LARS 2009

Papers

"Path Recognition for Outdoor Navigation"

• Submission No. 18
• Authors: Denis Wolf and Patrick Shinzato
• Affiliation: Universidade de Sao Paulo, Brazil
• Keywords: Mobile Robotics, Navigation
• Abstract: Navigation is a broad topic that has been receiving considerable attention from the mobile robotic community. In order to execute a safe navigation on outdoors it is necessary to identify parts of the terrain than can traversed by the robot and parts that should be avoided. This paper describes an analyses of a image-based terrain identification based on different features of visual information. Experimental tests using an outdoor robot and a video camera have been conducted in real scenarios to evaluate the proposed methods.
• Session: No. 1
• Presentation slides

"Terrain Mapping and Classification Using Support Vector Machines"

• Submission No. 19
• Authors: Alberto Yukinobu Hata and Denis Fernando Wolf
• Affiliation: Universidade de Sao Paulo, Brazil
• Keywords: Mobile Robots, Environment Mapping, Terrain Classification, Laser Range Finder
• Abstract: This paper describes a three-dimensional terrain mapping and classification technique to allow the operation of mobile robots in outdoor environments using laser range finders. We used Support Vector Machines to classify portions of mapped terrain into navigable, partially navigable and non-navigable, in order to detect safe places to robot traverse. Furtherly this classifier can be used to assist the robot navigation in unstructured lands. Experimental results obtained using real environments and robot could show the efficiency of the presented methods.
• Session: No. 1
• Presentation slides

"Image Skeletonizing Method applied to generation of Topological Maps"

• Submission No. 17
• Authors: Janderson R. Oliveira and Roseli A. F. Romero
• Affiliation: Universidade de Sao Paulo, Brazil
• Keywords: metric mapping, topological mapping, occupancy grid, image skeletonizing
• Abstract: The environment mapping has been a great challenge of many researchers of the mobile robot autonomous navigation area. There are two important paradigms for mapping, metric and topological mapping. Several methods that combine the advantages of each paradigm have been proposed. Thereby the aim of this work is to provide an alternative hybrid approach, simple and efficient, for the construction of topological maps in indoor environments. For this, the proposal is the combination of two techniques found in the literature, one of them is a technique known as Occupation Grid, and the other is an image skeletonization method used here for topological mapping. Several experiments have been performed for verifying the efficiency of the proposed approach. The results obtained by junction of these two techniques demonstrate that image skeletonization technique is simple and feasible for obtaining the topological lines corresponding to free space of the environment.
• Session: No. 1
• Presentation slides

"A Comparison of Gradient versus Color and Texture Analysis for Lane Detection and Tracking"

• Submission No. 33
• Authors: Rodolfo Tapia-Espinoza and Miguel Torres-Torriti
• Affiliation: Pontificia Universidad Catolica de Chile
• Keywords: lane detection, lane tracking, texture segmentation, Gabor filters, steerable filters
• Abstract: Accurate lane detection in real-time is a critical task in autonomous vehicle guidance and lane departure warning for driver assistance. Existing vision-based approaches rely mostly on some analysis of the spatial gradient of the image. However, if the road structure is not regular and well delimited, edges may not be easy to extract and other features must be employed. This paper evaluates the use of color and textural features as a way to improve the standard gradient-based lane detection. Textural features are generated using a bank of Gabor filters. A benefit of using color and texture is that the sky regions of the image, as well as side elements, can be detected. The results obtained from testing the approaches on city roads show that color and texture analysis yields a more accurate road segmentation.
• Session: No. 1
• Presentation slides

"Comparison of Panoramic Stereoscopic Sensors Based on Hyperboloidal Mirrors"

• Submission No. 34
• Authors: Felipe Bravo-Valenzuela and Miguel Torres-Torriti
• Affiliation: Pontificia Universidad Catolica de Chile
• Keywords: panoramic stereoscopy, catadioptric sensor, omnidirectional hyperolic vision sensor, omnidirectional stereoscopy
• Abstract: Panoramic distance maps can provide valuable information for mobile robot navigation. An approach to obtain such distance maps is to implement a stereocopic system using two hyperboloidal catadioptric sensors. This paper analyzes the effects over distance estimation of two typical arrangements of the mirrors. In one of the arrangements the mirrors are placed along a vertical axis, while in the other configuration the sensors are located on plane. Both configurations may yield similar distance estimation errors for an adequately chosen baseline and given measurement range of interest. However, depending on the application, some aspects can make one arrangement preferable over the other. In the vertical configuration, the epipolar lines correspond to radial lines in the omnidirectional image, and hence, the stereoscopic correspondance problem can be solved in a standard and simple way, but a drawback of this arrangement is the space required, which in the case of mobile robot and vehicular application translates into a reduced vertical clearance and poorer aerodynamics. The horizontal arrangement, on the other hand, requires less space if mounted on a mobile robot, a vehicle’s roof or its lateral mirrors, but involves complex parabolic, elliptic or hyperbolic epipolar curves that make the solution of the stereo correspondence problem computationally more expensive. Despite the evident implementation advantages and drawback of each approach, no rigurous quantitative information exits in the literature about which of the two configurations provides in general the most accurate 3D reconstruction of the environment.
• Session: No. 1
• Presentation slides

"An Evolutionary Sonification Process Based on Robotic Trajectories"

• Submission No. 22
• Authors: Artemis Moroni and Jonatas Manzolli
• Affiliation: Centro de Tecnologia da Informacao, Brazil; Interdisciplinary Nucleus of Sound Studies, Brazil
• Keywords: Audio systems, Biological systems, Robot vision systems, Robots
• Abstract: A new approach which integrates an interactive evolutionary interface for sound production, real world devices, such as mobile robots, and an omnidirectional vision system, is presented. Starting with an evolutionary composition system named VOX POPULI and later JaVOX, a hybrid environment named AURAL has evolved. It uses curves drawn in a GUI and trajectories produced by mobile robots to modify the fitness function of a real time sonic composition. An arena was constructed to allow the interaction between robots and to control the variations of sonic parameters producing real time MIDI data. The navigation supervisor system, the sonification environment and graphic results of real time experiments are described.
• Session: No. 1
• Presentation slides

"Humanoid motion representation by sensory state transitions"

• Submission No. 5
• Authors: Fabio DallaLibera, Takashi Minato, Hiroshi Ishiguro, Ademar Ferreira, Enrico Pagello and Emanuele Menegatti
• Affiliation: Universidade de Sao Paulo, Brazil; Japan Science and Technology Agency, Japan; Osaka University, Japan; Universita degli Studi di Padova, Italy
• Keywords: humanoid, robot motion
• Abstract: Given the complexity and the high number of degrees of freedom humanoid robot motions often are generated off-line, stored in the robot memory and then reproduced. In the simplest cases motor commands are just replayed in open-loop (i.e. feed-forward), in more advanced implementations few parameter are changed on fly to adapt the motion to the external conditions. Indeed, several authors proposed a large variety of techniques to exploit the input from sensory feedback to modify a reference trajectory, in order to cope with environment changes and disturbances. In this paper, we propose a motion representation for humanoid robots that includes the sensory feedback information in the motion representation itself. This motion description allows stabilization against disturbances and environmental changes, but does not require any design or tuning of the relationships between sensory inputs and movement modification. We present experimental results on a simulated small humanoid robot equipped with motor encoders and touch sensors covering the whole body.
• Session: No. 2
• Presentation slides

"Autonomous Information Fusion for Robust Obstacle Localization on a Humanoid Robot"

• Submission No. 24
• Authors: Mohan Mohan Sridharan and Xiang Li
• Affiliation: Texas Tech University, USA
• Keywords: Information fusion, Robot vision, Obstacle localization, Humanoid robot soccer
• Abstract: Recent developments in sensor technology have resulted in the deployment of mobile robots equipped with multiple sensors, in specific real-world applications. A robot equipped with multiple sensors, however, obtains information about different regions of the scene, in different formats and with varying levels of uncertainty. In addition, the bits of information obtained from different sensors may contradict or complement each other. One open challenge to the widespread deployment of robots is the ability to fully utilize the information obtained from each sensor, in order to operate robustly in dynamic environments. This paper presents a probabilistic framework to address autonomous multisensor information fusion on a humanoid robot. The robot exploits the known structure of the environment to autonomously model the expected performance of the individual information processing schemes. The learned models are used to effectively merge the available information. As a result, the robot is able to robustly detect and localize mobile obstacles in its environment. The algorithm is fully implemented and tested on a humanoid robot platform (Aldebaran Naos) in the robot soccer scenario.
• Session: No. 2

"Design of a Mobile Robot for RoboCup Middle Size League"

• Submission No. 4
• Authors: Andre Conceicao, Paulo Moreira and Paulo Costa
• Affiliation: Universidade Federal da Bahia, Brazil
• Keywords: Mobile robotics, Embedded and Mobile Hardware, Modeling, Robot Soccer
• Abstract: This article presents a design of a four-wheeled omnidirectional mobile robot for RoboCup Middle Size League. The mobile robot was built for the 5dpo-2000 Robotic Soccer team from the Department of Electrical and Computer Engineering at the University of Porto, Portugal. The robot's architecture and communication structure are presented. A nonlinear modeling and a motion analysis based on the dynamics, kinematics, and DC motors are analyzed. This model could find the nonlinear saturation elements, such as limits of current of the motors, amplitude of the applied voltage in the motors, and frictions related to equation of motion. Simulation results are provided to demonstrate the performance of the proposed project.
• Session: No. 2
• Presentation slides

"An Integrated Multi-Agent Decision Making Framework for Robot Soccer"

• Submission No. 32
• Authors: Pablo Guerrero, Javier Ruiz-del-Solar, Miguel Romero and Luis Alberto Herrera
• Affiliation: Universidad de Chile, Chile
• Keywords: Integrated Decision Making, Robot Soccer, Markov Decision Process
• Abstract: Decision making is one of the most interesting issues to be solved in the RoboCup domain. This paper presents an integrated multi-agent decision making framework for robot soccer which addresses the problems of where to kick the ball to, and how to position in the field when the robot does not have the ball. The decision making framework is based on an MDP model of a soccer game, and it is integrated in the sense that the same criterions are used by the different robots for the selection of both the ball destination and the robot position. The proposed decision making framework is tested in simulated robot soccer matches, and decisions made in selected situations are shown.
• Session: No. 2
• Presentation slides (SOON)

"Visual Odometry and Mapping for Underwater Autonomous Vehicles"

• Submission No. 16
• Authors: Silvia Botelho, Gabriel Oliveira, Paulo Drews and Monica Figueiredo
• Affiliation: Universidade Federal do Rio Grande, Brazil
• Keywords: Computer Vision, Self-organizing maps, Robotics
• Abstract: The use of Autonomous Underwater Vehicles (AUVs) for underwater tasks is a promising robotic field. These robots can carry visual inspection cameras. Besides serving the activities of inspection and mapping, the captured images can also be used to aid navigation and localization of the robots. Visual odometry is the process of determining the position and orientation of a robot by analyzing the associated camera images. It has been used in a wide variety of non-standard locomotion robotic methods. In this context, this paper proposes an approach to visual odometry and mapping of underwater vehicles. Supposing the use of inspection cameras, this proposal is composed of two stages: i) the use of computer vision for visual odometry, extracting landmarks in underwater image sequences and ii) the development of topological maps for localization and navigation. The integration of such systems will allow visual odometry, localization and mapping of the environment. A set of tests with real robots was accomplished, regarding online and performance issues. The results reveals an accuracy and robust approach to several underwater conditions, as illumination and noise, leading to a promissory and original visual odometry and mapping technique.
• Session: No. 2
• Presentation slides

"Short-Term Memory Mechanisms in Neural Network Learning of Robot Navigation Tasks: A Case Study"

• Submission No. 13
• Authors: Ananda Freire, Guilherme Barreto, Marcus Veloso and Themoteo Varela
• Affiliation: Universidade Federal do Ceara, Brazil; Federal Institute of Education, Technology and Science, Brazil
• Keywords: Neural Networks, Robot navigation, Short-term memory, Spatiotemporal classification, Elman network
• Abstract: This paper reports results of an investigation on the degree of influence of short-term memory mechanisms on the performance of neural classifiers when applied to robot navigation tasks. In particular, we deal with the well-known strategy of navigating by ``wall-following''. For this purpose, four standard neural architectures (Logistic Perceptron, Multilayer Perceptron, Mixture of Experts and Elman network) are used to associate different spatiotemporal sensory input patterns with four predetermined action categories. All stages of the experiments - data acquisition, selection and training of the architectures in a simulator and their execution on a real mobile robot - are described. The obtained results suggest that the wall-following task, formulated as a pattern classification problem, is nonlinearly separable, a result that favors the MLP network if no memory of input patters are taken into account. If short-term memory mechanisms are used, then even a linear network is able to perform the same task successfully.
• Session: No. 3
• Presentation slides

"Segment-wise optimal trajectory execution control for four-wheeled omnidirectional mobile robots"

• Submission No. 27
• Authors: Luis F. Lupian and Josue R. Rabadan-Martin
• Affiliation: Universidad La Salle, Mexico
• Keywords: Omnidirectional mobile robot, LQR control, optimal trajectory control, OpenGL
• Abstract: We introduce a novel technique for path execution control on a four-wheeled omnidirectional mobile robot which provides a segment-wise near-optimal solution based on LQR. The path is specified by a sequence of target regions that the mobile robot will have to visit in order. Unlike other similar solutions that can be found in the literature, our proposed technique is flexible enough to adjust the optimality index for different strategies such as low power consumption, high speed, high accuracy, and so on. We also show how the dynamic model for this kind of robot can be linearized by a change of variables on the control vector. The application that motivated this work is the RoboCup SmallSize League. However, the ideas presented in this paper can be easily extended to other similar applications. The proposed solution is tested against a benchmark simulation experiment and visualized inside an OpenGL virtual environment.
• Session: No. 3
• Presentation slides

"Formation control of cooperative robots with limited sensing using a virtual robot as reference"

• Submission No. 21
• Authors: Eduardo Francisco Caicedo, Eval Bladimir Bacca, Jonathan Hernandez and Alejandro Pustowka
• Affiliation: Universidad del Valle, Colombia
• Keywords: Formation control, Cooperative robots, Virtual robot, Behavior-based control, Colective behavior
• Abstract: The task of maintaining formations is of great importance in high level applications being executed by multi-robot systems, such as exploration, surveillance and cleaning among others. This article presents a novel strategy for formation maintenance in a group of mobile, cooperative, homogenous robots, which work together with a software agent known as “virtual robot”. This virtual agent acts as reference for the spatial localization of each robot, and facilitates the navigation tasks of the group. This new approach combines the characteristics of other well known formation tecniques: leader- and unit-center referenced formations. In such way, we obtained advantages of both approaches. The proccess of validating this strategy was obtained from both simulated and real multi-robot systems.
• Session: No. 3
• Presentation slides (SOON)

"Project of a Hardware and Software Architecture for an Unmanned Aerial Vehicle"

• Submission No. 20
• Authors: Djalma Maranhao and Pablo Alsina
• Affiliation: Universidade Federal do Rio Grande de Norte, Brazil
• Keywords: Hardware/Software architecture, master-slave, UAV, USB, linux
• Abstract: This paper describes a hardware/software architecture for a UAV (Unmanned Aerial Vehicle). This architecture follows the master-slave model and uses the USB standard as the communication interface. Using the unique USB features, it is possible to build a robot using a standard Linux as operational system, and even though, attend to certain time deadlines.
• Session: No. 3
• Presentation slides

"Dynamic Gesture Recognition for Human Robot Interaction"

• Submission No. 23
• Authors: Jong Lee-Ferng, Javier Ruiz-del-Solar, Rodrigo Verschae and Mauricio Correa
• Affiliation: Universidad de Chile
• Keywords: dynamic hand gesture recognition, human robot interaction, RoboCup @Home
• Abstract: In this article a robust and real-time dynamic hand gesture recognition system meant to allow a natural interaction with a service robot, in dynamic environments, is proposed. The main novelty of the proposed approach is the use of temporal statistics about the hand’s positions and velocities as basic information to recognize the gestures. The use of these features allows carrying out the final recognition using a standard Bayes classifier, instead of the traditional Hidden Markov Models. The gesture segmentation and recognition is achieved simultaneously by finding gesture’s candidate subsequences that give high scores when matched to a gesture. The system uses boosted classifiers to detect hands, and the mean-shift algorithm for their tracking. The system performance is validated in a digit recognition system database and real-world video sequences.
• Session: No. 3
• Presentation slides

"Face Recognition using Thermal Infrared Images for Human-Robot Interaction Applications: A Comparative Study"

• Submission No. 35
• Authors: Gabriel Hermosilla, Javier Ruiz-del-Solar, Rodrigo Verschae and Mauricio Correa
• Affiliation: Universidad de Chile
• Keywords: Face Recognition, Thermal Infrared Images, Human-Robot Interaction
• Abstract: The aim of this work is to carry out a comparative study of face-recognition methods for Human-Robot Interaction (HRI) applications using long wave infrared images. The analyzed methods are selected by considering their suitability for HRI use, and their performance in former comparative studies. The methods are compared using visible and infrared (8-12 µm) images. The results of this comparative study are intended to be a guide for developers of face recognition systems for HRI.
• Session: No. 3
• Presentation slides