Showing papers by "Ivan Petrović published in 2007"

Dynamic window based approach to mobile robot motion control in the presence of moving obstacles

Abstract: This paper presents a motion control method for mobile robots in partially unknown environments populated with moving obstacles. The proposed method is based on the integration of focused D* search algorithm and dynamic window local obstacle avoidance algorithm with some adaptations that provide efficient avoidance of moving obstacles. The moving obstacles are modelled as moving cells in the occupancy grid map and their motion is predicted by applying a procedure similar to the dynamic window approach. The collision points of the robot predicted trajectory and moving cells predicted trajectories form the new active obstacles in the environment, which should be avoided. The algorithms are implemented and verified using a Pioneer 3DX mobile robot equipped with laser range finder.

Hybrid Theory-Based Time-Optimal Control of an Electronic Throttle

Abstract: An electronic throttle is a dc-motor-driven valve that regulates air inflow into the combustion system of the engine. The throttle control system should ensure fast and accurate reference tracking of the valve plate angle while preventing excessive wear of the throttle components by constraining physical variables to their normal-operation domains. These high-quality control demands are hard to accomplish since the plant is burdened with strong nonlinear effects of friction and limp-home nonlinearity. In this paper, the controller synthesis is performed in discrete time by solving a constrained time-optimal control problem for the piecewise affine (PWA) model of the throttle. To that end, a procedure is proposed to model friction in a discrete-time PWA form that is suitable both for simulation and controller design purposes. The control action computation can, in general, be restated as a mixed-integer program. However, due to the small sampling time, solving such a program online (in a receding horizon fashion) would be very prohibitive. This issue is resolved by applying recent theoretical results that enable offline precomputation of the state-feedback optimal control law in the form of a lookup table. The technique employs invariant set computation and reachability analysis. The experimental results on a real electronic throttle are reported and compared with a tuned PID controller that comprises a feedforward compensation of the process nonlinearities. The designed time-optimal controller achieves considerably faster transient, while preserving other important performance measures, like the absence of overshoot and static accuracy within the measurement resolution

Mobile robot self-localization in complex indoor environments using monocular vision and 3D model

Abstract: In this paper, we consider the problem of mobile robot pose estimation using only visual information from a single camera and odometry readings. A focus is on building complex environmental models, fast online rendering and real-time complex and noisy image segmentation. The 3D model of the robot's environment is built using a professional freeware computer graphics tool named Blender and pre-stored in the memory of the robot's on-board computer. Estimation of the mobile robot pose as a stochastic variable is done by correspondences of image lines, extracted using Random Window Randomized Hough Transform line detection algorithm, and model lines, predicted using odometry readings and 3D environment model. The camera model and ray tracing algorithm are also described. Developed algorithms are experimentally tested using a Pioneer 2DX mobile robot.

Simple off-line Odometry Calibration of Differential Drive Mobile Robots

Abstract: Odometry is a widely used method for estimation of the momentary pose of a mobile robot with respect to its starting pose. It provides easily accessible real-time pose information between periodic absolute pose measurements or between pose corrections using additional sensors. Odometric localization accumulates errors in an unbounded fashion with quadratic increase of error variance with traversed distance. Odometry errors consist of systematic and non-systematic parts. While non-systematic errors cannot be predicted and therefore compensated, systematic errors can be compensated by means of calibration, which can be off-line and on-line. This paper describes an approach to off-line odometry calibration for differential drive mobile robots, which is based on simple experiments combined with optimization methods. Two variants of the proposed calibration method are examined: one with 3 calibration parameters and other one with 2 calibration parameters. Experimental results obtained using Pioneer 2 DX differential drive mobile robot show that both methods significantly increase accuracy of the pose estimation. Due to simplicity and suitability for later on-line adaptation, calibration method with two parameters is preferred.

Damping of Wind Turbine Tower Oscillations through Rotor Speed Control

Abstract: To enable wind turbines to produce power under great variety of wind conditions a sophisticated control system is needed. Wind turbine system is highly nonlinear and its dynamic changes rapidly with the change of wind speed. Besides this, wind turbine mechanical structure is very flexible due to its great height and tends to oscillate. All this makes the design of wind turbine control system very demanding task. In his paper three controller design methods are compared. It is shown how with adequate controller design it becomes possible to control the wind turbine under various operating conditions reducing the structural oscillations at the same time

Design of a Wind Turbine Pitch Controller for Loads and Fatigue Reduction

Abstract: Wind turbines are constantly growing in size and power output level. This growth results in significant increase of structural loads and fatigue that have become limiting factors in wind turbine design. It introduces a need to control and limit structural loads and fatigue during wind turbine operation. This can be done by careful control system design. In this paper we focus on variable speed pitch controlled wind turbine that is today an industrial standard. A design procedure for three types of pitch controller is described and their performance is tested by simulations. Like first classic PID controller is described that aims only at achieving satisfactory rotor speed control. This classic pitch controller is then augmented and reduction of tower oscillations is used as additional demand during controller design. Two controller concepts used for this purpose are compared – SISO and full state feedback controller. Both controllers are designed by pole placement method that assures closed loop system to behave in desired manner. Desired closed loop system behavior is carefully chosen in such a way that controlled system behaves as if its structural damping has increased. It results in reduced tower motion what leads to loads and fatigue reduction. All controllers described in this paper are linear so an adaptation algorithm is needed to account for nonlinear nature of wind turbine. For this purpose we propose the use of Takagi-Sugeno fuzzy model that calculates process model output as a combination of "local" linear models identified in chosen operating points.

Teleoperation Of Collaborative Mobile Robots With Force Feedback Over Internet

Abstract: A teleoperation system has been developed that enables two human operators to safely control two collaborative mobile robots in unknown and dynamic environments from any two PCs connected to the Internet by installing developed client program on them and by using simple force feedback joysticks. On the graphical user interfaces, the operators receive images forwarded by the cameras mounted on the robots, and on the the joysticks they feel forces forwarded by developed obstacle prevention algorithm based on the dynamic window approach. The amount and direction of the forces they feel on their hands depend on the distance and direction to the robot’ s closest obstacle, which can also be the collaborating robot. To overcome the instability caused by the unknown and varying time delay an event-based teleoperation method is employed to synchronize actions of each robot with commands from its operator. Through experimental investigation it is confirmed that developed teleoperation system enables the operators to successfully accomplish collaborative tasks in complex environments.

Estimation and Prediction in Load frequency Control

Abstract: Many controllers designed for load-frequency control (LFC) in power system are based on the knowledge of system state, which is generally not measured. This paper presents a method for estimation of disturbance and system state in LFC. In power systems, measurement signals and control signals are transmitted over long distances between controller and power units what introduces large time delays in the system. Therefore, a state prediction method to overcome the effects of the time delay is presented here. A power system consisting of three interconnected areas represented by thermal power units is simulated. Simulation results of state and disturbance estimation and also state prediction are shown here.

Platform based development of embedded systems for traction and power engineering applications - experiences and challenges

Abstract: The paper deals with development of platform based, embedded, hard real-time systems for traction and power engineering applications. Such systems are intended to have lifecyle of more than 20 years. Several cases of successful projects based on two typical platform architectures are presented. Both hardware and software platforms and components are described along with some topics that can help to identify advantages and disadvantages of such an attitude. Particular attention is given to the problem usually not taken into account-component obsolescence.

Correlation Based Approach to Mobile Robot Pose Tracking in Unknown Environments

Abstract: The usage of mobile robots in service sector is increasing in the past decade. There are already mobile robots as night guards, tourist guides, transportation units, etc. In order to efficiently perform its task, a mobile robot has to know precisely its current pose. Also to simplify the commissioning of a mobile robot system various algorithms for map building without a human intervention have been developed. In both cases a reliable and robust pose tracking module that can compensate the influence of systematic and non-systematic localization errors in real time is crucial. This paper presents a mobile robot pose tracking approach based on subsequent laser scan readings processed by usage of histograms and correlation comparison. Results obtained with a differential drive mobile robot are given to demonstrate the capabilities of the proposed approach.

Dynamic window based force reflection for safe teleoperation of a mobile robot via internet

Abstract: The paper presents an Internet-based teleoperation system that enables a human operator to safely control a mobile robot in unknown and dynamic environments. The operator controls the robot using a joystick and a graphical user interface which displays images forwarded from the camera mounted on the robot. A sonar ring on the robot circumference is used to measure obstacle range information, and a dynamic window algorithm is used to convert that information into a force, which is than reflected to the operator's hand via joystick, providing additional haptic information about obstacles in the robot vicinity. To overcome the instability caused by the unknown and varying time delay an event-based teleoperation system is employed to synchronize actions of the operator and teleoperated mobile robot. The experiments with the Pioneer 2DX mobile robot verified effectiveness of the developed system.

Global Vision Based Tracking of Multiple Mobile Robots in Subpixel Precision

Abstract: This paper presents a new global vision system for tracking of multiple mobile robots, with special emphasis on following requirements: (i) high measurement precision and accuracy, (ii) robustness to light intensity changes, (iii) ability to track unlimited number of robots, and (iv) high speed and real time operation. In order to fulfill these requirements we use an IEEE 1394 camera with 80 fps, which delivers images in so called Bayer format. We also design a specific robot mark that enables measurement of robot locations directly in Bayer format image with subpixel precision. The developed vision system fulfils all of given requirements, which was verified by carefully performed experiments.

Some issues of microprocessor-based power system stabilizer implementation

Abstract: In this paper, we present our experience gained through implementation of the IEEE PSS2B type of power system stabilizers (PSSs). The paper describes a particular case of implementing this type of PSS in microprocessor-based digital voltage regulator with limited processing resources. The following issues have been elaborated: (i) choice of sampling time and its influence on the performance of the PSS and on the processor utilization, (ii) selection of the lead-lags time constants and their influences on the noise amplification and (iii) scaling associated with the integer arithmetic. The above mentioned problems have been analyzed by simulations and experimentally on a laboratory power plant, and appropriate solutions have been proposed. Finally, the PSS2B has been implemented on real power plants, and a simple method for field verification of tuned PSS2B parameters is explained.

Two approaches to mobile robots simulator design

Abstract: This paper presents the design of a mobile robots simulator which is applicable for robot soccer or for general use. The simulator consists of a group of mobile robots, the ball and includes knowledge of their dynamic behavior modeling, collisions modeling and visualization. Two different approaches to the design of this simulator are given and compared. By the first approach the simulator physics background was completely developed by our team, which enabled us to get a better insight into the problem domain and gave us the possibility to efficiently solve some simulator specifics. First the model of ball and robot motion was derived and then complex approximate collisions models, where the real robot shape is taken into consideration. Some new ideas of collision formulation, realization and real robot shape inclusion are used. By the second approach the simulator was developed using freely available physics engine ODE ? Open Dynamics Engine. This engine already includes physical background for rigid bodies dynamic ; it is up to the user to define mechanical and physical parameters of the objects to simulate e.g. dimensions, masses, friction, joints and the like. The implementation of both simulators are described and a comparisons of their reality description, computational efficiency, effort and knowledge needed to built the simulator, advantages and disadvantages are given.

Mobile Robot Localization using Soft-reduced Hypotheses Tracking

Abstract: Mobile robot localization is the problem of determining the pose (position and orientation) of a mobile robot under complex measurement uncertainties. The Soft-reduced Hypotheses Tracking algorithm introduced here is based on the modified multiple model and exploits a soft gating of the measurements to reduce the computational requirements of the approach. The position part is based on an x- and y-histograms scan matching procedure, where x- and y-histograms are extracted directly from local occupancy grid maps using probability scalar transformation. The orientation part is based on the proposed obstacle vector transformation combined with polar histograms. Proposed algorithms are tested using a Pioneer 2DX mobile robot.