This paper presents a detailed characterization of the Hokuyo URG-04LX 2D laser range finder. While the sensor specifications only provide a rough estimation of the sensor accuracy, the present work analyzes issues such as time drift effects and dependencies on distance, target properties (color, brightness and material) as well as incidence angle. Since the sensor is intended to be used for measurements of a tubelike environment on an inspection robot, the characterization is extended by investigating the influence of the sensor orientation and dependency on lighting conditions. The sensor characteristics are compared to those of the Sick LMS 200 which is commonly used in robotic applications when size and weight are not critical constraints. The results show that the sensor accuracy is strongly depending on the target properties (color, brightness, material) and that it is consequently difficult to establish a calibration model. The paper also identifies cases for which the sensor returns faulty measurements, mainly when the surface has low reflectivity (dark surfaces, foam) or for high incidence angles on shiny surfaces. On the other hand, the repeatability of the sensor seems to be competitive with the LMS 200.

In this paper describes a pair of wall climbing robots that use magnetic wheels for adhesion. They are designed for inspecting the interior surfaces of gas-tanks in oversea ships. Environments like this were impossible to access by previous climbing robots, as they are made out of very thin sheet metal that cannot support a high robot mass and contain several types of difficult obstacles. In order to master these challenges, the system described in this work uses two robots in a "mother-child"-structure: the smaller robot is built very lightweight and simple, without the ability to steer or climb vertically. It just moves horizontally and uses ridges in the gas tank surface as guidance rails. In order to pass from one ridge to the next one, the smaller robot docks to the bigger "mother"-robot. This robot always stays in an area where the sheet metal is thicker and never enters the extremely fragile surfaces. Thus, its mass is not critical for the structural stability of the environment and the robot can hence be built big enough to be able to climb in all directions of gravity and to pass difficult obstacles along its path. In the present paper, the basic concept, the mechanical design of all important components and the proposed control strategy are described briefly, followed by test results of the most critical components. An outlook to similar applications to which the basic idea of this work can be successfully transferred concludes this article.

This paper describes a compact robot with two magnetic wheels in a bicycle arrangement, which is intended for inspecting the inner casing of pipes with complex shaped structures. The locomotion concept is based on an adapted magnetic wheel unit integrating two lateral lever arms. These arms allow for slightly lifting off the wheel in order to locally decrease the magnetic force, as well as laterally stabilizing the wheel unit. The robot has the main advantage to be compact and mechanically simple. It features 5 active degrees of freedom: 2 driven wheels each equipped with an active lifter-stabilizer and 1 steering unit. This paper also presents the design and implementation of a prototype robot and its high mobility is shown. It is able to pass 90deg convex and concave obstacles with any inclination regarding the gravity. Finally, it only requires limited space to maneuver, since turning on spot around the rear wheel is possible.

The pheromone trail laying and trail following behaviors of ants have proved to be an efficient mechanism to optimize path selection in natural as well as in artificial networks. Despite this efficiency, this mechanism is under-used in collective robotics because of the chemical nature of pheromones. In this paper we present a new experimental setup which allows to investigate with real robots the properties of a robotics systems using such behaviors. To validate our setup, we present the results of an experiment in which a group of 5 robots has to select between two identical alternatives a path linking two different areas. Moreover, a set of computer simulations provides a more complete exploration of the properties of this system. At last, experimental and simulation results lead us to interesting prediction that will be testable in our setup.