List of all recorded talks

  • An Evaluation of Sampling Path Strategies for an Autonomous Underwater Vehicle Authors: Ho, Colin; Mora, Andres; Saripalli, Srikanth
    A critical problem in planning sampling paths for autonomous underwater vehicles is balancing obtaining an accurate scalar field estimation against efficiently utilizing the stored energy capacity of the sampling vehicle. Adaptive sampling approaches can only provide solutions when real-time and a priori environmental data is available. Through utilizing a cost-evaluation function to experimentally evaluate various sampling path strategies for a wide range of scalar fields and sampling densities, it is found that a systematic spiral sampling path strategy is optimal for high-variance scalar fields for all sampling densities and low-variance scalar fields when sampling is sparse. The random spiral sampling path strategy is found to be optimal for low-variance scalar fields when sampling is dense.
  • Field Performance Evaluation of New Methods for In-Situ Calibration of Attitude and Doppler Sensors for Underwater Vehicle Navigation Authors: Troni, Giancarlo; Whitcomb, Louis
    We report a comparative performance evaluation, using at-sea field data, of recently reported methods for the problem of in-situ calibration of the alignment rotation matrix between Doppler sonar velocity sensors and inertial navigation sensors arising in the navigation of underwater vehicles. Most previously reported solutions to this alignment calibration problem require the use of absolute navigation fixes of the underwater vehicle, thus requiring additional navigation sensors and/or beacons to be located externally and apart from the underwater vehicle. We briefly review four recently reported alignment calibration methods employing only internal vehicle navigation sensors for velocity, acceleration, attitude, and depth. We report the results of comparative analysis of the performance of these recently reported methods and a previously reported method with navigation data from deep-water survey missions of the Sentry autonomous underwater vehicle conducted in March, 2011 in the Kermadec Arc in the Southern Pacific Ocean. The results reveal consistent differences in performance of the various methods when analyzed on navigation data from several different vehicle dives.
  • A Bio-Inspired Compliant Robotic Fish: Design and Experiments Authors: EL DAOU, Hadi; Salumae, Taavi; Toming, Gert; Kruusmaa, Maarja
    This paper studies the modelling, design and fabrication of a bio-inspired fish-like robot propelled by a compliant body. The key to the design is the use of a single motor to actuate the compliant body and to generate thrust. The robot has the same geometrical properties of a subcarangiform swimmer with the same length. The design is based on rigid head and fin linked together with a compliant body. The flexible part is modelled as a non-uniform cantilever beam actuated by a concentrated moment. The dynamics of the compliant body are studied and a relationship between the applied moment and the resulting motion is derived. A prototype that implements the proposed approach is built. Experiments on the prototype are done to identify the model parameters and to validate the theoretical modelling.
  • Ray-Tracing Codec for Structured Light 3D Camera Authors: Bui, Lam Quang; Lee, Sukhan
    In this paper, we present a new method for decoding pixel correspondences in structured light based 3D reconstruction, refer to here as Ray-Tracing codec. The key idea of Ray-Tracing codec is to correctly define the region boundaries in real number, for each layer of the Hierarchical Orthogonal Code (HOC) based on an accurate boundary estimator, and to inherit the correct region boundaries between layers sharing common boundaries. Furthermore, each region in lower layer is traced back to the upper layer for the correct correspondence between regions. This is an improvement over existing HOC decoding algorithms as the wrong decoded pixel correspondences can be greatly reduced. The experimental results have shown that the proposed Ray-Tracing codec significantly enhances the robustness and precision in depth imaging, compare with HOC and other well-known conventional approach. The proposed approach opens a greater feasibility of applying structured light based depth imaging to a 3D modeling of cluttered workspace for home service robots.
  • Coverage Optimized Active Learning for K-NN Classifiers Authors: Joshi, Ajay; Porikli, Fatih; Papanikolopoulos, Nikos
    Fast image recognition and classification is extremely important in various robotics applications such as exploration, rescue, localization, etc. k-nearest neighbor (kNN) classifiers are popular tools used in classification since they involve no explicit training phase, and are simple to implement. However, they often require large amounts of training data to work well in practice. In this paper, we propose a batch mode active learning algorithm for efficient training of kNN classifiers, that substantially reduces the amount of training required. As opposed to much previous work on iterative single sample active selection, the proposed system selects samples in batches. We propose a coverage formulation that enforces selected samples to be distributed such that all data points have labeled samples at a bounded maximum distance, given the training budget, so that there are labeled neighbors in a small neighborhood of each point. Using submodular function optimization, the proposed algorithm presents a near optimal selection strategy for an otherwise intractable problem. Further we employ uncertainty sampling along with coverage to incorporate model information and improve classification. Finally, we employ locality sensitive hashing for fast retrieval of nearest neighbors during classification, which provides 1-2 orders of magnitude speedups thus allowing real-time classification with large datasets.
  • Tissue Stiffness Simulation and Abnormality Localization Using Pseudo-Haptic Feedback Authors: Li, Min; Liu, Hongbin; Li, Jichun; Seneviratne, lakmal; Althoefer, Kaspar
    This paper introduces a new and low-cost tissue stiffness simulation technique for surgical training and robot-assisted minimally invasive surgery (RMIS) with pseudo-haptic feedback based on tissue stiffness maps provided by rolling mechanical imaging. Superficial palpation and deep palpation pseudo-haptic simulation methods are presented. Although without expensive haptic interfaces users receive only visual feedback (pseudo-haptics) when maneuvering a cursor over the surface of a virtual soft-tissue organ by means of an input device such as a mouse, a joystick, or a touch-sensitive tablet, the alterations to the cursor behavior induced by the method creates the experience of actual interaction with a tumor in the users’ minds. The proposed methods are experimentally evaluated for tissue abnormality identification. It is shown that users can recognize tumors with these two methods and the rate of correctly recognized tumors in deep palpation pseudo-haptic simulation is higher than superficial palpation simulation.
  • Prescribed Performance Tracking for Flexible Joint Robots with Unknown Dynamics and Elasticity Authors: Kostarigka, Artemis; Doulgeri, Zoe; Rovithakis, George
    In this paper a novel type of tracking controller for flexible joint robots is proposed. Joint elasticity is considered unknown and may be time varying. Robot and motor dynamics are also considered unknown. The controller guarantees link position performance specifications that have been a-priori set utilizing full state feedback. Simulation on a two link flexible joint robot validate the efficiency of the proposed control approach.
  • A Combined Potential Function and Graph Search Approach for Free Gait Generation of Quadruped Robots Authors: Geva, Yam; Shapiro, Amir
    This paper presents an algorithm for planning the foothold positions of quadruped robots on irregular terrain. The input to the algorithm is the robot kinematics, the terrain geometry, a required motion path, as well as initial posture. Our goal is to develop general algorithm that navigate quadruped robots quasi-statically over rough terrain, using an APF (Artificial Potential Field) and graph searching. The algorithm is planning a sequence set of footholds that navigates the robot along the required path with controllable motion characteristics. Simulations results demonstrate the algorithm in a planner environment.
  • Lateral and Feedback Schemes for the Inhibition of False-positive Responses in Edge Orientation Channels Authors: Park, Young-Bin; Suh, Il Hong
    Object recognition is one of the most important applications of robotics. For object recognition, edge orientation is widely used as a primitive visual feature. However, a classical filter-based approach passes not only edges inside target orientation band but also edges outside. This can thus cause problem in the estimation of the true orientation of edge. This study proposes a filtering scheme to reduce the false-positive responses, i.e. edges outside target orientation band, and investigate a solution inspired by biological vision. Motivated by several psychophysical and neuro-physiological findings, we present a computational framework based on the basic mechanisms of cortical processing, i.e. feed-forward, lateral and feedback stages. In the feed-forward stage, our model uses a classical filter-based method to allow as many true orientation edges to pass through as possible. False responses in orientation channels are then inhibited by lateral interaction. The remaining undesired responses are suppressed through the feedback stage. We evaluated the performance of our model against classical filter-based methods such as Gabor and Neumann filtering using several artificial and natural images. The results validated the effectiveness of our approach.
  • 3DNet: Large-Scale Object Class Recognition from CAD Models Authors: Wohlkinger, Walter; Aldoma, Aitor; Rusu, Radu Bogdan; Vincze, Markus
    3D object and object class recognition gained momentum with the arrival of low-cost RGB-D sensors and enables robotics tasks not feasible years ago. Scaling object class recognition to hundreds of classes still requires extensive time and many objects for learning. To overcome the training issue, we introduce a methodology for learning 3D descriptors from synthetic CAD-models and classification of never-before-seen objects at the first glance, where classification rates and speed are suited for robotics tasks. We provide this in 3DNet (, a free resource for object class recognition and pose estimation from point cloud data. 3DNet provides a large-scale hierarchical CAD-model databases with increasing numbers of classes and difficulty with 10, 50, 100 and 200 object classes together with evaluation datasets that contain thousands of scenes captured with a RGB-D sensor. 3DNet further provides an open-source framework based on PCL for testing new descriptors and benchmarking of state-of-the-art descriptors together with pose estimation procedures to enable robotics tasks such as search and grasping.