TechTalks from event: Technical session talks from ICRA 2012

Conference registration code to access these videos can be accessed by visiting this link: PaperPlaza. Step-by-step to access these videos are here: step-by-step process .
Why some of the videos are missing? If you had provided your consent form for your video to be published and still it is missing, please contact support@techtalks.tv

Learning and Adaptation Control of Robotic Systems II

  • Online Learning of Varying Stiffness through Physical Human-Robot Interaction Authors: Kronander, Klas; Billard, Aude
    Programming by Demonstration offers an intuitive framework for teaching robots how to perform various tasks without having to preprogram them. It also offers an intuitive way to provide corrections and refine teaching during task execution. Previously, mostly position constraints have been taken into account when teaching tasks from demonstrations. In this work, we tackle the problem of teaching tasks that require or can benefit from varying stiffness. This extension is not trivial, as the teacher needs to have a way of communicating to the robot what stiffness it should use. We propose a method by which the teacher can modulate the stiffness of the robot in any direction through physical interaction. The system is incremental and works online, so that the teacher can instantly feel how the robot learns from the interaction. We validate the proposed approach on two experiments on a 7-Dof Barrett WAM arm.
  • Reinforcement Planning: RL for Optimal Planners Authors: Zucker, Matthew; Bagnell, James
    Search based planners such as A* and Dijkstra’s algorithm are proven methods for guiding today’s robotic systems. Although such planners are typically based upon a coarse approximation of reality, they are nonetheless valuable due to their ability to reason about the future, and to generalize to previously unseen scenarios. However, encoding the desired behavior of a system into the underlying cost function used by the planner can be a tedious and error-prone task. We introduce Reinforcement Planning, which extends gradient based reinforcement learning algorithms to automatically learn useful surrogate cost functions for optimal planners. Reinforcement Planning presents several advantages over other learning approaches to planning in that it is not limited by the expertise of a human demonstrator, and that it acknowledges the domain of the planner is a simplified model of the world. We demonstrate the effectiveness of our method in learning to solve a noisy physical simulation of the well-known “marble maze” toy.
  • Adaptive Collaborative Estimation of Multi-Agent Mobile Robotic Systems Authors: Nestinger, Stephen; Demetriou, Michael
    Collaborative multi-robot systems are used in a vast array of fields for their innate ability to parallelize domain problems for faster execution. These systems are generally comprised of multiple identical robotic systems in order to simplify manufacturability and programmability, reduce cost, and provide fault tolerance. This work takes advantage of the homogeneity and multiplicity of multi-robot systems to enhance the convergence rate of adaptive dynamic parameter estimation through collaboration. The collaborative adaptive dynamic parameter estimation of multi-robot systems is accomplished by penalizing the pair-wise disagreement of both state and parameter estimates. Consensus and convergence is based on Lyapunov stability arguments. Simulation studies with multiple Pioneer 3-DX systems provides verification of the proposed theoretic collaborative adaptive parameter estimation predictions.
  • Lingodroids: Learning Terms for Time Authors: Heath, Scott Christopher; Schulz, Ruth; Ball, David; Wiles, Janet
    For humans and robots to communicate using natural language it is necessary for the robots to develop concepts and associated terms that correspond to the human use of words. Time and space are foundational concepts in human language, and to develop a set of words that correspond to human notions of time and space, it is necessary to take into account the way that they are used in natural human conversations, where terms and phrases such as ‘soon’, ‘in a while’, or ‘near’ are often used. We present language learning robots called Lingodroids that can learn and use simple terms for time and space. In previous work, the Lingodroids were able to learn terms for space. In this work we extend their abilities by adding temporal variables which allow them to learn terms for time. The robots build their own maps of the world and interact socially to form a shared lexicon for location and duration terms. The robots successfully use the shared lexicons to communicate places and times to meet again.
  • Teaching Nullspace Constraints in Physical Human-Robot Interaction Using Reservoir Computing Authors: Nordmann, Arne; Rüther, Stefan; Wrede, Sebastian; Steil, Jochen J.
    A major goal of current robotics research is to enable robots to become co-workers that collaborate with humans efficiently and adapt to changing environments or workflows. We present an approach utilizing the physical interaction capabilities of compliant robots with data-driven and model-free learning in a coherent system in order to make fast reconfiguration of redundant robots feasible. Users with no particular robotics knowledge can perform this task in physical interaction with the compliant robot, for example to reconfigure a work cell due to changes in the environment. For fast and efficient training of the respective mapping, an associative reservoir neural network is employed. It is embedded in the motion controller of the system, hence allowing for execution of arbitrary motions in task space. We describe the training, exploration and the control architecture of the systems as well as present an evaluation on the KUKA Light-Weight Robot. Our results show that the learned model solves the redundancy resolution problem under the given constraints with sufficient accuracy and generalizes to generate valid joint-space trajectories even in untrained areas of the workspace.
  • A Bayesian Nonparametric Approach to Modeling Battery Health Authors: Joseph, Joshua; Doshi, Finale; Roy, Nicholas
    The batteries of many consumer products are often both a substantial portion of the item's cost and commonly a first point of failure. Accurately predicting remaining battery life can lower costs by reducing unnecessary battery replacements. Unfortunately, battery dynamics are extremely complex, and we often lack the domain knowledge required to construct a model by hand. In this work, we take a data-driven approach and aim to learn a model of battery time-to-death from training data. Using a Dirichlet process prior over mixture weights, we learn an infinite mixture model for battery health. The Bayesian aspect of our model helps to avoid over-fitting while the nonparametric nature of the model allows the data to control the size of the model, preventing under-fitting. We demonstrate our model's effectiveness by making time-to-death predictions using real data from iRobot Roomba batteries.

Parts Handling and Manipulation

  • Design of Parts Handling and Gear Assembling Device Authors: Yamaguchi, Kengo; Hirata, Yasuhisa; Kaisumi, Aya; Kosuge, Kazuhiro
    Many one-degree-of-freedom (1-DOF) grippers have been used in factories. This paper focuses on the design of the 1-DOF parts handling device for picking up small objects robustly and agilely and realizing assembly tasks. In our conventional research, we proposed a concept for the handling device, which cages an object without letting the object escape from its tips before closing them completely and then grasps the object robustly at a unique position of the tips. In this paper, we propose a method for designing the shape of the device's tips by considering not only the caging and self-alignment of the object but also the gear assembly task. We also develop the robust and agile pick-up device (RAPiD) with tips designed by the new method and present experimental results that illustrate the ability of RAPiD to handle and assemble gears.
  • Optimal Admittance Characteristics for Planar Force-Assembly of Convex Polygonal Parts Authors: Wiemer, Steven; Schimmels, Joseph
    Robots are not typically used for assembly tasks in which positioning requirements exceed robot capabilities. To address this limitation, a significant amount of work has been directed toward identifying desirable mechanical behavior of a robot for force-guided assembly. Most of this work has been directed toward the `standard' peg-in-hole assembly problem. Little has been done to identify the specific behavior necessary for reliable assembly for different types of polygonal parts, and little has been done relating assembly characteristics to classes of part geometries. This paper presents the best passive admittance and associated maximum coefficient of friction for planar force-assembly of a variety of different polygonal parts, specifically pegs with rectangular, trapezoidal, triangular, and pentagonal cross sections. The results show that force-guided assembly can be reliably achieved at higher values of friction when parts are shorter and wider. For all geometries considered, force-guided assembly is ensured for any value of friction less than 0.8 when the optimal admittance is used; and, for some geometries, for any value of friction less than 15.
  • The Effect of Anisotropic Friction on Vibratory Velocity Fields Authors: Umbanhowar, Paul; Vose, Thomas; Mitani, Atsushi; Hirai, Shinichi; Lynch, Kevin
    This paper explores the role of anisotropic friction properties in vibratory parts manipulation. We show that direction-dependent surface friction properties can be used in conjunction with a vibrating plate to help design friction-induced velocity fields on the surface of the plate. Theoretical, simulation, and experimental results are presented quantifying the anisotropic friction effects of textured surfaces such as micromachined silicon and fabrics.
  • Sparse Spatial Coding: A Novel Approach for Efficient and Accurate Object Recognition Authors: Leivas, Gabriel; Nascimento, Erickson; Wilson Vieira, Antonio; Campos, Mario Montenegro
    Successful state-of-the-art object recognition techniques from images have been based on powerful methods, such as sparse representation, in order to replace the also popular vector quantization (VQ) approach. Recently, sparse coding, which is characterized by representing a signal in a sparse space, has raised the bar on several object recognition benchmarks. However, one serious drawback of sparse space based methods is that similar local features can be quantized into different visual words. We present in this paper a new method, called Sparse Spatial Coding (SSC), which combines a sparse coding dictionary learning, a spatial constraint coding stage and an online classification method to improve object recognition. An efficient new off-line classification algorithm is also presented. We overcome the problem of techniques which make use of sparse representation alone by generating the final representation with SSC and max pooling, presented for an online learning classifier. Experimental results obtained on the Caltech 101, Caltech 256, Corel 5000 and Corel 10000 databases, show that, to the best of our knowledge, our approach supersedes in accuracy the best published results to date on the same databases. As an extension, we also show high performance results on the MIT-67 indoor scene recognition dataset.
  • Humanoid's Dual Arm Object Manipulation Based on Virtual Dynamics Model Authors: Shin, Sung Yul; Lee, Jun won; Kim, ChangHwan
    In order to implement promising robot applications in our daily lives, robots need to perform manipulation tasks within the human environments. Especially for a humanoid robot, it is essential to manipulate a variety of objects with different shapes and sizes to assist humans in the human environments. This paper presents a method of manipulating objects with humanoid robot's dual arms. The robot is usually asked to control both the motion and force to manipulate the objects. We propose a novel concept of control method based on the virtual dynamics model (VDM), which enables the robot to perform both tasks of reaching to an object and grasping it under the uniform control system. Furthermore, the impedance model based on the VDM controller also enables the robot to safely grasp an object by reducing the impact at the contact point. The proposed algorithm is implemented on the humanoid robot, Mahru, with independent joint controller at each motor. Its performance is demonstrated by manipulating different types of objects.
  • A Kernel-Based Approach to Direct Action Perception Authors: Kroemer, Oliver; Ugur, Emre; Oztop, Erhan; Peters, Jan
    The direct perception of actions allows a robot to predict the afforded actions of observed objects. In this paper, we present a non-parametric approach to representing the affordance-bearing subparts of objects. This representation forms the basis of a kernel function for computing the similarity between different subparts. Using this kernel function, together with motor primitive actions, the robot can learn the required mappings to perform direct action perception. The proposed approach was successfully implemented on a real robot, which could then quickly learn to generalize grasping and pouring actions to novel objects.

Space Robotics

  • Automatic Rock Recognition from Drilling Performance Data Authors: Zhou, Hang; Hatherly, Peter; Monteiro, Sildomar; Ramos, Fabio; Oppolzer, Florian; Nettleton, Eric; Scheding, Steven
    Automated rock recognition is a key step for building a fully autonomous mine. When characterizing rock types from drill performance data, the main challenge is that there is not an obvious one-to-one correspondence between the two. In this paper, a hybrid rock recognition approach is proposed which combines Gaussian Process (GP) regression with clustering. Drill performance data is also known as Measurement While Drilling (MWD) data and a rock hardness measure - Adjusted Penetration Rate (APR) is extracted using the raw data in discrete drill holes. GP regression is then applied to create a more dense APR distribution, followed by clustering which produces discrete class labels. No initial labeling is needed. Comparisons are made with alternative measures of rock hardness from MWD data as well as state-of-the-art GP classification. Experimental results from an actual mine site show the effectiveness of our proposed approach.
  • Evaluation of the Reconfiguration Effects of Planetary Rovers on their Lateral Traversing of Sandy Slopes Authors: Inotsume, Hiroaki; Sutoh, Masataku; Nagatani, Keiji; Yoshida, Kazuya
    Rovers that are used to explore craters on the Moon or Mars require the mobility to negotiate sandy slopes, on which slippage can easily occur. Such slippage can be reduced by actively readjusting the attitude of the rovers. By changing attitude, rovers can modify the position of their center of gravity and the wheel-soil contact angle. In this study, we discuss the effects of attitude changes on downhill sideslip based on the slope failure mechanism and experiments on reconfiguring the rover attitude and wheel angles. We conducted slope-traversing experiments using a wheeled rover under various roll angles and wheel angles. The experimental results show that the contact angle between wheels and slopes has a dominant influence on sideslip when compared with that of readjusting the rover's center of gravity.
  • Evaluation of Influence of Surface Shape of Locomotion Mechanism on Traveling Performance of Planetary Rovers Authors: Sutoh, Masataku; Nagaoka, Kenji; Nagatani, Keiji; Yoshida, Kazuya
    The surfaces of both the Moon and Mars are covered with loose soil, with numerous steep slopes along their crater rims. Therefore, one of the most important requirements imposed on planetary rovers is their ability to minimize slippage while climbing steep slopes, i.e., the ability to generate a drawbar pull with only a small amount of slippage. To this end, the wheels/tracks of planetary rovers typically have parallel fins called lugs (i.e., grousers) on their surface. Recent studies have reported that these lugs can substantially improve the traveling performances of planetary rovers. Therefore, in this study, we conducted experiments using lightweight two-wheeled and mono-tracked rovers to provide a quantitative confirmation regarding the influence of lugs on the traveling performances of planetary rovers. Based on our experimental results, we confirmed that, although an increase in the number of lugs contributes to the high traveling performance of wheeled rovers, it does not contribute much to that of tracked rovers. Furthermore, an increase in lug height improves the traveling performances of both types of rovers.
  • The Robonaut 2 Hand Designed to Do Work with Tools Authors: Bridgwater, Lyndon; Ihrke, Chris; Diftler, Myron; Abdallah, Muhammad; Radford, Nicolaus; Rogers, Jonathan; yayathi, Sandeep; Askew, Roger, Scott; Linn, Marty
    The second generation Robonaut hand has many advantages over its predecessor. This mechatronic device is more dexterous and has improved force control and sensing giving it the capability to grasp and actuate a wider range of tools. It can achieve higher peak forces at higher speeds than the original. Developed as part of a partnership between General Motors and NASA, the hand is designed to more closely approximate a human hand. Having a more anthropomorphic design allows the hand to attain a larger set of useful grasps for working with human interfaces. Key to the hands improved performance is the use of lower friction drive elements and a redistribution of components from the hand to the forearm, permitting more sensing in the fingers and palm where it is most important. The following describes the design, mechanical/electrical integration, and control features of the hand. Lessons learned during the development and initial operations along with planned refinements to make it more effective are presented.
  • Autonomous Detection of Volcanic Plumes on Outer Planetary Bodies Authors: Lin, Yucong; Bunte, Melissa; Saripalli, Srikanth; Greeley, Ronald
    We experimentally evaluated the efficacy of var- ious autonomous supervised classification techniques for de- tecting transient geophysical phenomena. We demonstrated methods of detecting volcanic plumes on the planetary satellites Io and Enceladus using spacecraft images from the Voyager, Galileo, New Horizons, and Cassini missions. We successfully detected 73-95% of known plumes in images from all four mission datasets. We increased the detection rate by using a training subset. Additionally, we showed that the same tech- niques are applicable to differentiating geologic features, such as plumes and mountains, which exhibit similar appearances in images.
  • Gravity-Independent Mobility and Drilling on Natural Rock Using Microspines Authors: Parness, Aaron; Frost, Matthew; Thatte, Nitish; King, Jonathan
    To grip rocks on the surfaces of asteroids and comets, and to grip the cliff faces and lava tubes of Mars, a 250 mm diameter omni-directional anchor is presented that utilizes a hierarchical array of claws with suspension flexures, called microspines, to create fast, strong attachment. Prototypes have been demonstrated on vesicular basalt and a‘a lava rock supporting forces in all directions away from the rock. Each anchor can support >160 N tangent, >150 N at 45, and >180 N normal to the surface of the rock. A two-actuator selectively-compliant ankle interfaces these anchors to the Lemur IIB robot for climbing trials. A rotary percussive drill was also integrated into the anchor, demonstrating self-contained rock coring regardless of gravitational orientation. As a harder-than-zero-g proof of concept, 20mm diameter boreholes were drilled 83 mm deep in vesicular basalt samples, retaining a 12 mm diameter rock core in 3-6 pieces while in an inverted configuration, literally drilling into the ceiling.