Showing papers presented at "International Conference on Control, Automation, Robotics and Vision in 2014"

Medical image classification with convolutional neural network

Abstract: Image patch classification is an important task in many different medical imaging applications. In this work, we have designed a customized Convolutional Neural Networks (CNN) with shallow convolution layer to classify lung image patches with interstitial lung disease (ILD). While many feature descriptors have been proposed over the past years, they can be quite complicated and domain-specific. Our customized CNN framework can, on the other hand, automatically and efficiently learn the intrinsic image features from lung image patches that are most suitable for the classification purpose. The same architecture can be generalized to perform other medical image or texture classification tasks.

Towards dense moving object segmentation based robust dense RGB-D SLAM in dynamic scenarios

Abstract: Based on the latest achievements in computer vision and RGB-D SLAM, a practical way for dense moving object segmentation and thus a new framework for robust dense RGB-D SLAM in challenging dynamic scenarios is put forward. As the state-of-the-art method in RGB-D SLAM, dense SLAM is very robust when there are motion blur or featureless regions, while most of those sparse feature-based methods could not handle them. However, it is very susceptible to dynamic elements in the scenarios. To enhance its robustness in dynamic scenarios, we propose to combine dense moving object segmentation with dense SLAM. Since the object segmentation results from the latest available algorithm in computer vision are not satisfactory, we propose some effective measures to improve upon them so that better results can be achieved. After dense segmentation of dynamic objects, dense SLAM can be employed to estimate the camera poses. Quantitative results from the available challenging benchmark dataset have proved the effectiveness of our method.

Wind-optimal path planning: Application to aircraft trajectories

Abstract: In this paper, an algorithm to plan a continuous wind-optimal path is proposed, and simulations are made for aircraft trajectories. We consider a mobile which can move in a two dimensional space. The mobile is controlled only by the heading direction, the speed of the mobile is assumed to be constant. The objective is to plan the optimal path avoiding obstacles and taking into account wind currents. The algorithm is based on Ordered Upwind Method which gives an optimality proof for the solution. The algorithm is then extended to spherical coordinates in order to be able to handle long paths.

A new feedback linearization LQR control for attitude of quadrotor

Abstract: This paper presents a new robust and stable controller for attitude of quadrotor. We proposed a combination of feedback linearization and LQR (linear quadratic regulator) control strategy. To our best acknowledge, this method is firstly employed to stabilize the attitude of the quadrotor with disturbance. Numerical simulations demonstrate the stabilizations and robustness of the control system under nominal conditions. Furthermore, a bounded disturbance was added to the system in flight test, and the results also illustrate the successful behavior of the controller proposed.

Feature extraction and tracking of a weld joint for adaptive robotic welding

Abstract: Recent advances in automation and sensor technology have enabled the use of industrial robots for complex tasks that require intelligent decision making. Vision sensors have been the most successfully used sensor in many high value industrial applications. Over the recent years, weld seam tracking has been a topic of interest, as most of the existing robotic welding systems operate on basis of pre-programmed instructions. Such automated systems are incapable of adapting to unexpected variations in the seam trajectory or part fit-up. Applications such as tungsten inert gas (TIG) welding of aerospace components require high tolerances and needs intelligent decision making. Such decision making procedure has to be based on the weld groove geometry at any instance. In this study, a novel algorithm along with an automated system was developed for estimating the joint profile and path tracking of a three dimensional (3D) weld groove. A real-time position based closed-loop system was developed with a six axis industrial robot and a laser triangulation based sensor. The system was capable of finding the 3D weld joint profile and position in real-time, and make intelligent decisions accordingly. Raw data from a vision sensor was processed through a novel algorithm to obtain X and Z co-ordinates at an accuracy of 8.3μm and 43μm respectively at an acquisition speed of 2.5 profiles per second. The algorithm was also capable of measuring the weld gaps with an accuracy of 28μm. Finally, the developed system was successfully used for three dimensional seam tracking, and demonstrates an accuracy of ±0.5mm at a tracking a speed of 2mm/s.

Multiple obstacle detection and tracking using stereo vision: Application and analysis

Abstract: Vision systems provide a large functional spectrum for perception applications and, in recent years, they have demonstrated to be essential in the development of Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicles. In this context, this paper presents an on-road objects detection approach improved by our previous work in defining the traffic area and new strategy in obstacle extraction from U-disparity. Then, a modified particle filtering is proposed for multiple object tracking. The perception strategy of the proposed vision-only detection system is structured as follows : First, a method based on illuminant invariant image is employed at an early stage for free road space detection. A convex hull is then constructed to generate a region of interest (ROI) which includes the main traffic road area. Based on this ROI, an U-disparity map is built to characterize on-road obstacles. In this approach, connected regions extraction is applied for obstacles detection instead of standard Hough Transform. Finally, a modified particle filter framework is employed for multiple targets tracking based on the former detection results. Besides, multiple cues, such as obstacle's size verification and combination of redundant detections, are embedded in the system to improve its accuracy. Our experimental findings demonstrates that the system is effective and reliable when applied on different traffic video sequences from a public database.

Smart grid cyber security

Abstract: The trend of integrating power systems with advanced computer and communication technologies has introduced serious cyber security concerns, especially in a smart grid environment where the cyber system is no longer regarded as 100% reliable to support power system communications and control as before. Power system security therefore extends to potential cyber security domain in the smart grid era. Risks from the cyber system as well as non-conventional physical power system contingencies start to contributing to the overall grid security. This will be particularity important considering the potential risks from targeted attacks on vulnerable system components which may bring done the overall system. The presentation gives an overview of the work done by the research team on power system security, including conventional stability as well as cyber security assessment. A framework for smart grid cyber security and vulnerability assessment will be illustrated as well. The framework includes two main components, which are respectively cyber system security assessment and fast power system security assessment. Complex networks theory and data mining based approaches are also employed to identify the vulnerable components of the physical power system. The proposed cyber system models can be integrated with existing power system models to study the complex interactions between the cyber and physical parts of the smart grid. Advanced modeling tools are proposed to model cyber attacks and evaluate their impacts on smart grid security have been developed as well.

Automated three-stage nucleus and cytoplasm segmentation of overlapping cells

Abstract: Developing segmentation techniques for overlapping cells has become a major hurdle for automated analysis of cervical cells. In this paper, an automated three-stage segmentation approach to segment the nucleus and cytoplasm of each overlapping cell is described. First, superpixel clustering is conducted to segment the image into small coherent clusters that are used to generate a refined superpixel map. The refined superpixel map is passed to an adaptive thresholding step to initially segment the image into cellular clumps and background. Second, a linear classifier with superpixel-based features is designed to finalize the separation between nuclei and cytoplasm. Finally, edge and region based cell segmentation are performed based on edge enhancement process, gradient thresholding, morphological operations, and region properties evaluation on all detected nuclei and cytoplasm pairs. The proposed framework has been evaluated using the ISBI 2014 challenge dataset. The dataset consists of 45 synthetic cell images, yielding 270 cells in total. Compared with the state-of-the-art approaches, our approach provides more accurate nuclei boundaries, as well as successfully segments most of overlapping cells.

Lane marking based vehicle localization using particle filter and multi-kernel estimation

Abstract: Vehicle localization is the primary information needed for advanced tasks like navigation. This information is usually provided by the use of Global Positioning System (GPS) receivers. However, the low accuracy of GPS in urban environments makes it unreliable for further treatments. The combination of GPS data and additional sensors can improve the localization precision. In this article, a marking feature based vehicle localization method is proposed, able to enhance the localization performance. To this end, markings are detected using a multi-kernel estimation method from an on-vehicle camera. A particle filter is implemented to estimate the vehicle position with respect to the detected markings. Then, map-based markings are constructed according to an open source map database. Finally, vision-based markings and map-based markings are fused to obtain the improved vehicle fix. The results on road traffic scenarios using a public database show that our method leads to a clear improvement in localization accuracy.

Predicting traffic speed in urban transportation subnetworks for multiple horizons

Abstract: Traffic forecasting is increasingly taking on an important role in many intelligent transportation systems (ITS) applications. However, prediction is typically performed for individual road segments and prediction horizons. In this study, we focus on the problem of collective prediction for multiple road segments and prediction-horizons. To this end, we develop various matrix and tensor based models by applying partial least squares (PLS), higher order partial least squares (HO-PLS) and N-way partial least squares (N-PLS). These models can simultaneously forecast traffic conditions for multiple road segments and prediction-horizons. Moreover, they can also perform the task of feature selection efficiently. We analyze the performance of these models by performing multi-horizon prediction for an urban subnetwork in Singapore.

An evidential sensor model for Velodyne scan grids

Abstract: For the development of driving assistance systems and autonomous vehicles, a reference perception equipment including navigable space determination and obstacles detection is a key issue. The Velodyne sensor which provides high definition and omnidirectional information can be used for this purpose. Nevertheless, when scanning around the vehicle, uncertainty necessarily arises due to unperceived areas and noisy measurements. This paper proposes an inverse evidential model for the Velodyne in order to exploit its measurements in a 2D occupancy grid mapping framework. The evidential sensor model interprets the data acquired from the Velodyne and successively maps it to a Carthesian evidential grid using a fusion process based on the least commitment principle to guarantee information integrity. Experimental results prove that this approach can handle efficiently the uncertainties of the sensor and thus a highly reliable local reference map near the vehicle can be built for every timestamped perception system that needs evaluation or calibration.

Image-based plant stornata phenotyping

Abstract: We propose in this paper a fully automatic approach for image-based plant stornata phenotyping. Given a microscopic image of a plant leaf surface, our goal is to automatically detect stornata cells and measure their morphological and structural features, such as stornata opening length and width, and size of the guard cells. The main challenge in developing such tool is the lack of contrast between the stornata cell region and its surrounding background. Our approach uses template matching to detect individual stornata cells and local analysis to measure stornata features within the detected stornata regions. It is fully automatic and computationally efficient. Thus, it will enable plant biologists to perform large scale analysis of stornata morphology, which in turn will help in developing understanding and controlling plant's response to various environmental stresses (e.g. drought and soil salinity).

Dynamic non-prehensile object transportation

Abstract: When possible, non-prehensile transportation (ie transporting objects without grasping them) can be faster and more efficient than prehensile transportation However, the need to explicitly consider reaction and friction forces yields kino-dynamic constraints that are difficult to take into account by traditional planning algorithms Based on the recently developed Admissible Velocity Propagation algorithm, we propose here a fast and general non-prehensile transportation scheme Our contribution is twofold First we show how to cast the dynamic balance constraints of a 3D object (eg a bottle) into a form compatible with the AVP algorithm Second, we extend the AVP-RRT algorithm into a more efficient AVP-biRRT algorithm, which makes use of the idea of concurrently growing two trees, one rooted at the starting configuration and one rooted at the goal configuration We also show both in simulations and on a real robot how our algorithm allows planning fast and dynamic trajectories for the non-prehensile transportation of a bottle

Multi-label classification method based on extreme learning machines

Abstract: In this paper, an Extreme Learning Machine (ELM) based technique for Multi-label classification problems is proposed and discussed. In multi-label classification, each of the input data samples belongs to one or more than one class labels. The traditional binary and multi-class classification problems are the subset of the multi-label problem with the number of labels corresponding to each sample limited to one. The proposed ELM based multi-label classification technique is evaluated with six different benchmark multi-label datasets from different domains such as multimedia, text and biology. A detailed comparison of the results is made by comparing the proposed method with the results from nine state of the arts techniques for five different evaluation metrics. The nine methods are chosen from different categories of multi-label methods. The comparative results shows that the proposed Extreme Learning Machine based multi-label classification technique is a better alternative than the existing state of the art methods for multi-label problems.

High-throughput 3D reconstruction of plant shoots for phenotyping

Abstract: In this paper we report on development and analysis of a high throughput 3D reconstruction set-up for reconstructing cereal plants grown in pots for their phenotypic analysis. We are motivated with the idea of accurate and high-throughput reconstruction of shoots. We have developed a turntable setup for high-throughput and high accuracy 3D reconstruction of plant shoots, where the plants stays stationary and the camera moves on a circular path to take multi-view digital images. The leaves of cereal plants like wheat, barley, corn, etc. are very delicate and their movement during imaging causes errors in the volumetric reconstruction of the shoots. The synchronization of the turntable and camera has been designed by us. A novel and easy to use camera calibration for single axis motion is designed and implemented. Visual hull algorithm has been used for 3D reconstruction. We analyse results of reconstruction for three different modes of image acquisition against ground truth with respect to time taken for imaging and the number of images used. In results we obtain phenotypic quality 3D volumetric reconstruction with an acquisition time of less than a minute per potted plant.

Human detection in uncluttered environments: From ground to UAV view

Abstract: Nowadays pedestrian detectors are fast, scale-robust and quite efficient. Embedded within a UAV such a detector would open new possibilities. In this paper the very well known HOG detector is adapted for UAV use and a new kind of training dataset is proposed in order to increase the detector's angular robustness. A more appropriate set of detection windows, together with a new detection pipeline, is proposed in order to reduce the search space and consequently reduce the computation time. Tests conducted using the improved detector show significantly better results on aerial images.

Geometry-based static hand gesture recognition using support vector machine

Abstract: Static hand gesture recognition plays an important role in developing a system for human-computer interaction. Besides, such systems can be also used by the deaf community in order to convey information through gestures instead of words. A vision-based processing of hand gesture recognition consists of three main stages: preprocessing, feature extraction and identification. In this paper, the first stage involves following two sub-stages: segmentation which locates hand using color information and extracts its silhouette; separation that separates arm, the part with less information, based on geometrical properties. In the second stage, features which extracted from hand-without-arm are general (ratio of width to height, wrist angle and number of fingers) and detailed (calculated based on fingertips and cross sections) characteristics. Finally, support vector machine model with "max-wins" voting strategy is used to classify the hand gestures. The experiment is conducted on color image dataset of Polish Ministry of Science and Higher Education, with 89.5% classification accuracy.

Optimal economic dispatch by fast distributed gradient

Abstract: Concerning on optimal economic dispatch, interior point method via 6-logarithmic barrier is employed to reformulate the cost function of power generation Fully distributed technology-enabled algorithm is developed to solve the economic dispatch More specifically, the minimum connected dominating set based distributed algorithm aims at efficiently allocating the task of supply-demand balance for the whole power grid A fast gradient based distributed optimization method is designed to fast converge to optimal solution The simulations illustrate the effectiveness and good performance of our algorithms

L 1 adaptive control for quadcopter: Design and implementation

Abstract: Unmanned Aerial Vehicles have a lot of potentials in outdoor applications. However, uncertainties such as wind disturbances and mass change when performing some particular tasks, greatly affect their tracking performance. This paper presents a methodology using L 1 adaptive control to address some of the robustness issues of the quadcopter in outdoor flight which significantly improves the performance comparing to the baseline controller. Simulation and flight tests verify the potential of the presented controller.

Saturated feedback control for an automated parallel parking assist system

Abstract: This paper considers the parallel parking problem of automatic front-wheel steering vehicles. The problem of stabilizing the vehicle at desired position and orientation is seen as an extension of the tracking problem. A saturated control is proposed which achieves quick steering of the system near the desired position of the parking spot with desired orientation and can be successfully used in solving parking problems. In addition, in order to obtain larger area of the starting positions of the vehicle with respect to the parking spot for the first reverse maneuver of the parallel parking, an approach of using saturated control with two different levels of saturation is proposed. The vehicle can be automatically parked by using one or multiple maneuvers, depending on the size of the parking spot. Simulation results are presented to confirm the effectiveness of the proposed control schemes.

Mechanical development and control of a miniature nonholonomic spherical rolling robot

Abstract: In this paper, a miniature nonholonomic spherical rolling robot capable of navigating over two dimensional surfaces is described. This 55 gram spherical robot consists of a 6 cm diameter external spherical shell driven by an internal two-wheeled differential drive cart. A gravity powered pendulum effect is produced as the internal device climbs up the internal surface of the shell, propelling the robot forward up to a speed of 0.16m/s. We derived its dynamic model using Lagrangian, and studied its dynamics and performance under various applied torques. The spherical robot is built and its overall mechanical, hardware and control architecture are elaborated. Experiments are conducted to evaluate the robot open and closed loop performance on a linear trajectory, captured using an optical motion capture system. We showed that with our implemented PD controller, the robot can follow the desired orientation.

Design of retuning fractional PID controllers for a closed-loop magnetic levitation control system

Abstract: In this paper, we study the problem of fractional-order PID controller design for an unstable plant — a laboratory model of a magnetic levitation system. To this end, we apply model based control design. A model of the magnetic levitation system is obtained by means of a closed-loop experiment. Several stable fractional-order controllers are identified and optimized by considering isolated stability regions. Finally, a nonintrusive controller retuning method is used to incorporate fractional-order dynamics into the existing control loop, thereby enhancing its performance. Experimental results confirm the effectiveness of the proposed approach. Control design methods offered in this paper are general enough to be applicable to a variety of control problems.

Identification of potential biomarkers in the hippocampus region for the diagnosis of ADHD using PBL-McRBFN approach

Abstract: Attention Deficiency Hyperactivity Disorder (ADHD) as a disruptive behavior disorder is receiving lots of attention because of its complexity and need for early detection. This paper presents a study on identification of potential biomarkers in the diagnosis of ADHD based on the structural-MRI of the brain obtained through ADHD-200 competition data set. The region of the brain considered here is "hippocampus". The grey matter probability of the T1 images is segmented followed by tissue alignment and inter subject normalization. Then, the voxels of the hippocampus are segregated using a region-of-interest mask, and the grey matter tissue probability values are obtained. These values are then used as features to classify ADHD patients against typically developing controls using a projection based learning algorithm for a meta-cognitive radial basis function network (PBL-McRBFN) and compared the results with that of support vector machines. Initially we take all the voxels of hippocampus for our study and then we have selected the most relevant voxels as a biomarker using Chi-square approach and developed a classifier to diagnosis ADHD. The results clearly highlight that use of hippocampus from the structural-MRI is sufficient to diagnosis ADHD to certain degree of confidence.

Material classification based on thermal and surface texture properties evaluated against human performance

Abstract: Effective robotic grasping and manipulation requires knowledge about the surface properties of an object and the environment in which it is located. Physical contact with materials using tactile sensors can enable the retrieval of detailed information about the material, i.e. compressibility, surface texture and thermal properties. This paper describes a system used to classify a wide range of materials based on their thermal properties and surface texture. Following acquisition of data from a sophisticated tactile sensor, the system uses principal component analysis (PCA) to extract features from the data which are used to train an Artificial Neural Network (ANN) to classify materials, first into groups and then as individual materials. The system is compared with human performance and the results demonstrate that the proposed system performed better than humans by almost 10%.

An adaptive predictive control based on a quasi-ARX neural network model

Abstract: A quasi-ARX (quasi-linear ARX) neural network (QARXNN) model is able to demonstrate its ability for identification and prediction highly nonlinear system. The model is simplified by a linear correlation between the input vector and its nonlinear coefficients. The coefficients are used to parameterize the input vector performed by an embedded system called as state dependent parameter estimation (SDPE), which is executed by multi layer parceptron neural network (MLPNN). SDPE consists of the linear and nonlinear parts. The controller law is derived via SDPE of the linear and nonlinear parts through switching mechanism. The dynamic tracking controller error is derived then the stability analysis of the closed-loop controller is performed based Lyapunov theorem. Linear based adaptive robust control and nonlinear based adaptive robust control is performed with the switching of the linear and nonlinear parts parameters based Lyapunov theorem to guarantee bounded and convergence error.

Task oriented area partitioning and allocation for optimal operation of multiple industrial robots in unstructured environments

Abstract: When multiple industrial robots are deployed in field applications such as grit blasting and spray painting of steel bridges, the environments are unstructured for robot operation and the robot positions may not be arranged accurately. Coordination of these multiple robots to maximize productivity through area partitioning and allocation is crucial. This paper presents a novel approach to area partitioning and allocation by utilizing multiobjective optimization and voronoi partitioning. Multiobjective optimization is used to minimize: (1) completion time, (2) proximity of the allocated area to the robot, and (3) the torque experienced by each joint of the robot during task execution. Seed points of the voronoi graph for voronoi partitioning are designed to be the design variables of the multiobjective optimization algorithm. Results of three different simulation scenarios are presented to demonstrate the effectiveness of the proposed approach and the advantage of incorporating robots' torque capacity.

Improved analysis of a multirobot graph exploration strategy

Abstract: In this paper, we present an algorithm for exploring an unknown graph with opaque edges by multiple robots. We show that this algorithm is near optimal on graphs with n vertices and superlinear number of edges (i.e., ω(n) edges), and give an adversarial construction to show that the algorithm does not perform well on cyclic graphs with O(n) edges.

Artificial neural network based automatic ship berthing combining PD controlled side thrusters — A combined controller for final approaching to berth

Abstract: Manoeuvring ship during berthing has always required vast experience, skill and knowledge to provide desired necessary actions. Presence of environmental disturbances as well as decreased manoeuvrability in low speed often makes the whole procedure so sophisticated that even slight mistake may results catastrophic disaster. By knowing the fact that Artificial Neural Network (ANN) has the ability to replicate human brains and good enough for controlling such multi-input multi-out nonlinear system, at the beginning of this research, consistent teaching data are created using Non Linear Programing (NPL) method and a new concept named ‘virtual window’ is introduced. Later on, considering gust wind disturbances, two separate multilayer feed forward networks are trained using back propagation technique for command rudder and propeller revolution output. After being successful in simulation works, real time berthing experiments are carried out for Esso Osaka 3-m model where the ship is planned to successfully stop within a distance of 1.5L from actual pier to ensure safety. Finally, as a current status, PD controlled side thrusters are included in order to shake hand with current controller to align the ship with pier considering wind up to 1.5 m/s for model ship.

Image segmentation based on histogram of depth and an application in driver distraction detection

Abstract: This study proposes an approach to segment human object from a depth image based on histogram of depth values. The region of interest is first extracted based on a predefined threshold for histogram regions. A region growing process is then employed to separate multiple human bodies with the same depth interval. Our contribution is the identification of an adaptive growth threshold based on the detected histogram region. To demonstrate the effectiveness of the proposed method, an application in driver distraction detection was introduced. After successfully extracting the driver's position inside the car, we came up with a simple solution to track the driver motion. With the analysis of the difference between initial and current frame, a change of cluster position or depth value in the interested region, which cross the preset threshold, is considered as a distracted activity. The experiment results demonstrated the success of the algorithm in detecting typical distracted driving activities such as using phone for calling or texting, adjusting internal devices and drinking in real time.

An event-triggered based robust control of robot manipulator

Abstract: This paper proposes a framework to design an event-triggered based robust control law for nonlinear uncertain robot manipulator. Load variations and unmodeled system dynamics of manipulator are the primary sources of both system and input uncertainties. A static event-triggering rule is employed to realize the proposed robust control law. Derivation of static event-triggering rule with a positive inter-event time and corresponding stability criteria for uncertain manipulator dynamics are the key contribution of this paper. Validation of proposed control technique is carried out numerically on a two-link SCARA type robot manipulator. Simulation results show that measurement error norm is always bounded by the state dependent threshold and also ensures that asymptotic convergence of manipulator states in the presence of both system and input uncertainty.