Skip to main content
main-content
Top

2013 | Book

Intelligent Autonomous Systems 12

Volume 1 Proceedings of the 12th International Conference IAS-12, held June 26-29, 2012, Jeju Island, Korea

Editors: Sukhan Lee, Hyungsuck Cho, Kwang-Joon Yoon, Jangmyung Lee

Publisher: Springer Berlin Heidelberg

Book Series : Advances in Intelligent Systems and Computing

share
SHARE
insite
SEARCH

About this book

Intelligent autonomous systems are emerged as a key enabler for the creation of a new paradigm of services to humankind, as seen by the recent advancement of autonomous cars licensed for driving in our streets, of unmanned aerial and underwater vehicles carrying out hazardous tasks on-site, and of space robots engaged in scientific as well as operational missions, to list only a few. This book aims at serving the researchers and practitioners in related fields with a timely dissemination of the recent progress on intelligent autonomous systems, based on a collection of papers presented at the 12th International Conference on Intelligent Autonomous Systems, held in Jeju, Korea, June 26-29, 2012. With the theme of “Intelligence and Autonomy for the Service to Humankind, the conference has covered such diverse areas as autonomous ground, aerial, and underwater vehicles, intelligent transportation systems, personal/domestic service robots, professional service robots for surgery/rehabilitation, rescue/security and space applications, and intelligent autonomous systems for manufacturing and healthcare. This volume 1 includes contributions devoted to Autonomous Ground Vehicles and Mobile Manipulators, as well as Unmanned Aerial and Underwater Vehicles and Bio-inspired Robotics.

Table of Contents

Frontmatter

Autonomous Ground Vehicles and Mobile Manipulators

Frontmatter
On-Line Obstacle Detection Using Data Range for Reactive Obstacle Avoidance

This paper deals with the reactive navigation in clustered environment. It proposes an online and adaptive elliptic trajectory to perform smooth and safe mobile robot navigation. These trajectories use limit-cycle principle already applied in the literature [3]. The main contribution proposed here is to perform this navigation in a completely reactive way while using only range sensor data. At this aim, each obstacle to avoid is surrounded by an ellipse and its parameters are obtained online while using the sequential range data and appropriate method to identify the enclosed ellipse. Different methods to obtain these ellipse parameters are presented and implemented. A specific criterion is taken into account to obtain always smooth change in these parameters. A large number of simulations permit to show the efficiency of our proposal for the navigation in cluttered environment.

José Miguel Vilca, Lounis Adouane, Youcef Mezouar
Obstacle Avoidance Based on Plane Estimation from 3D Edge Points by Mobile Robot Equipped with Omni-directional Camera

In this paper, we propose a method for a mobile robot to avoid obstacles in its environment using an omni-directional camera. The method makes an environment map consisting of 3D edge points obtained from omni-directional camera images and estimates the locations of planes by analysing these 3D edge points so that the robot can avoid walls as obstacles. The method has the advantage that it can generate a 3D map in environments constructed by textureless planes. Experimental results show the effectiveness of the proposed method.

Kazushi Watanabe, Ryosuke Kawanishi, Toru Kaneko, Atsushi Yamashita, Hajime Asama
A Generalized Neural Network Approach to Mobile Robot Navigation and Obstacle Avoidance

Navigation is one of the most important problems in developing and designing intelligent mobile robots. To locally navigate and autonomously plan a path to arrive to a desired destination, Artificial Neural Networks (ANNs) are employed to model complex relationships between inputs and outputs or to find patterns in data as they provide more suitable solutions than the traditional methods. However, current neural network navigation approaches are limited to one kind of robot platform and range sensor, and usually are not extendable to other types of robots with different range sensors without the need to change the network structures. In this paper, we propose a general method to interpret the data from various types of 2-dimensional range sensors and a neural network algorithm to perform the navigation task. Our approach can yield a global navigation algorithm which can be applied to various types of range sensors and robot platforms. Moreover, this method contributes positively to reducing the time required for training the networks.

S. Hamid Dezfoulian, Dan Wu, Imran Shafiq Ahmad
Drivable Road Recognition by Multilayered LiDAR and Vision

This paper presents the processing of the drivable road recognition by multilayered LiDAR and vision. Multilayered LiDAR information used for detecting a planetary region with boundaries and vision processing gives colored lane information for safe driving control. During navigating the road, EKF result on two different information are fused for robust and reliable navigation. This sensor fusing technique makes the autonomous navigation system to be robust and useful in real environment not only on regular road intersection but also unpaved ground way.

Suhyeon Gim, Ilyas Meo, Yongjin Park, Sukhan Lee
Field Trial Results of Autonomous Road Crossing Mobile Robot

We have developed a fully-integrated outdoor mobile robot that is capable of crossing roads autonomously in real world urban environments. To do this, the robot travels along pedestrian sidewalks autonomously, continually detects pedestrian push button boxes and navigates to it when one is detected. It then activates the push-button using an onboard-finger, moves to the crossing zone and crosses the road after detecting the zebra stripes and pedestrian lights. In this paper, we report the results of preliminary field trial experiments where the robot was deployed in a real world environment and its performance was evaluated.

Aneesh Neeschal Chand, Shin’ichi Yuta
Timed Trajectory Generation Combined with an Extended Kalman Filter for a Vision-Based Autonomous Mobile Robot

Planning collision-free trajectories requires the combination of generation and modulation techniques. This is especially important if temporal stabilization of the generated trajectories is considered. Temporal stabilization means to conform to the planned movement time, in spite of environmental conditions or perturbations. This timing problem has not been addressed in most current robotic systems, and it is critical in several robotic tasks such as sequentially structured actions or human-robot interaction. This work focuses on generating trajectories for a mobile robot, whose goal is to reach a target within a constant time, independently of the world complexity. Trajectories are generated by nonlinear dynamical systems. Herein, we extend our previous work by including an Extended Kalman Filter (EKF) to estimate the target location relative to the robot. A simulated hospital environment and a Pioneer 3-AT robot are used to demonstrate the robustness and reliability of the proposed approach in cluttered, dynamic and uncontrolled scenarios. Multiple experiments confirm that the inclusion of the EKF preserves the timing properties of the overall architecture.

Jorge Bruno Silva, Cristina P. Santos, João Sequeira
Autonomous Navigation of a Personal Transporter within Moving Human Groups Using Reactive Control

If a vehicle has to drive autonomously in an area with mostly pedestrian traffic and thus needs to share the free space with pedestrians it has to safely navigate in this environment. A pedestrian area lacks fixed lanes and definite traffic rules which leads to more complex paths. Classical path planning in a partially known environment with multiple moving pedestrians is a hard planning problem. Just the task of accurately predicting the individual path of multiple pedestrians in a group accounting for all interactions between the different persons over more than a few seconds is as good as impossible. Using this faulty prediction as a foundation for a path planning algorithm will produce a flawed plan. However such an environment is easy for humans or animals to navigate, avoidance of other persons and moving in a group of pedestrians happens almost unconsciously. In this paper a simple way of autonomously navigating such an environment by using a pure reactive control based on a few simple rules is explored. This method doesn’t need a completely known environment, doesn’t plan ahead, is simple to implement and can react quick to changes in the environment.

Florian Steinhardt, Marcus Strand, J. Marius Zöllner
Simultaneous Control of Translational and Rotational Motion for Autonomous Omnidirectional Mobile Robot Considering Shape of the Robot and Movable Area by Heights

This paper presents a real time collision avoidance method for an autonomous omnidirectional mobile robot considering shape of the robot and movable area by heights based on simultaneous control of translational and rotational motion. Service robots which have been developed in recent years have arms to work and execute tasks. In these robots, the size of width is sometimes not equal to that of depth by heights. In order to avoid obstacles considering safety and mobility for the robots, it is necessary to evaluate shape of the robot and movable area by heights. To evaluate them, the robot model is defined by heights. Evaluating of the robot model and the movable area for each height, if the robot is unable to move keeping a safe distance from the obstacles, the robot determines the suitable orientation angle considering the minimum length from the center of the robot model to that outer shape. In this paper, the novel control method based on the fuzzy potential method is presented. To verify the effectiveness of the proposed method, several numerical simulations are carried out.

Ayanori Yorozu, Takafumi Suzuki, Tetsuya Matsumura, Masaki Takahashi
Development of an Autonomous Vehicle for High-Speed Navigation and Obstacle Avoidance

This paper introduces the autonomous vehicle named Pharos that participated in 2010 Autonomous Vehicle Competition organized by Hyundai-Kia Automotive Group. Pharos was developed for high-speed on/off-road unmanned driving avoiding diverse patterns of obstacles. For the high speed travelling up to 60 Km/h, long range terrain perception, real-time path planning and high speed vehicle motion control algorithms are developed. This paper describes the major hardware and software components of our vehicle.

Jee-Hwan Ryu, Dmitriy Ogay, Sergey Bulavintsev, Hyuk Kim, Jang-Sik Park
Advanced Perception for Robots in a Closed World Environment

A basic requirement for all mobile robots is a precise, accurate and fast perception of the environment to allow intelligent behavior through interaction with its surrounding. This paper introduces a new method for a reliable, fast and efficient purely omni-directional vision based mobile robot navigation for a closed world environment. The proposed method enables a robot that is driving at a high-speed to detect and classify different objects of different colors by only calibrating one dominant environment color. The correctness and the effectiveness of the introduced approach was evaluated successfully in the highly dynamical RoboCup Middle Size Soccer scenario.

Mahmoud El Shaikh, Andreas Koch, Bernd Eckstein, Kai Häussermann, Oliver Zweigle, Paul Levi
An Android Remote Call Vehicle Service for OSGi-Based Unmanned Vehicle Using by a Mobile Device

Almost every country is interested in electronic vehicle uses of green energy. Smart device appeared by developing mobile devices. Thus, the desire to overcoming our rapid changeable world, we believe that the study of balancing way between Smart Technology and Vehicle must be necessary. To achieve this, we should support open interface by adopting OSGi framework in a vehicle. And for mobile device user or caller, we should design and implement android application of mobile devices. This helps cooperation to the OSGi framework and Android. Especially, we propose the Android-based Remote Call Vehicle Service (ARC-VS) for mobile user or caller.

Choon-Sung Nam, Sukhan Lee, Dong-Ryeol Shin
Visual Memory Update for Life-Long Mobile Robot Navigation

A central clue for implementation of visual memory based navigation strategies relies on efficient point matching between the current image and the key images of the memory. However, the visual memory may become out of date after some times because the appearance of real-world environments keeps changing. It is thus necessary to remove obsolete information and to add new data to the visual memory over time. In this paper, we propose a method based on short-term and long term memory concepts to update the visual memory of mobile robots during navigation. The results of our experiments show that using this method improves the robustness of the localization and path-following steps.

Jonathan Courbon, Hemanth Korrapati, Youcef Mezouar
Topological Mapping with Image Sequence Partitioning

Topological maps are vital for fast and accurate localization in large environments. Sparse topological maps can be constructed by partitioning a sequence of images acquired by a robot, according to their appearance. All images in a partition have similar appearance and are represented by a node in a topological map. In this paper, we present a topological mapping framework which makes use of image sequence partitioning (ISP) to produce sparse maps. The framework facilitates coarse loop closure at node level and a finer loop closure at image level. Hierarchical inverted files (HIF) are proposed which are naturally adaptable to our sparse topological mapping framework and enable efficient loop closure. Computational gain attained in loop closure with HIF over sparse topological maps is demonstrated. Experiments are performed on outdoor environments using an omni-directional camera.

Hemanth Korrapati, Jonathan Courbon, Youcef Mezouar
Kinect Enabled Monte Carlo Localisation for a Robotic Wheelchair

Proximity sensors and 2D vision methods have shown to work robustly in particle filter-based Monte Carlo Localisation (MCL). It would be interesting however to examine whether modern 3D vision sensors would be equally efficient for localising a robotic wheelchair with MCL. In this work, we introduce a visual Region Locator Descriptor, acquired from a 3D map using the Kinect sensor to conduct localisation. The descriptor segments the Kinect’s depth map into a grid of 36 regions, where the depth of each column-cell is being used as a distance range for the measurement model of a particle filter. The experimental work concentrated on a comparison of three different localization cases. (a) an odometry model without MCL, (b) with MCL and sonar sensors only, (c) with MCL and the Kinect sensor only. The comparative study demonstrated the efficiency of a modern 3D depth sensor, such as the Kinect, which can be used reliably for wheelchair localisation.

Theodoros Theodoridis, Huosheng Hu, Klaus McDonald-Maier, Dongbing Gu
Succinct Landmark Database

Recently developed robotic mapping techniques enable the acquisition of large scale landmark databases. This paper explores an approach for succinct landmark database, which memorizes a large collection of point landmarks while allowing to random access the location of i-th landmark. Our approach combines and extends three independent compression techniques: space coding, succinct data structure, and exemplar-based scene compression. Experiments using real datasets evaluate effectiveness of the presented techniques in terms of compactness, access speed, and accuracy of landmark database.

Kanji Tanaka
Thermal 3D Mapping of Building Façades

Never before in history were humans as dependant on energy as we are today. But the natural ressources are limited and a waste of energy has drastic influences on the environment. In their Action Plan for Energy Efficiency [6] the European Commission estimates that the largest and cost-effictive energy savings potential lies in residential (≈ 27%) and commercial (≈ 30%) buildings. To eliminate heat and air conditioning losses in buildings and factories heat and air leaks need to be localized and identified. Imagine the availability of a complete 3D model of every building that architects can use to analyze the heat insulation of buildings and to identify necessary modifications. In these 3D models temperature peaks are not only detectable but also their extent is visible. A robot equiped with a 3D laser scanner, a thermal camera, and a color camera constitutes the basis for our approach. The data from all three sensors and from different locations are joined into one high-precise 3D model that shows the heat distribution. This paper describes the setup of the hardware and the methods applied to create the 3D model, including the automatic co-calibration of the sensors. Challenges unique to the task of thermal mapping of outdoor environments are discussed.

Dorit Borrmann, Jan Elseberg, Andreas Nüchter
Moving Object Detection Using Monocular Vision

This paper presents an algorithm for moving object detection (MOD) in robot visual simultaneous localization and mapping (SLAM). The algorithm is designed based on the defining epipolar constraint for the corresponding feature points on image plane. An essential matrix obtained using the state estimator is utilized to represent the epipolar constraint. Meanwhile, the method of speeded-up robust feature (SURF) is employed in the algorithm to provide a robust detection for image features as well as a better description of landmarks and of moving objects in visual SLAM system. Experiment is carried out on a hand-held monocular camera to validate the performances of the proposed algorithm. The results show that the integration of MOD and SURF is efficient for robot navigating in dynamic environments.

Yin-Tien Wang, Kuo-Wei Chen, Ming-Jang Chiou
Linear Kalman Filter for Dead Time Affected Measurement Signals Implemented in a Small Scale Automated Guided Vehicle

This paper describes an efficient way to calculate state estimations with a linear Kalman filter, if the measurement signal is subject to a fix dead time. In addition the presented solution can handle measurement signals with arbitrarily high sample times. After the derivation of the algorithm an application with high measurement signal dead time is explained. Finally a simulation depicts the result of the algorithm being implemented on the described application.

F. Lütteke, J. Franke
Balance Control of a Variable Centroid Inverted Pendulum Robot

A basic problem has been found in two wheels inverted pendulum based mobile manipulator. It is difficult to keep balance as the manipulator operates and it is necessary to find a method to resolve it. This research presents a simple and feasible method for this kind of problem. It involves a common PD control and a method to calculate the compensatory angle for a variable centroid inverted pendulum robot. A 4 DOF mobile manipulator based on inverted pendulum has been developed to analyze the problem, and a simulation is also used to certify the method in this research. The experiment and its results show the performance of the method. It is possible to make the robot keep balance by using the compensatory angle even though the position of centroid is time variant.

Gang Wang, Sang Yong Lee, Seok Won Kang, Jang-Myung Lee
Stabilization for Truck-Trailer Mobile Robot System via Discrete LPV T-S Fuzzy Models

The stability analysis and synthesis problem of discrete nonlinear Truck-Trailer Mobile Robot (TTMR) system is studied in this paper. The Linear Parameter Varying (LPV) model and Takagi-Sugeno (T-S) fuzzy model are employed to construct the discrete nonlinear TTMR system. There are few researchers discussing the stability analysis and synthesis for the discrete LPV T-S fuzzy models. Therefore, the problem discussed in this paper is worthy to receive the attention of control engineers.

Wen-Jer Chang, Po-Hsun Chen
Discovering New Motor Primitives in Transition Graphs

In this paper we propose a methodology for discovering new movement primitives in a database of example trajectories. The initial trajectory data, which is usually acquired from human demonstrations or by kinesthetic guiding, is clustered and organized into a binary tree, from which transition graphs at different levels of granularity are constructed. We show that new movements can be discovered by searching the transition graph, exploiting the interdependencies between the movements encoded by the graph. By connecting the results of the graph search with optimized interpolation and statistical generalization techniques, we can construct a complete representation for new movement primitives, which were not explicitly present in the original database of example trajectories.

Miha Deniša, Aleš Ude
Design of a Space Robot System to Simulate Climbing of Astronaut Based on Binocular Vision System

The exploration and development of space science nowadays is explosive, the tasks which astronauts need to achieve in space are also more dangerous and complex. Therefore, space robot is needed to implement the EVA of astronauts to do these space missions (EVA—Extravehicular activity). Firstly, we designed a kind of space robot with vision system and also we discussed a vision information processing method based on a binocular stereo vision system which is used on our space robot. Secondly, we designed a passive experimental platform which can simulate the weightless environment in space for space robot to simulate climbing of astronaut, and a method is adopted to improve the stability of the passive mechanism by analyzing force and vibration condition. Lastly, based on this robot system, space robot could simulate the extravehicular movement of astronaut based on its binocular vision system, and this validated presented methods in this paper.

Que Dong, Yu He, Hongjie Li, Bo Wei, Xiaopeng Chen, Hui Li, Zhihong Jiang, Qiang Huang
Fast 6D Odometry Based on Visual Features and Depth

The availability of affordable RGB-D cameras which provide color and depth data at high data rates, such as Microsoft MS Kinect, poses a challenge to the limited resources of the computers onboard autonomous robots. Estimating the sensor trajectory, for example, is a key ingredient for robot localization and SLAM (Simultaneous Localization And Mapping), but current computers can hardly handle the stream of measurements. In this paper, we propose an efficient and reliable method to estimate the 6D movement of an RGB-D camera (3 linear translations and 3 rotation angles) of a moving RGB-D camera. Our approach is based on visual features that are mapped to the three Cartesian coordinates (3D) using measured depth. The features of consecutive frames are associated in 3D and the sensor pose increments are obtained by solving the resulting linear least square minimization system. The main contribution of our approach is the definition of a filter setup that produces the most reliable features that allows for keeping track of the sensor pose with a limited number of feature points. We systematically evaluate our approach using ground truth from an external measurement systems.

Salvador Domínguez Quijada, Eduardo Zalama, Jaime Gómez García-Bermejo, Rainer Worst, Sven Behnke
Moving Region Segmentation Using Sparse Motion Cue from a Moving Camera

This paper presents a method for pixel-wise segmentation of moving regions using sparse motion cues on an image from a freely moving camera. The main idea is to utilize residual motion, i.e., motion relative to a background, on sparse grid points. Our algorithm consists of three parts: global motion estimation, characterization of points based on sparse motion cue, and pixel-wise labeling of moving regions. Experimental results on real image sequences are presented, showing the effectiveness of the proposed method.

Jungwon Kang, Sijong Kim, Taek Jun Oh, Myung Jin Chung
Fast and Robust Multi-people Tracking from RGB-D Data for a Mobile Robot

This paper proposes a fast and robust multi-people tracking algorithm for mobile platforms equipped with a RGB-D sensor. Our approach features an efficient point cloud depth-based clustering, an HOG-like classification to robustly initialize a person tracking and a person classifier with online learning to manage the person ID matching even after a full occlusion. For people detection, we make the assumption that people move on a ground plane. Tests are presented on a challenging real-world indoor environment and results have been evaluated with the CLEAR MOT metrics. Our algorithm proved to correctly track 96% of people with very limited ID switches and few false positives, with an average frame rate of 25 fps. Moreover, its applicability to robot-people following tasks have been tested and discussed.

Filippo Basso, Matteo Munaro, Stefano Michieletto, Enrico Pagello, Emanuele Menegatti
Multi-agent Based Optic Flow

In this article, the authors present a novel algorithm for computing optic flow using a multi-agent based feature point tracking method. In this multi-agent based optic flow method, feature points which are invariant to scale, orientation and illumination changes are extracted and tracked in parallel using independent agents. Each agent is run by a separate light-weight thread which can be implemented using parallel processes on a multicore processor. The agents use a Kalman filter to predict the frame to frame position of the feature points in the image, producing position and velocity data for each feature point, which can then be used to perform optic flow, while simultaneously producing feature descriptors that can be used for object recognition and stereopsis. We show that in a parallel implementation, this algorithm provides significant performance advantages over other feature point tracking object recognition methods. It therefore may provide a plausible basis for a unified computer vision architecture including optic flow, object recognition, and stereopsis.

Kiwon Sohn, Paul Oh, M. Anthony Lewis
3D Face Recognition Based on Curvature Feature Matching, with Expression Variation

In this paper, we try to improve face recognition system by taking better advantage of the inherent 3D nature of the face. Face recognition can be greatly improved because the abundant 3D face features can be obtained from different angles. During the simulation, we try to extract the curvatures of the eyes, nose and mouth, which can be used as features for face recognition. The Gaussian curvature is an important component of our work. The distribution of this curvature is used to construct the feature vectors. In order to raise the recognition rate, the projection method is used to intensify the edge information. The mesh modification method is also applied to the 3D mesh models. Finally, the distance between the 3D normalized curvatures of the features is compared between the query and database images for recognition. Even when the facial expression of the query image has changed, we can still achieve a 92% recognition rate with our 3D face recognition algorithms.

Shu-Wei Lin, Shu-Shen Hao, Jui-Lun Chang, Sheng-Yi Li
Advances in 3D Camera: Time-of-Flight vs. Active Triangulation

Over the last decade, numerous 3D camera techniques have been proposed and advanced dramatically. One main approach is time-of-flight (TOF) and the other is active triangulation. Each has its own strengths and weaknesses. In this paper, we overview the principle of each method and compare the advantages and disadvantages in detail, and introduce several commercially available 3D cameras and their characteristics.

Daesik Kim, Sukhan Lee
A Novel 3D Registration Algorithm Using Parallel-Light Association

This paper presents a novel method for free-form registration of multiple point clouds. The method adopts a parallel-light data association design inspired from torchlight structure which improves the correctness of point correspondence. When two sets of point clouds are placed together, assume a set of parallel light beams are passing through them. Each light beam will pass the point clouds twice, one on each data set. The Euclidean distance on each light beam between the two sets are taken as measurement of the separation. The fitness is the reciprocal of the mean distance of all light beams. When the two sets are optimally aligned, the fitness is maximized. Hence, the registration problem is reduced to a six degree of freedom search. Preprocessing and acceleration modules such as Genetic Algorithm (GA) are introduced to reduce the exploration space and execution time. Unlike the Iterative Closest Point (ICP) algorithm, the proposed algorithm does not require pre-alignment information. Secondly, ICP does not perform well when the overlapped area between two sets are not sufficiently large. And the proposed algorithm does not suffer from this partial overlapping problem. Based on various experiments with real data, the proposed method has superior performance compared to ICP.

Han Wang, Ying Ying
Real-Time Model Based Visual Servoing Tasks on a Humanoid Robot

Several model based techniques have been used to apply various domestic service tasks on humanoid robots ( through teleoperation, learning, ...). But for many reasons, it is more suitable to study the interaction between the robot and its environment using the Sensor Based Control in these cases. In this paper we present a work of integration of real-time visual servoing techniques in performing self localization and different manipulation tasks on a humanoid robot in closed loop.

Real-time model based tracking techniques are used to apply 3D visual servoing tasks on the Nao humanoid robot. Elementary tasks used by the robot to perform a concrete scenario are detailed with their corresponding control laws. Experimental results are presented for the following tasks: self-localization of the robot while walking, head servoing for the visibility task, detection, tracking and manipulation of environment’s objects.

Amine Abou Moughlbay, Enric Cervera, Philippe Martinet
Visual Gyroscope for Omnidirectional Cameras

At present, algorithms for attitude estimation with omnidirectional cameras are predominantly environment-dependent. This constitutes a significant limitation to the applicability of such techniques. This study introduces an approach aimed at general mobile camera attitude estimation. The approach extracts features to directly estimate three-dimensional movements of a humanoid robot from its head-mounted camera. By doing so, it is not subject to the constraints of Structure from Motion with epipolar geometry, which are currently unattainable in real-time. The central idea is: movements between consecutive frames can be reliably estimated from the identity on the unit sphere between external parallel lines and projected great circles. After calibration, parallel lines match optical flow tracks. The point of infinity corresponds to the expansion focus of the movement. Simulations and experiments validate the ability to distinguish between translation, pure rotation, and roto-translation.

Nicola Carlon, Emanuele Menegatti
Panoramic 3D Reconstruction with Three Catadioptric Cameras

In this paper, we present a new system setup combining three catadioptric cameras. Our application is in autonomous vehicles. This camera setup allows for panoramic stereo vision of the environment and therefore proves to be useful for ego motion estimation and localization by 3D feature points all around the vehicle. The three catadioptric cameras are arranged in a triangle on top of the vehicle and are horizontally aligned. In the paper we discuss two methods for 3D scene point reconstruction with the system. We perform experiments for the 3D reconstruction in a simulated environment and evaluate the accuracy by means of a Monte-Carlo-Simulation. The proposed system improves the mean accuracy of the 3D reconstruction significantly compared to a system with two cameras.

Miriam Schönbein, Holger Rapp, Martin Lauer
Omnidirectional Vision for Indoor Spatial Layout Recovery

In this work, we study the problem of recovering the spatial layout of a scene from a collection of lines extracted from a single indoor image. Equivalent methods for conventional cameras have been proposed in the literature, but not much work has been done about this topic using omnidirectional vision, particulary powerful to obtain the spatial layout due to its wide field of view. As the geometry of omnidirectional and conventional images is different, most of the proposed methods for standard cameras do not work and new algorithms with specific considerations are required. We first propose a new method for vanishing points (VPs) estimation and line classification for omnidirectional images. Our main contribution is a new approach for spatial layout recovery based on these extracted lines and vanishing points, combined with a set of geometrical constraints, which allow us to detect floor-wall boundaries regardless of the number of walls. In our proposal, we first make a 4 walls room hypothesis and subsequently we expand this room in order to find the best fitting. We demonstrate how we can find the floor-wall boundary of the interior of a building, even when this boundary is partially occluded by objects and show several examples of these interpretations.

J. Omedes, G. López-Nicolás, J. J. Guerrero
EPI Analysis of Fish-Eye Images

This paper proposes a method to get depth information from image sequences obtained from a moving fish-eye camera. The motion is assumed to be along the optical axis of the fish-eye camera. In this case, a point on a fish-eye image moves to radial direction. This is one of the epipolar constraints that are very effective to find corresponding points in an image sequences. Epipolar-plane image (EPI) is an image having epipolar constraint in a volume of accumulating an image sequence. The gradient of a curve on EPI has depth information of the corresponding measuring point. Measurement accuracy of the proposed method is examined over a wide-angle range.

Kenji Terabayashi, Toru Morita, Hiroya Okamoto, Takaaki Oiwa, Kazunori Umeda
An Adaptive Fuzzy Based System for Time Critical Real World Applications

From a control theory point of view, robotics and artificial intelligence offer exceptionally complex problems. Often the examined control systems have a high number of inputs and outputs, show a black-box behaviour and can not be systematically analyzed due to a vague output and long reaction times. Adaptive behaviour control and enhancement during runtime of robots moving with high speed is such a problem, with the added requirement of real-time capability. In this paper, an adaptive pre-calculated fuzzy system is proposed as a possible solution for this task. The basic structure, construction process and adaption mechanism are described, furthermore the runtime for various dimensions is benchmarked as efficiency aspects are a major contribution of the approach.

Christoph Kattmann, Oliver Zweigle, Kai Häussermann, Paul Levi
Ontology Representation and Instantiation for Semantic Map Building by a Mobile Robot

To offer sustainable robotic services, service robots must accumulate knowledge by using recognition results and choose a action for services intelligently. Robust knowledge instantiation and update by using imperfect sensing data such as misidentification of perception is a main issue to implement semantic robot intelligence. In this paper, robust knowledge acquisition method is proposed to enable robots to detect falsity of object recognition for robust knowledge instantiation, where spatial reasoning, temporal reasoning, movable properties and data confidences are considered.

Gi Hyun Lim, Chuho Yi, Il Hong Suh, Dong Wook Ko, Seung Woo Hong
Recognizing Hardware Faults on Mobile Robots Using Situation Analysis Techniques

Current research in the field of robotics deals with planning and executing of increasingly complex tasks as well as applications in which autonomous mobile robot systems interact with highly dynamic environments. In order to operate successfully in a highly dynamic physical world a mobile robot needs awareness to adapt its behavior according to the environment. This awareness also has to cope with uncertainty due to imperfect hardware sensors and actuators or network delays of the robot itself. Thus another important ability of a mobile robot is to ”understand” its internal state to detect any kind of hardware faults in the system. In this work a highly customizable framework for situation awareness is presented. For the evaluation of this framework we take advantage of its generality and assume fault diagnosis as a more specific subclass of the general situation awareness process to extend the situation detection framework for fault detection and diagnosis in a multi robot system.

Oliver Zweigle, Benjamin Keil, Markus Wittlinger, Kai Häussermann, Paul Levi
A Decision System for Aircraft Faults Diagnosis Based on Classification Trees and PCA

Aircrafts are complex systems that require permanent and precise monitoring and troubleshooting. The automation of these tasks is thus of a high importance. This paper presents an intelligent decision system for faults diagnosis of aircrafts. The system relies on decision trees, being easier to interpret, quicker to learn than other data-driven methods, and able to work even with missing pieces of information. The used C4.5 algorithm automatically “learns” the best decision tree by performing a search through the set of possible trees according to the available training data. And Principal Component Analysis (PCA) is used to decrease the input data’s dimension. Compared to other methods, the proposed one is more advantageous and some presented evaluations demonstrate its abilities. High correct faults detection rates and low missed detection and false alarm rates are obtained. Such a decision system is highly useful for engineering consulting services, accumulating the knowledge for the operational rules of diagnosis, and the design of new aircrafts.

ZeFeng Wang, Jean-Luc Zarader, Sylvain Argentieri, Karim Youssef
A Sociology of Intelligent, Autonomous Cothinkers and Coagents

Scientific and technological progress has brought robots where machine-based cognition and cooperation abilities start to emerge; not only between robots and humans but also among multiple robots themselves. In order to technically improve performances in latter case, as well as, by analogy, to better understand how humans can interact with one another and grow communities, time as come to further, scientifically and technically develop sociology-related knowledge and ontologies. Critical theoretical bases for cognition have been built and demonstrated, both in the human and machine-based cases, providing valuable contributions in this regard. Now sociable competences are considered, allowing for incrementally binding individuals and small groups into holistic units of increasing scope. Ultimately, what is also considered here is a kind of common, meta-human, secular framework where robots and humans can best co-think and co-act. Concepts have now been complemented and validated by real size implementation and experimentation in the domain of homes, as well as industrial and public environments. This should motivate the reader to get familiar with the proposed formal, quantitative MCS framework, thereby getting better insight in judgment and better ability to quantify requirements.

Jean-Daniel Dessimoz, Jean-François Gauthey, Hayato Omori
An Algorithm Model for Gross Cognitive Reappraisal Strategy

In this paper, we use the mathematical model of Finite State Machine to describe the conversion process of individual’s emotional state based on Gross’s process model of emotion regulation strategies. In the model, we abstract Gross cognitive reappraisal strategy into a quantitative parameter and propose a kind of preliminary algorithm description for the influence cognitive reappraisal strategy has on the emotional conversion process. At last, we make simulation experiment for the algorithm model, and the experimental results show that the proposed algorithm proposed can effectively describe the relationship between reappraisal strategy and emotion-generative process.

Xiaolan Peng, Lun Xie, Xin Liu, Zhiliang Wang
Actor Studio: Development of User Friendly Action Editing System for Cultural Performance Robots

Social robots are developed for social interacting between humans and robots by various ways. One of the ways is sharing cultural performances such as singing, dancing, and acting. Various technologies are required for developing cultural robots, and one of the most important things is generating contents. We develop a user friendly action editing system with graphical user interface. It is comfort method to generate long contents of performance according to musical notes or scenarios. Using this method, we generate action data for some performances with real robot system in the theater.

Ho Seok Ahn, Dong-Wook Lee, Dongwoon Choi, Duk-Yeon Lee, Manhong Hur, Hogil Lee
RoboEarth Action Recipe Execution

The ability of reusing existing task execution plans is an important step towards autonomous behavior. Today, the reuse of sophisticated services allowing robots to act autonomous is usually limited to identical robot platforms and to very similar application scenarios. The approach presented in this paper proposes a way to mitigate this limitation by storing and reusing task plans on a global accessible database. We describe the task execution engine for the RoboEarth project to demonstrate its ability to execute tasks in a flexible and reliable way.

Daniel Di Marco, Moritz Tenorth, Kai Häussermann, Oliver Zweigle, Paul Levi
Exchanging Action-Related Information among Autonomous Robots

In this paper, we describe representations and inference techniques that are used in the RoboEarth system for the web-based exchange of information between robots. We present novel representations for environment maps that combine expressive semantic environment models with techniques for selecting suitable maps from the web-based RoboEarth knowledge base. We further propose techniques for improving class-level object models with additional information as needed for distributed learning of object properties. In an integrated experiment, we show that the system enables robots to perform mobile manipulation tasks including the retrieval of suitable environment maps and the estimation and exchange of object property information.

Moritz Tenorth, Michael Beetz
Ubiquitous Network Robot Platform for Realizing Integrated Robotic Applications

This paper introduces a common infrastructure for robotic applications to support our daily life: the

Ubiquitous Network Robot Platform

(UNR-PF). UNR-PF bridges the gaps among service application providers, customers and robotic devices such as robots, sensors and smartphones. For application programmers, UNR-PF provides a common API so that programmers can realize applications in a way independent to the actual variation in devices. UNR-PF also provides applications to span across multiple areas so that customers can receive seamless support in different scenes based on their abilities. Here, we first describe the motivation as well as requirements for UNR-PF based on past field studies. Based on these requirements, several real-world case studies has been performed which showed the effectiveness of the proposed concept of UNR-PF for continuously supporting our daily activities across multiple areas. We describe recent activities toward standardizing the key elements of UNR-PF and discuss future works.

Shuichi Nishio, Koji Kamei, Norihiro Hagita

Unmanned Aerial and Underwater Vehicles and Bio-inspired Robotics

Frontmatter
Rapid Prototyping Framework for Visual Control of Autonomous Micro Aerial Vehicles

Rapid prototyping environments can speed up the research of visual control algorithms. We have designed and implemented a software framework for fast prototyping of visual control algorithms for Micro Aerial Vehicles (MAV). We have applied a combination of a proxy-based network communication architecture and a custom Application Programming Interface. This allows multiple experimental configurations, like drone swarms or distributed processing of a drone’s video stream. Currently, the framework supports a low-cost MAV: the Parrot AR.Drone. Real tests have been performed on this platform and the results show comparatively low figures of the extra communication delay introduced by the framework, while adding new functionalities and flexibility to the selected drone. This implementation is open-source and can be downloaded from

www.vision4uav.com/?q=VC4MAV-FW

Ignacio Mellado-Bataller, Pascual Campoy, Miguel A. Olivares-Mendez, Luis Mejias
UKF Applied for Position Estimation of Underwater-Beacon Precision

A location estimation algorithm based on underwater beacon was proposed in this paper. Location estimation of beacon is prerequisite to the underwater localization and communication. Unscented Kalman Filter is verified through MATLAB simulations and experiments in real-time environments. And performance of Unscented Kalman Filter was evaluated through underwater-BEACON. In This experiments, MCU of Underwater-BEACON was used DSP(TMS320F28335). INS(Gyroscope Sensor, Accelerometer Sensor, Magnetic Compass) and GPS was Used for the location estimation of underwater BEACON. UKF combined GPS signals and INS signals to design Underwater vehicle’s nonlinearity and non-normal distribution. Underwater-BEACON applies UKF. It can see that fewer errors were occurring.

Ba-Da Yoon, Ha-Nul Yoon, Sung-He Choi, Jang-Myung Lee
Smart Filter Design for the Localization of Robotic Fish Using MEMS Accelerometer

This paper presents the design of smart filter for the 2D localization of robotic fish using low-cost MEMS (Micro-Electro Mechanical System) accelerometer. The main purpose of the paper is to minimize the drift error that is inevitable in the double integration process in accelerometer-only navigation system. The proposed approach relies on two parts: 1) an effective calibration method to remove the major part of the deterministic sensor errors and, 2) a novel smart filtering scheme based on fuzzy-logic in order to accurately estimate a 2D position with an accelerometer triad. In addition, we compare the results of the fuzzy logic based on 2D position estimation system with simulation result from a conventional Kalman Filter.

Tae Suk Yoo, Sang Cheol Lee, Sung Kyung Hong, Young Sun Ryuh
View Planning of a Multi-rotor Unmanned Air Vehicle for Tree Modeling Using Silhouette-Based Shape Estimation

The use of a multi-rotor unmanned air vehicle (UAV) in image acquisition tasks is promising for three-dimensional (3D) object modeling. Such an autonomous data acquisition system can be useful to handle the geometric complexity of objects such as trees and the inherent difficulties of image capture. In this paper, we address the problem of view planning for a camera-equipped multi-rotor UAV to acquire an adequate set of images that leads to more detailed and complete knowledge of the 3D tree model. The proposed algorithm based on shape-from-silhouette methods incorporates both expected new visual information and vehicle movement. Occupancy estimation for volumetric object model serves as a baseline measure of new information. The outlined approach determines next best views across the viewpoint space bounded by the sensor coverage and the capability of the UAV with minimal a priori knowledge of the object. Simulation studies conducted with virtual reality environments show the effectiveness of the algorithm.

Dae-Yeon Won, Ali Haydar Göktoğan, Salah Sukkarieh, Min-Jea Tahk
Planar Evasive Aircrafts Maneuvers Using Reinforcement Learning

In this paper, the reinforcement learning technique is proposed to implement evasive strategies for aircrafts during engagement. A simplified point-mass model is used to describe the aircraft and the missile equations of motion. The missile follows the pure proportional navigation guidance (PPNG) law to attack the aircraft. Q-learning algorithm which is a form of reinforcement learning is suggested to learn the evasive maneuvers. The performance of the proposed approach is analyzed with numerical simulations. It is shown that the aircraft evades from a missile properly by reinforcement learning with bang-bang type action profiles.

Dongjin Lee, Hyochoong Bang
Design and Implementation of Sensor Modules Enabling Round-the-Clock Underwater Operations

The legacy technologies of water quality measurement show the weakness that it manually or semi-automatically measures various water quality information at just only small parts of a large area. In this paper, we propose new water quality sensor apparatus which can consistently provide high resolution water quality data as well as attachable and PnP type deployable to underwater vehicles. The water quality sensor module, as a result of going through two water system tests in Han River and Gabchun and benchmarking with the commercial water quality meter, shows not only outstanding performance but also receives an official experimental achievement by the supervisor of the Korea Testing Laboratory.

Joongki Park, Juchan Sohn, Sunghoon Kim
A Standard Error Detection Mechanism for Underwater Acoustic Sensor Networks

Underwater Acoustic Sensor Networks (UWASNs) have some constant factors like energy limitations and node mobility. Energy is a very crucial issue as any kind of recharging is not possible in underwater environment. Energy consumption in each section is important. Saving a small amount of energy helps the life span of the sensor nodes as well as the lifetime of the networks. Especially for static UWASNs, long-term non-time critical application energy efficiency is of major concern. Beside energy efficient protocol design, error control is important in communication as the issue of data integrity is becoming increasingly important. Allowable error rate and undetected error probability in a network is also calculated for noisy channel. Error control is a system design technique that can fundamentally change the trade-offs in a communication system design. An error detection mechanism with low power consumption is also important for saving energy. In this paper, we have tried to find out how we can save our precious energy during communication among sensor nodes and bio mimetic fish-robots. Especially in security and error checking operation in underwater environment based on different error detection and correction mechanisms.

Imtiaz Ahmed Khan, Nam-Yeol Yun, Sardorbek Muminov, Soo-Hyun Park, Chang-Hwa Kim
Analysis on the Robotic Fish Propulsion for Various Caudal Fin Shapes

In this paper, propulsion characteristic of various shape caudal fin is analyzed. Using added mass theory, some caudal fins for robotic fish propulsion are analyzed. From this analysis, torque and propulsion characteristic for caudal fins with various shapes can be investigated. This method can be adopted to identify the caudal fin performance more simply and faster than computational method, although with less accuracy.

Dongwon Yun, Jinho Kyung, Chanhum Park
Development of a 3-DOF Fish Robot ‘ICHTHUS V5’

In recent, there is a rising interest on studying fish-like underwater robots because of real fish’s great maneuverability and high energy efficiency. However, researches about the fish-like underwater robots have not been investigated so much and there are still diverse problems in respect of using of the fish robot in the real environment such as in the river for detecting water pollution. For example, the fish robot has a short operating time and cannot move narrow passage such as swimming between aquatic plants. Therefore, this paper mainly describes a development of robotic fish which can be used for water quality monitoring system. The fish robot ‘Ichthus V5’ has a 3-DOF serial link-mechanism and is developed in KITECH. Furthermore, we propose a dynamic equation of the fish robot to use the underwater environment. We added several sensors to navigate autonomously in the real environment like river. Also, we added two kinds of sensor to detect temperature, electric conductivity, pH (hydrogen ion concentration) of water. Therefore, the developed system can be applied to environmental monitoring system for detect pollution or quality of river.

Gi-Hun Yang, Hyunjin Lee, Young Sun Ryuh
Research on Bio-mechanism Robotics by Robotic Fish Fin Technology

In the surgical operation, the research and development of the surgical robot that has both safety and effectiveness is needed in consideration of the physical mental strain decrease of the surgeon and the patient. Through the observation of the motion of fish, it has been found that fish swim efficiently using flexibility of pliable fins, giving the actuator the name of flexible oscillating fin propulsion system. The author proposed the application to the realm of healing the elastic vibration wing promotion system. The flexible forceps robot for the surgical operation of a vibrating flexible fin promotion system application was developed.

Zusong Gu, Ikuo Yamamoto, Tomokazu Hiratsuka
Simox: A Robotics Toolbox for Simulation, Motion and Grasp Planning

Software development plays a major role besides hardware setup and mechanical design when it comes to building complex robots such as mobile manipulators or humanoids. Different requirements have to be addressed depending on the application. A low-level controller for example must be implemented for realtime use, whereas a task planning component will interact with the robot on a higher abstraction level. Hence, developing robotics software is subject to several constraints such as performance and robustness.

N. Vahrenkamp, M. Kröhnert, S. Ulbrich, T. Asfour, G. Metta, R. Dillmann, G. Sandini
Directional Manipulability to Improve Performance Index of Dual Arm Manipulator for Object Grasping

Directional manipulability measure is used to improve the performance index of dual arm manipulator, instead of general manipulability measure. In this paper, a performance index of dual arm manipulator has been proposed and manipulability measure is an important part of it. General manipulability measure evaluates for the ability of manipulator to act in either direction. But, for a given task, abilities in either direction are desired and this is called desired manipulability measure. Although the general manipulability measure probably is bigger than desired manipulability measure, only the intersection part of actual and desired manipulability ellipsoids applies to positively influence performance index of dual arm manipulator. Simulation results show the difference and efficiency of these manipulability measures.

Hu Chen, Suk-In Lee, Jin-Hyun Do, Jang-Myung Lee
Gait Analysis for a Human with a Robot Walking Helper

With the growth of elderly population in our society, intelligent walking aids will play an important role in providing functional mobility to humans. In this paper, we propose a model to compute gait of humans walking with a robot helper. This model is aimed at designing a control system for the robot walking helper. The human model includes both the single support phase and impacts. Since a human will be walking along with the robot with its help, geometrical constraints and interaction forces are included. To achieve stable walking, zero moment point (ZMP) is utilized in the analysis and friction constraint is included within the reaction force from the ground. Simulations are performed to obtain optimal gait trajectories, the human applied joint torques, and the supporting forces from the robot walking helper.

Chun-Hsu Ko, Kuu-Young Young, Sunil K. Agrawal
Global Pose Estimation with Adaptive GPS/IMU Fusion

The aim of this study is to fuse GPS(Global Positioning System)/IMU(Inertial Measurement System) by using Kalman filter. We acquired the 6D pose data and compared the accuracy of 3D world model by using the data with Kalman filter and 3D world model by without filtering. Using proposed Kalman filter method, we obtain the exact pose data. This indicates that reconstructed 3D world model, using the proposed method is more accurate.

Taek Jun Oh, Myung Jin Chung
Sensor-Based Incremental Boustrophedon Decomposition for Coverage Path Planning of a Mobile Robot

This paper presents a sensor-based coverage algorithm with which a robot covers an unknown rectilinear region while simultaneously constructing cell decomposition. In this algorithm, cell boundaries are indicated by combined lines of critical edges those are sensed partial contours of wall and objects in the workspace. The robot uses the laser scanner to sense critical edges. The proposed algorithm incrementally (one by one) decomposes unknown environment into cells. The constructing order of the cells is very important in this incremental cell decomposition algorithm. In order to decide next target cell from candidate cells, the robot checks redundancy of the planned path and possible position of ending points of the current cell. The key point of the algorithm is memorizing the covered space to define the next target cell from possible cells more than one. Path generation within the defined cell is determined to minimize the number of turns because the number of turns is the main factor to save time for coverage. Therefore, the long boundary of cell should be chosen as a main path of the robot. Verification of this algorithm is done by the simulation under LABVIEW environment.

Dugarjav Batsaikhan, Adiyabaatar Janchiv, Soon-Geul Lee
Distributed Traffic Signal Control Using PSO Based on Probability Model for Traffic Jam

In this article, a new traffic signal control method is proposed. The Bayesian Network (BN) model and the Cellular Automaton (CA) model are used to build up a probability model for traffic jam. And then using the Particle Swarm Optimization (PSO) based on the probability model, the optimal traffic signal can be obtained. Finally, the effectiveness of the proposed method is shown with a micro-traffic simulator.

Cheng-You Cui, Hee-Hyol Lee
Design of a Sensing Limited Autonomous Robotic System

This paper describes the development and implementation of a behavioral-based solution for a sensing-limited robotic system in area coverage problem using the LEGO Mindstorms NXT robotics kit. The main aim is to investigate how area coverage algorithms can be implemented on a robot with limited sensing and processing capabilities to cover a given area efficiently without localization or map building as well as to compare these algorithms with each other. Interestingly, there has been limited research done in this aspect, particularly in the efficiencies of current heuristic based algorithms in commercially available robots. In this project, three behaviors: Random Walk, Spiral, and Weave are proposed. A robot was constructed to mimic a sensing limited robot and it was used to carry out the proposed behaviors to determine the most efficient behavior. During experiment, it was found that aspects such as the parameter being measured and placement of obstacles in the environment affected the perception and performance by the robot. Results showed that the zigzag motion of Weave was the most efficient movement of the three, performing consistently well under varied environments and measurements.

Lim Han Yang Benjamen, Marcelo H. Ang Jr.
On Sub-modularization and Morphological Heterogeneity in Modular Robotics

Modular robots are a kind of robots built from mechatronic modules, which can be assembled in many different ways allowing the modular robot to assume a wide range of morphologies and functions. An important question in modular robotics is to which degree modules should be heterogeneous. In this paper we introduce two contributing factors to heterogeneity namely morphological heterogeneity and sub-functional modularization. Respectively, the ideas are to create modules with significantly different morphologies and to spread sub-functionality across modules. Based on these principles we design and implement the Thor robot and evaluate it by participating in the ICRA Planetary Robotic Contingency Challenge. The Thor robot demonstrates that sub-functional modularity and morphological heterogeneity may increase the versatility of modular robots while reducing the complexity of individual modules, which in the longer term may lead to more affordable modular robots.

A. H. Lyder, K. Stoy, R. F. M. Garciá, J. C. Larsen, P. Hermansen
Balancing Control of Unicycle Robot

This paper proposes controlling balance of unicycle robot by using PID control. The robot consists of the Pitch and Roll. And the robot measures the Pitch and Roll angle which are obtained from IMU sensor. In this paper, Assuming that the Pitch and Roll is decoupled, design the controller separately. Because controller’s performance is confirmed from simulation, the dynamic equations of the robot are derived for simulation. Finally, designed controller is applied to have a test of the unicycle robot.

In-Woo Han, Jae-Won An, Jang-Myung Lee
Autonomy for Mobility on Demand

We describe the development of our autonomous personal vehicle that attempts to provide mobility on demand service to address the first- and last-mile problem. We discuss the challenges faced for such a system in a campus environment and discuss our approach towards mitigating them. The autonomous vehicle has operated over 30km of autonomous operation in a campus environment interacting with pedestrian and human driven vehicles.

Z. J. Chong, B. Qin, T. Bandyopadhyay, T. Wongpiromsarn, B. Rebsamen, P. Dai, E. S. Rankin, M. H. Ang Jr.
Interleaving Planning and Control of Mobiles Robots in Urban Environments Using Road-Map

This paper presents a robot solution that allows to automatically reach a set of goals attributed to a robot. The challenge is to design autonomous robots assigned to perform missions without a predefined plan. We address the stochastic salesman problem where the goal is to visit a set of points of interest. A stochastic Road-Map is defined as a topological representation of an unstructured environment with uncertainty on the path achievement. The Road-Map allows us to split deliberation and reactive control. The proposed decision making uses a computation of Markov Decision Processes (MDPs) in order to plan all the reactive tasks to perform while there are goals not yet reached. Finally, from a brief explanation on how the approach could be extend to multi-robot missions, experiments in real conditions permit to evaluate the proposed architecture for multi-robot stochastic salesmen missions.

Guillaume Lozenguez, Lounis Adouane, Aurélie Beynier, Abdel-Illah Mouaddib, Philippe Martinet
Aggressive Manuevering of Unmanned Helicopters: Learning from Human Based on Neural Networks

“Teaching by Showing” control of a small helicopter’s aggressive maneuvering often needs inner aided controllers based on helicopter’s dynamics, which is very complex to identify. In this paper, a neural network based control is proposed, based on the identification of the relationship between the pilot’s control and flight states, and it is a model-free control method. Flight test is done in simulation environment based on real flight data. The results show the effectiveness of the neural network based controller for aggressive flight control.

Dalei Song, Chong Wu, Juntong Qi, Jianda Han
Robust Object Recognition in Unstructured Environments

One of the main goals of Robotics research is to help human beings in their daily tasks. However, physical interaction in everyday human scenarios requires robot systems endowed with rich sensory-motor skills and multisensory feedback that enables them to exhibit levels of adaptability high enough to achieve their goals. In this context, vision plays a main role since it provides rich information about the state of the environment. Although it is an active research area, there are still some challenging issues to be solved. Among them, the research presented in this paper addresses the object recognition for manipulation tasks in unstructured environments from a visual input. We have designed a mechanism that provides a

background model

covering the whole system’s peripersonal space such that the objects of interest can be always identified without any information about the system. Different objects and scene conditions have been considered to assess approach’s performance by showing robust object recognition in all the cases.

Ester Martínez-Martín, Angel P. del Pobil
Mapping of Incremental Dynamic Environment Using Rao-Blackwellized Particle Filter

This research is a preliminary research for real time autonomous robot in unknown incremental dynamical environment. A general method for mapping incremental dynamic environment using Multiple Target tracking (MTT) was proposed in this research. Rao-Blackwellized Particle Filter (RBPF) was used for the multiple moving obstacles tracking problem. Firstly data association problem was solved via Multiple Hypothesis Tracking (MHT) data association by a new method. The new MHT method can use extra information except using only position of targets. Particle Filter is used in the method. Each particle is assumed as an obstacle map. Then tracking problem for each obstacle in the particle is solved by Extended Kalman Filter (EKF). Finally the particle which has highest weight is assumed as the dynamic map. Additionally a new resampling method was proposed in this research. The algorithm can cope with new obstacles and false detection according to the pure particle filter. Obstacles are assumed as human in this research hence their velocities are determined randomly up to human walking speed. Furthermore the robot moves approximately at human walking speed. A graphical user interface program was constituted in MATLAB so different states are surveyed.

Alper Öner
Texture-Based Crowd Detection and Localisation

This paper presents a crowd detection system based on texture analysis. The state-of-the-art techniques based on co-occurrence matrix have been revisited and a novel set of features proposed. These features provide a richer description of the co-occurrence matrix, and can be exploited to obtain stronger classification results, especially when smaller portions of the image are considered. This is extremely useful for crowd localisation: acquired images are divided into smaller regions in order to perform a classification on each one. A thorough evaluation of the proposed system on a real world data set is also presented: this validates the improvements in reliability of the crowd detection and localisation.

Stefano Ghidoni, Grzegorz Cielniak, Emanuele Menegatti
Polar Histogram Based Sampling Method for Autonomous Vehicle Motion Planning

In this paper we present a sampling based motion planning algorithm for an autonomous vehicle, which allowed our vehicle to navigate smoothly at high speed with limited computation resources. A new sampling method, limiting candidate states, is introduced to reduce computation burden, associated with sampling based motion planning algorithms. The proposed method is experimentally evaluated driving an autonomous vehicle at speeds up to 60 km/h. It showed how advantages of both sampling based and space discretization based planning algorithms can be combined in one method, providing short planning time in higher dimensional configuration space and good performance in narrow and cluttered environment.

Dmitriy Ogay, Jee-Hwan Ryu, Eun-Gyung Kim
A Multiresolution Approach for Real-Time Motion Planning under Differential Constraints

In this paper we present a complex approach to resolve a contradiction in planning between space discretization based algorithms and sampling RRT-like algorithms in a way, appropriate for a real time motion planning with limited computation resources. The former type planning algorithms have advantage that they can systematically explore discretized space, and are less vulnerable to the problem of navigation in narrow passages, but from the other side, to achieve satisfactory results in terms of path smoothness and traversability, higher resolution and additional dimensions in configuration space might be required, which makes use of such algorithms problematic for real time motion planning, because of curse of dimensionality. RRT-like sampling based algorithms have advantage, that they can explore fast in high dimensional configuration space, but they may spent indefinitely long time if they encounter narrow passages. For the safe motion it is desired that motion planner gives response to the plan execution part in a limited time. The proposed solution implements this feature in a more efficient way than existing solutions. The proposed solution was tested in a real autonomous driving, including parking with backward motion.

Dmitriy Ogay, Jee-Hwan Ryu, Eun-Gyung Kim
A Study of Path Planning Algorithm Based on the Survival Probability

After the development of fusion technology of mechanics, electronics and IT, there are a lot of researches about the autonomous vehicle. For this kind of vehicle to move automatically, it needs the reference input that the vehicle can follow, and this reference input comes from a path planning algorithm. A* is one of the well-known global path planning algorithm that finds the optimal path on the given map using the heuristic cost function. However, A* algorithm gives the vehicle some reference points on the path not the continuous reference path. And A* may not guarantee the path will not collide the corner of the obstacle. To overcome this problem, in this paper, the hybrid path planning algorithm is suggested using B-spline equation and new heuristic cost function with survival probability. At last, we developed the simulation program and the results are shown.

Min-Ho Kim, Chi-Beom Noh, Jung-Hun Heo, Min-Cheol Lee
Gaze Control-Based Navigation Architecture for Humanoid Robots in a Dynamic Environment

Due to the limited information from the environment using a local vision sensor, gaze control research is very important for humanoid robots. In addition, multiple objectives for navigation have interactive relationships among them. From this point of view, this paper proposes a gaze control-based navigation architecture using fuzzy integral and fuzzy measure for humanoid robots. Four criteria are employed along with their partial evaluation functions in order to determine the final gaze direction. By employing fuzzy integral approach for the global evaluation for candidate gaze directions, effective gaze control considering the interactive phenomena among criteria is accomplished and verified through a simulation using a developed simulator for HanSaRam-IX (HSR-IX).

Jeong-Ki Yoo, Jong-Hwan Kim
A Multimodal Distributed Intelligent Environment for a Safer Home

This paper presents an ambient intelligence system aimed at increasing the security of elderly people in their living environment. The proposed system processes both video and audio signals to detect dangerous events and trigger automatic warnings. The system is implemented through a software middleware for multimedia streaming and processing in which the information is processed in several, distributed software nodes. This paper focuses on the fall detection video processing node and on the sound classification and localization audio processing node. The system has been tested both in a laboratory and in a real-world scenario, demonstrating high performance levels.

Salvatore M. Anzalone, Stefano Ghidoni, Emanuele Menegatti, Enrico Pagello
Comparison between Photo Interrupter and Giant Magnetoresistive Sensor for Auto Focusing System in the Digital Camera

Photo interrupter (PI) sensors are used to focus the lens in the optical devices such as digital camera, close circuit camera etc. It is cheap and easy to control. Despite of the advantages it has some limitations such as it is difficult to assemble, needs small tolerance and lacks high speed and accuracy. Magnetoresistive (MR) especially giant magnetoresistive (GMR) sensor provides the solution of PI sensor’s weak points. Furthermore, using GMR sensor shows better performances than PI sensor for the focusing system.

Sakura Sikander, Han Nam Lee, Hee-Je Kim
Natural Terrain Detection and SLAM Using LIDAR for UGV

This paper describes a natural terrain detection algorithm and a SLAM algorithm using a LIDAR sensor for an unmanned ground vehicle. We describe how features are detected from natural terrain, and then we localize the vehicle’s position and compose a map with the detected features. The LIDAR equipped on the experimental vehicle to scan natural terrain. The scan data is included many kinds of intrinsic disturbance on uneven terrain: a banded tree, a branch of a tree, uniform size of bush, undefined or unexpected objects. We apply a RANSAC (RANdom SAmple Consensus) algorithm to discriminate ground point cloud data and object point cloud data, and then separate bush point cloud data and tree point cloud data by two combination algorithms; GMM (Gaussian Mixture Model) and EM (Expectation Maximization). Both GMM and EM algorithms are for extracting features and classifying groups, respectively. We propose the double FCM (Fuzzy C-mean clustering) algorithm to robustly estimate the number of trees and its position. The Extended Kalman Filter approach to simultaneous localization and mapping (EKF-SLAM) is composed of extracted tree features. The

mahalanobis

distance is applied to remain consistency for feature correspondence which is for data association. Finally, we show the results which is experienced in a tree-filled mountain.

Kuk Cho, SeungHo Baeg, Sangdeok Park
Positioning Accuracy Improvement of Laser Navigation Using Unscented Kalman Filter

This paper presents positioning improvement of a laser navigation system (LNS) using unscented Kalman filter (UKF) for an automatic guided vehicle (AGV). The existing AGVs mainly used a magnetic system or an inductive system as a guidance system. However, those have high cost and difficult maintenance according to change of environment, and can drive only the designated path which sensors are installed on. The laser guidance system is developed to solve those problems, but it has also problems which is slow response time and low accuracy. Therefore, we propose a sensor fusion method for the AGV. The sensors used in this paper are encoders, a gyro and the LNS, and they are fused by UKF. To analyze the performance of the proposed system, we designed a fork-type AGV for ourselves, and performed the experiment that was repeated 5 times under the same working conditions. In experimental results, we verified that the proposed method could improve positioning accuracy of the LNS effectively. In addition, it was appropriate to apply a real AGV system for autonomous driving.

Jungmin Kim, Kyunghoon Jung, Jaeyong Kim, Hajun Song, Sungshin Kim
Improvement of Position Accuracy of Magnetic Guide Sensor Using Kalman Filter

This paper is represented to research of method to improve the performance of magnetic guide sensor using Kalman filter. The magnetic guide sensor is calculating the center position of the AGV (Automatic Guided Vehicle) by to measure the magnetic information of a magnetic substance that is magnetic tape, magnet spot. The existing magnetic guide sensor is the device that calculates the center position of the AGV using the magnetic force of the measured data that is one pole to measure each axis at the histogram algorithm. But, the existing method is unfit for requiring the high accuracy such as industrial setting because of interference between sensors and the effect by disturbance. Therefore, in this paper proposed method that increases the position accuracy of the magnetic guide sensor using Kalman filter. To verify the proposed method, we use the AGV to install magnetic guide sensor. And it compare the positioning accuracy of the propose method and the commercialized magnetic guide sensor. As a result, the proposed method was found 24.78% to improve the positioning accuracy of the proposed method than that of the commercialized magnetic guide sensor.

Eunkook Jung, Jungmin Kim, Hyunhak Cho, Junha Lee, Sungshin Kim
Indoor Flight Testing and Controller Design of Bioinspired Ornithopter

Indoor flight testing of a bioinspired ornithopter is conducted in this study and the dominant flight state variables such as body pitch angle, forward flight speed, altitude, wings and tail motions of the freely flying ornithopter are simultaneously measured by using a three-dimensional visual tracking system. A control-oriented system model of the ornithopter in trimmed level flight is established based on the recorded inputs and outputs dataset and the system matrices are fitted in a least-squares sense. To reduce the amplitude of the ornithopter body oscillations, the identified linear time-invariant system model is formulated to a disturbance-rejection problem and an optimal controller minimizing the quadratic performance index is designed. The continuous wing motion defined as the known disturbance deteriorates the pitch balance with respect to the center of gravity; however, the designed feedforward and feedback controller periodically activates the ornithopter tail and successfully reduces the magnitudes of the body oscillations.

Jun-Seong Lee, Jae-Hung Han
Effect of Passive Body Deformation of Hawkmoth on Flight Stability

In this study, the effect of passive body deformation on flight stability during insect flapping flight is investigated numerically. We developed a flexible body dynamic solver for a three-dimensional flexible beam model and coupled it with an

in-house

fluid dynamics solver. With this integrated model, hawkmoth free flights are simulated and analyzed systematically with six cases, in which the joint stiffness between thorax and abdomen varied from extremely rigid to very flexible. Our results indicate that the passive body deformation works likely altering the aerodynamic torque, the body attitude and the flight trajectory. We further found that the most stable flight can be achieved by a moderate joint stiffness, in which the body attitude remains approximately around the initial angle of 40 degree. This points to the importance that the flexible body and its passive deformation during flapping-wing flight are capable to enhance stable flight and flight control.

Ryusuke Noda, Masateru Maeda, Hao Liu
ACS-PRM: Adaptive Cross Sampling Based Probabilistic Roadmap for Multi-robot Motion Planning

In this paper we present a novel approach to multi-robot motion planning by using a probabilistic roadmap (PRM) based on adaptive cross sampling (ACS). The proposed approach, we call ACS-PRM, consists of three steps, which are C-space sampling, roadmap building and motion planning. Firstly, a sufficient number of points should be generated in C-space on an occupancy grid map by using an adaptive cross sampling method. Secondly, a roadmap should be built while the potential targets and milestones are extracted by post-processing the result of sampling. Finally, the motion of robots should be planned by querying the constructed roadmap. In contrast to previous approaches, our ACS-PRM approach is designed to plan separate kinematic paths for multiple robots to minimize the problem of congestion and collision in an effective way so as to improve the planning efficiency. Our approach has been implemented and evaluated in simulation. The experimental results demonstrate the total planning time can be significantly reduced by our ACS-PRM approach compared with previous approaches.

Zhi Yan, Nicolas Jouandeau, Arab Ali Cherif
On-Line Learning of the Visuomotor Transformations on a Humanoid Robot

In infant primates, the combination of looking and reaching to the same target is used to establish an implicit sensorimotor representation of the peripersonal space. This representation is created incrementally by linking together correlated signals. Also, such a map is not learned all at once, but following an order established by the temporal dependences between different modalities, which is imposed by the choice of the vision as master signal. Indeed, visual feedback is used both to correct gazing movements and to improve eye-arm coordination. Inspired by these observations we have developed a framework for building and maintaining an implicit sensorimotor map of the environment. In this work we present how this framework can be extended to allow a humanoid robot to update on-line the sensorimotor transformations among visual, oculomotor and arm-motor cues.

Marco Antonelli, Eris Chinellato, Angel P. Del Pobil
Adaptive Face Recognition for Low-Cost, Embedded Human-Robot Interaction

This paper presents an accelerated AdaBoost face detection algorithm and an incremental PCA-based face recognition algorithm for human robot interactive applications. The accelerated AdaBoost algorithm utilizes an image resizing technique and a skin color filter for detecting face regions. To track a detected face precisely and efficiently while also recognizing the face, a hybrid face tracking approach is applied based on an adaptive skin color mode and an estimated potential face area. In addition, a novel adaptive face recognition method is implemented by automatically upgrading the set of sample faces of a

known

person and collecting new samples of an

unknown

person with incrementally enhanced recognition performance. These algorithms are well suited for embedded systems, such as socially interactive robots, because of their cost and time efficiency and little pre-training required for reliable performance.

Yan Zhang, Kenneth Hornfeck, Kiju Lee
Backmatter
Metadata
Title
Intelligent Autonomous Systems 12
Editors
Sukhan Lee
Hyungsuck Cho
Kwang-Joon Yoon
Jangmyung Lee
Copyright Year
2013
Publisher
Springer Berlin Heidelberg
Electronic ISBN
978-3-642-33926-4
Print ISBN
978-3-642-33925-7
DOI
https://doi.org/10.1007/978-3-642-33926-4

Premium Partner