and these planners often struggle to solve large-scale problems.  This talk will present a new language for path planning problems called SAT-TSP, which seeks to strike a balance between expressivity and efficiency. The language, while not as expressive as LTL, allows a user to naturally express many motion planning problems for single and multiple robots.  The talk will summarize solver approaches for problems in this language and give a performance comparison. Several solvers show a link to the Generalized Traveling Salesman Problem (GTSP) and recent work on developing GTSP solvers will also be discussed.
  • Talk: Fall 2014 GRASP Seminar - Kris Hauser, Duke University, "Practical Robot Motion Planning"
    Date: Friday, November 14, 2014 - 11am to 12pm
    Presenters: Kris Hauser
    Motion planning -- the problem of computing physical actions to complete a specified task -- has inspired some of the most theoretically rigorous and beautiful results in robotics research.  But as robots proliferate in real-world applications like household service, driverless cars, warehouse automation, minimally-invasive surgery, search-and-rescue, and unmanned aerial vehicles, we are beginning to see the classical theory fall behind.  The clean assumptions of theory are at odds with the dirty reality: robots must handle large amounts of noisy sensor data, uncertainty, underspecified models, nonlinear and hysteretic dynamic effects, exotic objective functions and constraints, and real-time demands.  This talk will present recent efforts to bring motion planners to bear on real robots, in the context of three projects: 1) ladder climbing in the DARPA Robotics Challenge; 2) intelligent user interfaces for human-operated robots; and 3) navigation amongst many moving obstacles.  I will present new planning algorithms and architectures whose performance is backed both by theoretical guarantees and empirical evaluation.
  • Talk: Fall 2014 GRASP Seminar - Hong Zhang, University of Alberta, "Scalable Visual Mapping"
    Date: Friday, November 21, 2014 - 11am to 12pm
    Presenters: Hong Zhang
    One area of significant progress in robotics research in recent years has been the use of visual sensing by mobile robots to map an environment and negotiate routes within the environment. This growth is mostly attributed to the rich textural information in visual sensory data compared with the traditional range data, as well as to the great strides made in the development of efficient and robust computer vision and image processing algorithms. In this talk, I will give an overview of the recent progress in appearance-based visual SLAM (simultaneous localization and mapping) algorithms, and highlight our research in scalable and accurate visual SLAM algorithms using whole image descriptors. As well, I will describe our current work in dealing with dynamically changing environments, and our involvement in the NSERC Canadian Field Robotics Network (NCFRN).
  • Talk: GRASP Special Seminar: Peter Henry, University of Washington, "Mapping and Modeling with RGB-D Cameras"
    Date: Monday, December 1, 2014 - 11am to 12pm
    Presenters: Peter Henry
    Alternate Location: Levine 307
    In the past few years, we have seen the availability of low-cost RGB-D cameras, such as the Microsoft Kinect and Asus Xtion Pro Live, which deliver high framerate dense depth information in addition to color.  My research is focused on how these cameras may be used for 3D camera tracking, mapping, and the creation of high-quality models of indoor environments.  I will also talk about some applications of my work to interactive modeling, quadcoptor mapping, and object discovery. Primary papers: RGB-D Mapping: Using Kinect-style Depth Cameras for Dense 3D Modeling of Indoor Environments.  IJRR 2012. http://homes.cs.washington.edu/~peter/papers/3d-mapping-ijrr-12-preprint.pdf Patch Volumes: Segmentation-based Consistent Mapping with RGB-D Cameras.  3DV 2013. http://homes.cs.washington.edu/~peter/papers/patch-volumes-3dv-2013.pdf Toward Online 3-D Object Segmentation and Mapping.  ICRA 2014. http://homes.cs.washington.edu/~peter/papers/icra-2014-herbst.pdf Other papers available at: http://homes.cs.washington.edu/~peter/
  • Talk: GRASP Special Seminar: Brian Kulis, Ohio State University, "Small-Variance Asymptotics for Large-Scale Learning"
    Date: Tuesday, December 2, 2014 - 2pm to 3pm
    Presenters: Brian Kulis
    Alternate Location: Levine 307
    It is widely known that the Gaussian mixture model is related to k-means by “small-variance asymptotics”: as the covariances of the clusters shrink, the EM algorithm approaches the k-means algorithm and the negative log-likelihood approaches the k-means objective. Similar asymptotic connections exist for other machine learning models, including dimensionality reduction (probabilistic PCA becomes PCA), multiview learning (probabilistic CCA becomes CCA), and classification (a restricted Bayes optimal classifier becomes the SVM). The asymptotic non-probabilistic counterparts to the probabilistic models are almost always more scalable, and are typically easier to analyze, making them useful alternatives to the probabilistic models in many situations. I will explore how we can extend such asymptotics to a richer class of probabilistic models, with a focus on large-scale graphical models, Bayesian nonparametric models, and time-series data. I will develop the necessary mathematical tools needed for these extensions and will describe a framework for designing scalable optimization problems derived from the rich probabilistic models. Applications are diverse, and include topic modeling, network evolution, and deep feature learning.
  • Talk: GRASP Special Seminar: Katerina Fragkiadaki, UC Berkeley (GRASP Alum), "Learning to Parse Videos"
    Date: Wednesday, December 3, 2014 - 3pm to 4pm
    Presenters: Katerina Fragkiadaki
    Alternate Location: Levine 307
    In this talk, we look into the problem of segmenting, tracking, and  extracting 3D time varying shape and camera poses for non-rigid objects in monocular videos.  Our method segments and tracks objects and their parts using past segmentation and tracking experience from a training set, and uses the segmented point trajectories of each object to extract 3D shape assuming a low-rank shape prior.  We segment using motion boundaries and learnt saliency detection, and outperform by a margin the previous state-of-the-art in challenging video scenes.  We ``learn to track’’ using a novel tracking loss in a distance learning framework, and  outperform color and texture histograms as well as deep feature matching learnt from Imagenet Classification or Detection tasks.  We extract dense 3D object models from realistic monocular videos, a problem typically studied with lab acquired datasets, pre-segmented objects and oracle trajectories.
  • Talk: GRASP Special Seminar: Matei Ciocarlie, Columbia University, "Versatility in Robotic Manipulation: the Long Road to Everywhere"
    Date: Thursday, December 4, 2014 - 11am to 12pm
    Presenters: Matei Ciocarlie
    Alternate Location: Levine 307
    In this talk, I will present a number of methods for increasing the versatility of mobile manipulators, from novel hand designs and grasp planning algorithms to Human-in-the-Loop manipulation and its applications in assistive robotics. I will introduce the concept of eigengrasps as the bases of a linear hand posture subspace, and use it to show that, from a grasp planning perspective, a hand does not have to be complex in order to be versatile. I will then present the Velo Gripper, a novel underactuated design that uses passive adaptation to the object and the environment to complement the traditional sense-plan-and-act loop. Beyond the end-effector, I will present sensing and planning methods for manipulation under uncertainty and through clutter. Finally, I will present our results in Human-in-the-Loop manipulation, enabling non-experts to operate robots performing complex tasks in difficult environments. Among other applications, this approach can enable mobile robots to assist people with motor impairments. I will present results where our robot, operated by a motor impaired person through a head-tracker single-button mouse, performed varied and unscripted manipulation tasks in a real home, and acted as a medium for social interaction.
  • Talk: Fall 2014 GRASP Seminar - Sertac Karaman, Massachusetts Institute of Technology, "On High-performance Motion in Stochastic Environments"
    Date: Friday, December 5, 2014 - 11am to 12pm
    Presenters: Sertac Karaman
    Consider a robotic vehicle that is traveling in a cluttered environment or attempting to fulfill tasks that require visiting several locations. What is the maximum speed that this vehicle can achieve and maintain for a long time? How does this speed depend on the agility, perception, actuation, or computation capabilities of the vehicle? To answer these questions, we formulate various control and planning problems in stochastic obstacle fields or stochastic reward fields; subsequently, we establish novel connections between these problems and suitable fundamental problems of statistical mechanics. In particular, we point out critical phenomena, phase transitions, and universality classes in certain control and planning problems. With the help of these results, we propose efficient algorithms with provable performance guarantees.
  • Talk: GRASP / PRECISE Seminar: Nuno Martins, University of Maryland, "Remote and Distributed Estimation over Shared Networks: New Results and Open Problems"
    Date: Wednesday, December 10, 2014 - 11am to 12pm
    Presenters: Nuno Martins
    Alternate Location: Levine 307
    This talk will focus on the design of distributed estimation systems that are formed by multiple non-collocated components. A shared network is used to disseminate information among the components. I will discuss two recent results: Assuming that the network is characterized by an incomplete directed communication graph, the first result characterizes the existence of omniscience-achieving schemes for which all components that observe only a portion of the output of an underlying plant can estimate the entire state with error that vanishes asymptotically. Our approach hinges on key concepts from decentralized control that are systematic and constructive. The second result characterizes the structure of certain optimal policies for the case in which the number of components exceeds the maximal number of simultaneous transmissions that the network can accept. In order to obtain a tractable framework for which design principles can be characterized analytically, I will consider the case in which there are two estimators that rely on information sent to them by two sensors that access dissimilar measurements. I will show the optimality of certain threshold-based policies, establish a connection with a problem of optimal quantization for which the distortion is non-uniform across representation symbols, present numerical approaches, discuss interpretations of the results and list related open issues.