We present a novel interval-based visual-inertial LiDAR SLAM (i-VIL SLAM) method that solely assumes sensor errors to be bounded and propagates the error from the input sources to the estimated map and trajectory using interval analysis. The method allows us to restrict the solution set of the robot poses and the position of the landmarks to the set that is consistent with the measurements. If the error limits are not violated, it is guaranteed that the estimated set contains the true solution. The accumulation of the uncertainty is stabilized by anchoring poses derived from GNSS/INS data. Furthermore, for the first time we compare confidence ellipses determined by a classical SLAM graph optimization approach with the interval estimates of the robot poses provided by our method. In this work, we experimentally show that the marginal co-variances computed by the classical SLAM graph optimization are too overconfident and underestimate the uncertainty of the poses. While the 99.9 %-ellipsoids derived from the marginal covariances of the poses only enclose less than 64 % of the ground truth in the worst case, our method provides interval bounds for the pose parameters that enclose the ground truth for more than 96 % of all frames.

To fuse information from a 3D Light Detection and Ranging (LiDAR) sensor and a camera, the extrinsic transformation between the sensor coordinate systems needs to be known. Therefore, an extrinsic calibration must be performed, which is usually based on features extracted from sensor data. Naturally, sensor errors can affect the feature extraction process, and thus distort the calibration result. Unlike previous works, which do not consider the uncertainties of the sensors, we propose a set-membership approach that takes all sensor errors into account. Since the actual error distribution of off- the-shelf sensors is often unknown, we assume to only know bounds (or intervals) enclosing the sensor errors and accordingly introduce novel error models for both sensors. Next, we introduce interval-based approaches to extract corresponding features from images and point clouds. Due to the unknown but bounded sensor errors, we cannot determine the features exactly, but compute intervals guaranteed to enclose them. Subsequently, these feature intervals enable us to formulate a Constraint Satisfaction Problem (CSP). Finally, the CSP is solved to find a set of solutions that is guaranteed to contain the true solution and simultaneously reflects the accuracy of the calibration. Experiments using simulated and real data validate our approach and show its advantages over existing methods.

Localization and mapping are essential tasks in mobile robotics. A purely odometry-based pose estimation accumulates errors caused by factors such as unequal wheel-diameters or wheel-slippage and is, therefore, inaccurate. The resulting error can be corrected by recognizing previously observed places and constraining the robots' relative poses. A common challenge of such place recognition methods arises from the incorrect association of different places, which can corrupt the resulting pose correction and, consequently, any map update. Therefore, we propose a novel place recognition method that enables the selection of a correct place recognition hypothesis from a set of possible matches.In this work, we make use of the fact that multiple robots' trajectories run parallel in many scenarios, like in warehouse corridors or on public roads. In such situations, place recognition can be regarded as a sequential task along the robots' trajectories. We study the problem of repetitive, periodic and indistinguishable landmarks for place recognition on regularly-spaced guideposts on German rural roads. By interpreting place recognition hypotheses as nodes of a k-partite graph, we introduce our novel selection method, the Hierarchic Single Cluster Graph Partitioning, that enables the robust selection of the correct hypotheses by finding the optimal path within the graph. The selected place recognition information is used to build a map incorporating multiple observations acquired from multiple vehicles.

To properly fuse IMU and camera information for robotics applications, the relative timestamp offset between both sensors' data streams has to be considered. However, finding the exact timestamp offset is often impossible. Thus, it is necessary to additionally consider the offset's uncertainty if we want to produce reliable results. In order to find the offset and its uncertainty, we determine orientation estimates from IMU and camera under interval uncertainty. Subsequently, these intervals are used as a common representation for our bounded-error approach that finds an interval enclosing the true offset while also modeling the uncertainty. Calibration data can be acquired in a few seconds using a simple setup of IMU, camera and camera target. Results using both simulated and real data demonstrate that we are able to determine the offset to an accuracy of 20 ms with a computation time that is suitable for future online applications. Here, our approach could be used to monitor the timestamp offset in a guaranteed way. Additionally, our method can be adapted to determine an interval for the rotation between both sensors. While this increases the computation time drastically, it also enhances the accuracy of the timestamp offset to less than 10 ms.

LiDAR scanners are commonly used for mapping and localization with mobile robots. But, they cannot see through occlusions, as it occurs in harsh environments, containing smoke, fog or dust. Radar scanners can overcome this problem, but they have lower range and angular resolution, and cannot represent an environment in the same quality. In the following article, we present the integration of fused LiDAR and radar data into a SLAM cycle and continue our work from [1], where we presented first results regarding a feature based and a scan matching-based approach for SLAM in environments with changing visibility using LiDAR and radar sensors. New content in this article, the data fusion takes place on scan level as well as on map level and aims to result in an optimum map quality considering the visibility situation. Additionally, we collected more data during an indoor experiment involving real fog (see Fig. 1). Besides the structure of the environment, we can model aerosol concentration with fused LiDAR and Radar data in parallel to the mapping process with a finite difference model without involving a smoke or gas sensor. Overall, our method allows the modeling of the structure of an environment including dynamic distribution of aerosol concentration.

S.30-36

Many mobile robots nowadays use thermal imaging cameras (TICs) in order to enhance the environment model that is created during exploration tasks. In conventional thermography, thermal images always have to be carefully revised by human operators, which is not practicable in autonomous applications. Unknown surface emissivities are the main source of misinterpretations in thermal images. In this work, we present two methods dealing with these misinterpretations by exploiting the TIC’s changing point of view. While the first approach classifies the regarded material in order to estimate improved surface temperature values, the second one is capable of detecting and removing thermal reflections. The spatial relationship between the thermal images and the regarded surface is made by using a rigidly mounted sensor stack consisting of a TIC and a 3D laser range finder, whose extrinsic calibration is described. During evaluation, we demonstrate the functionality of both approaches.

Technologies for fully automated driving are currently a hot topic for both industry and academia. To achieve a full automation, self-driving cars need a precise localization module. Most of existing localization approaches consist of two main steps: map generation and actual localization that uses the map obtained at the first step. The localization quality of a system directly depends on the capabilities of its sensors, the quality of a map, as well as correctness and effectiveness with which a localization method uses new observations. Therefore, to provide the best results, such systems are equipped with ever-increasing number of sensors. However, extraction of relevant information from sensor data is still challenging. This paper focuses on two localization components: landmark tracking and fusion. We consider landmarks corresponding to poles and road surface markings. They are extracted from the data provided by four fisheye cameras placed around a car and a front lidar. Detection of landmarks depends on characteristics of a sensor: its quality, delay (time), and output rate (frequency). Therefore, we have developed a tracking and fusion module based on the Sequential Probability Ratio Test which is used for both map generation and localization steps. This module was evaluated in a number of driving tests and the results showed high map quality and low localization error.

This paper will introduce two radar-based scanning-methods and evaluate their application in robotic mapping. Both approaches base upon a rotary joint, but with a fundamentally different angle estimation method to estimate object locations inside the scanning area. The first part of this paper describes the relevant theory behind both techniques and presents our considerations on erroneous influences. The focus of the second part of this paper is laying on experiments. We discuss the results of our experiments and take a look on the usability of both methods for occupancy grid mapping.

This paper describes two methods for determining the extrinsic calibration parameters of a projector with respect to the robot hand. One of them simultaneously solves the transformation between a camera with respect to the robot base. Self-calibration means that no sort of calibration rig like a chessboard is needed. Since the projector has no exteroceptive capabilities, a camera is placed in the environment or rigidly attached to the robot base to detect the projected pattern. At different robot configurations correspondences between the camera and projector are established to recover the transformation between them up to an unknown scale factor. The common known formulations AX = XB and AX = ZB can be arranged in a linear form with respect to the unknown extrinsic parameters and scale factors, and solved in least square sense.

A widely used technique for constructing two dimensional maps employing range sensors is occupancy grid mapping assuming normal distributed sensor errors. An alternative to the grid map model with its fixed grid cell size are variable resolution grid maps, e.g. quadtrees. In this paper, the authors propose an approach for building occupancy quadtree maps assuming unknown but bounded sensor errors. Therefore, they examine possible types of sensor uncertainty when using laser rangefinders. They show that the majority of possible types of sensor errors can be covered much better by bounded error models than by probabilistic models. Hence, a novel inverse sensor model has been developed that incorporates measurement and pose uncertainty in a mathematical straightforward way using interval analysis. With this model, an approach for incrementally building occupancy quadtree maps is proposed. A first real world experiment has shown the applicability of the approach. Moreover, the authors compare the map with its probabilistic grid map counterpart. The bounded error quadtree has proved to be conservative but more reliable than the conventional probabilistic grid map.

We present a method to aggregate and classify the 2D outline of arbitrary shaped objects from 3D-range data. As a mobile robot only captures a limited part of an object in a single measurement, the raw data of several measurements are combined to build a data set as comprehensive as possible. Doing so the classification uses all previous measurements. As the classification uses local curvature based features it can classify partial and complete shapes. The approach combines a spline approximation method with a scalable method for shape registration, combination and classification. Experimental results illustrate the ability of combining and classifying real world objects on a mobile robot.

The OMAC Packaging Machine Language Working Group has published a recommendation of a uniform state model for managing the different operation states and the user actions with packaging machinery. This proposed state model with well-defined states and transitions shall overcome the integration problems of the past and present concerning the different semantics of software and user interfaces of packaging machinery. This is addressed by providing an abstract view to the functionality of a packaging machine and its controller implementation. This shall enable the simple integration of machines from different vendors into packaging lines. In a comparable way, the IEC 61499 offers an abstract architectural model for designing distributed controllers based on function blocks. This paper uses the language features offered by function blocks according to IEC 61499 in order to provide a platform-independent controller implementation of the proposed OMAC state models. Two design alternatives will be presented. The designs are work in progress and are subject to further refinement in the future.

This paper aims at the systematic determination of faulty and problematic execution control charts (ECC) of IEC 61499 basic function blocks. The liveliness of a function block and the potential state transitions fundamentally depend on the execution semantics of function blocks. Many work has been performed in order to clarify the execution semantics, mainly by reducing the diversity of implementation alternatives.

This paper presents a localization system for mobile robots that are able to navigate autonomously in industrial environments like factory and exhibition halls. Previous approaches show that ceiling structures like beams, pipes and lighting installation are well-suited for self-localization in large halls. With this paper we describe how to use ceiling structures, measured with a 3D laser range sensor, as natural landmarks. Due to the novel relative representation of 3D ceiling features this method can be used in a wide range of industrial halls with sloped, stepped and even curved ceiling structures. In addition to this new algorithm the paper presents experimental results with an autonomous fork lift working without the need for artificial landmarks or extra infrastructure.

This paper introduces a localization based on GPS and laser measurements for urban and non-urban outdoor environments. In this approach, the GPS pose is Kalman filtered using wheel odometry and inertial data and tightly integrated into a Monte Carlo localization based on 3D laser range data and a line feature reference map. By applying to this kind of sensor fusion, global localization as well as precise position tracking in close distance to buildings is enabled where only poor GPS observations are available. Following the description of the localization system, real world experiments demonstrate the functionality of the presented approach.

In the past many solutions for simultaneous localization and mapping (SLAM) have been presented. Recently these solutions have been extended to map large environments with six degrees of freedom (DoF) poses. To demonstrate the capabilities of these SLAM algorithms it is common practice to present the generated maps and successful loop closing. Unfortunately there is often no objective performance metric that allows to compare different approaches. This fact is attributed to the lack of ground truth data. For this reason we present a novel method that is able to generate this ground truth data based on reference maps. Further on, the resulting reference path is used to measure the absolute performance of different 6D SLAM algorithms building a large urban outdoor map.

This paper identifies different aspects of modeling techniques used for the design and implementation of industrial distributed real-time control applications. These cover technical means for structuring, modeling and expressing timing, and representing concurrency, as well as graphic expressiveness and intuitive comprehension. These criteria are applied to consider one by one IEC 61499 function blocks, UML statecharts and Petri net based CNet. The paper shows how the addressed features are expressed using each modeling technique. Finally, the assets and drawback of the considered description means are compared

This paper presents two laser scanner based approaches to locate and pick-up pallets with the aim of automating forklift trucks. In contrast to camera based systems our approaches are independent of luminance conditions which can be highly variable in industrial environments. Whereas one approach uses pallets modified with adhesive reflectors which enables very flexible handling of pallets, the second approach detects pallets based on geometric characteristics without any modifications to the environment. Experimental results are also reported including a run of an autonomous fork-lift truck presented at the logistics fair CeMAT 2005 in Hannover.

This video introduces a new perceptual model for Monte Carlo Localization (MCL). In our approach a 3D laser scanner is used to observe the ceiling. The MCL matches ceiling structures like beams, columns, air condition and lightning installation against a world model containing line and point features. Thus the localization is not effected by clutter or any kind of dynamic obstacles on the ground level. Different experiments show that our system can be used to localize robustly in factory buildings and halls. Experimental results include a 5 day run of an autonomous fork lift truck at a logistics fair in Hannover.

This paper introduces a new perceptual model for Monte Carlo localization (MCL). In our approach a 3D laser scanner is used to observe the ceiling. The MCL matches ceiling structures like beams, columns, air condition and lightning installation against a world model containing line and point features. Thus the localization is not effected by clutter or any kind of dynamic obstacles on the ground level. Different experiments show that our system can be used to localize robustly in factory buildings and halls. Experimental results include a 5 day run of an autonomous fork lift truck on a logistics fair in Hannover

In this paper, the open source project RTnet is presented. RTnet provides a customisable and extensible framework for hard real-time communication over Ethernet and other transport media. The paper describes architecture, core components, and protocols of RTnet. FireWire is introduced as a powerful alternative to Ethernet, and its integration into RTnet is presented. Moreover, an overview of available and future application protocols for this networking framework is given.

A component based modeling language with the component interfaces derived from the elements of the Petri net theory called CNet is presented in this paper. In contrast to IEC 61499 function blocks that define the flow of events and data separately, CNet models them combined as colored tokens. The event-discrete behavior of CNet components is described by a special class of colored Petri nets with timed arcs and sharpened semantics in order to allow automatic generation of possibly concurrent Java code. In this paper, the state space of CNet components is analyzed and a novel event-discrete "handler-based" execution model that implicitly models an automaton is introduced.

This paper presents a combination of a 3D laser sensor and a line-base SLAM algorithm which together produce 2D line maps of highly cluttered indoor environments. The key of the described method is the replacement of commonly used 2D laser range sensors by 3D perception. A straightforward algorithm extracts a virtual 2D scan that also contains partially occluded walls. These virtual scans are used as input for SLAM using line segments as features. The paper presents the used algorithms and experimental results that were made in a former industrial bakery. The focus lies on scenes that are known to be problematic for pure 2D systems. The results demonstrate that mapping indoor environments can be made robust with respect to both, poor odometry and clutter.

This paper presents a navigation system for mobile service robots working in urban environments. The system combines a 3D laser sensor with 2D algorithms for path planning and simultaneous localization and mapping (SLAM). In contrast to other map representations the colored 2D map, first presented in this paper, is able to hold information adapted to both localization and path planning. The functionality of our approach is demonstrated by an experiment for online navigation with 3D data.

This paper presents a security concept for real-time operating systems based on strong access control mechanism. For that purpose, the system is divided into a minor real-time and a major non-real-time part. The general and formal domain and type enforcement model is applied to both divisions. This paper offers an outlook on emerging applications scenarios and an examination of the effect of the proposed concept under expected attacks. In addition, requirements and practical aspects of an implementation are discussed.

We present the concepts of our MoRob (modular educational robotic toolbox) project, which aims to provide a robot platform for university teaching and research. Characteristics are scalability, modularity, flexibility and ease of use. Within the context of this project, robots are used in student projects. Lego Mindstorms has been successfully applied at an introductory level, and insights from this are incorporated into a new masters degree in systems design, using other, higher-level platforms. With these, MSc. students will have the tools to tackle design tasks which are much more complex and sophisticated than the tasks that can be achieved with Lego. This also implies that, applying our concepts, a much broader range of engineering concepts can be taught. Our new system will be applicable as a vehicle for experimentation in a variety of courses. Therefore, a standardized teaching environment will emerge which will be easy to use in different applications.

Robot navigation in poorly structured and uneven outdoor environments is an unsolved problem. Thus we present a SLAM (simultaneous localization and mapping) approach that is based on "leveled range scans ". The combines 3D perception with 2D localization and mapping. In this way established path planning and 2D navigation algorithms can be used in uneven terrain without the computational costs of full three dimensional modeling. The paper describes the processing steps data acquisition, obstacle segmentation, generation of leveled range scans and SLAM with these scans. Additionally, the paper shows experimental results in man-made outdoor environments as they are typical for service robots used by civilians.