Gregor Novak

Throwing Objects -- A bio-inspired Approach for the Transportation of Parts

In this paper a new way of transporting parts in production systems is described. Based on the fact that in nature the highest speeds are reached at flight, a bio-inspired technical approach for the fast transportation of objects is derived. Objects shall be thrown by a throwing device and are caught again by a catching device. Such a transport approach requires besides proper working throwing and catching devices a high precision object detection mechanism. For object detection a 3D photonic mixer device (PMD) sensor is proposed. By assembling of this sensor on the catching device, it can detect the flying objects on their trajectories like a raptor which is hunting a prey. The technical feasibility of this concept is investigated and possible applications are described.

Line-based landmark recognition for self-localization of soccer robots

This paperl describes visual landmark extraction for self localization of an auitonomolus mobile robot in at wellknown dynamic environment. The gradient based Hough transform provides the strongest groupings ofcollineair pixels having roughly the same edge orientcation. Grouips of pixels are then processed to calcuilate length and end points of line segments, which together with the length anid direction of the normcal completes the description of the line segmnents. This is followed by classification offield markinigs which are bright lines and aircs on a dark background forming double edges of opposite gradient. Corners, junctions and line intersections are determined by the interpretation of detected line segments. Simulation results illustrate the performance of the method u(sing both real and sYnthetic

An introduction to a vision system used for a MiroSOT robot soccer system

In a MiroSOT robot soccer system a global vision system mounted above the playground is the only sensor capable of providing information about the situation on the playground. It is the basis for the decision making and path planning process. Therefore the accuracy and speed of the vision system is the major key factor for the success of a team. This paper gives an overview of the common systems and algorithms

NEURAL NETWORK PREDICTIVE TRAJECTORY TRACKING OF AN AUTONOMOUS TWO-WHEELED MOBILE ROBOT

Abstract As a prerequisite for precise trajectory tracking of a two-wheeled mobile robot, accurate control of the velocity and the curvature along a predefined trajectory is vital. After offline training, a neural network is used for nonlinear predictive control. To make the system more robust against modeling inaccuracies and other disturbance influences, the control error is integrated and used to adjust the control variables calculated by the predictor.

Simple stereo vision system for real-time object recognition for an autonomous mobile robot

Intelligent sensors for mobile robots play an important role in many technical applications. In this paper, a real-time stereo object recognition system for a tiny mobile robot is presented, capable of detecting objects in real-time at a frame rate of up to 60frames/s. In order to get an all around view the stereo camera system is mounted on a pivoted head. We propose an object recognition algorithm that is optimized for the detection of well known objects on deeply embedded systems. The detection algorithm is based on a combination of edge and color detection and uses a fixed model for each object to be recognized. The stereo algorithm is a simple matching of two independently performed algorithms for each camera. Results of the ball recognition application show that its relative coordinates are found within <10ms

Optimized Control Methods for Capturing Flying Objects with a Cartesian Robot

Throwing or shooting is a new approach for the transportation of objects within production systems. Since gantry robots are often applied to load and unload machines, in a research project a Cartesian robot was used for capturing flying objects. A camera system is measuring the objects positions during a throw in subsequent periods of time. Based on these measurements it can predict the capturing point with an increasing accuracy. So it can direct the robot during a throw to the capturing point also with an increasing accuracy. In this paper control methods are proposed which allow in such an application fast and smooth motions of the robot to the final capturing point. For the evaluation of the control methods two scenarios, a simplified and a realistic one, are defined.

Roby-go, a prototype for several MiroSOT soccer playing robots

For a competitive robot soccer team a perfect working robot is a fundamental part. The developed mobile mini robot is a two wheeled differentially driven (2WDD) robot , distinguished by its simple, compact and modular construction. Due to the open architecture of the robot it is possible to use it not only as a soccer player, in the category MiroSOT, but also as a mobile platform in a number of several other tasks, for example as an independent testing bed for other multi agent systems

Layered Multi-Agent Reactive-Behavior Learning in a Robotic Soccer

Abstract Robotic soccer is an interesting test bench for the field of self-organizing and cooperating multi-agent systems. This paper deals with learning of two basic low-level behaviors that will enable the robotic player to participate further in higher-level collaborative and adversarial learning situations. First, a ball interception and obstacle avoidance behavior is learned. Then the acquired skills are incorporated into a next higher-level multi-agent learning scenario, namely the shooting ball behavior. The proposed control scheme for these behaviors consists of a trajectory generator with a layered structure, which supplies data to a trajectory-tracking controller.

Embedded Real-Time Ball Detection Unit for the YABIRO Biped Robot

Estimation of objects in a 3D space is a fundamental problem in computer vision and robotics. This paper describes an algorithm and its implementation for a vision module as a sensor of a biped robot (YABIRO). The embedded vision sensor is able to estimate the position of objects like spheres in 3D space. Objects are defined with their size and color in a model. The vision sensor detects the positions or at least the directions to the objects and stores them in a history. The algorithm includes a new voting system for detected objects, based on how trustable the detection was, and a new edge filter to terminate edges on the circle border for the circle detection. The systems frame rate depends on the area of interest and lies between 5 Hz and 20 Hz. With a mechanical size of 36times32mm it is smaller than a matchbox

A Real-Time Image Recognition System for Tiny Autonomous Mobile Robots

Intelligent sensors for mobile robots play an important role in many technical applications. In this paper a real-time image recognition system for a tiny autonomous mobile robot is presented, capable of detecting objects in real-time at a frame rate of up to 60 frames/s. The image recognition module has very low power consumption of less than 250 mW and fits into a package of only 35 × 35 mm including a CMOS camera and a low power, high performance signal processor. We propose an object recognition algorithm that is optimized for deeply embedded systems used in energy and performance constrained devices. The algorithm is based on a combination of edge and color detection and uses a fixed model for each object to be recognized. Results of the ball recognition application show that its relative polar coordinates are found within 11 ms.