High definition (HD) maps can fail by becoming outdated. To still use them safely for automated driving, they need to be verified or updated, both requiring methods for change detection. We propose two significant improvements for HD map change detection that do not require a highly accurate localization prior as localization quickly fails or cannot be trusted in an outdated map. Given a very coarse localization prior, we group stored or measured map features in spatially and semantically separable feature groups. These feature groups are not only intuitive, like the sequence of leftmost dashed lane markings, but changes are also highly correlated within them. The first contribution improves the way internal consistency of each feature group is assured by using boosted classification trees. Additionally, a mutual evaluation scheme is added for all seemingly unchanged feature groups. Always one feature group is used for localization by feature alignment while each other group's alignment is checked for compatibility. Two voting schemes are presented that allow a more or less sensitive change detection on the level of proposed groups. In contrast to almost all other approaches, our approach allows to use still valid parts of the map for automated driving and to update the changed parts. We evaluate our approach on a previously published map verification dataset [1], showing that the number of undetected map changes can be reduced by up to 31 % compared to state of the art using boosted classification trees, at the same time reducing false positive rates by up to 50 %. The additional mutual evaluation step is able to uncover a whole category of previously undetectable changes and reduces undetected changes by an extra 15 %.

High definition (HD) maps have proven to be a necessary component for safe and comfortable automated driving (AD) [1]. Naïvely verifying HD maps requires an accurate localization prior in order to correctly associate measurements with map data. In periodic environments, such as highways, localization results are often ambiguous - in particular in longitudinal direction. To still be able to verify an HD map, we propose the use of quasi-continuous 1D signals that can be computed without pointwise association. These signals can be chosen to change significantly when the map has changed while they only change rarely or slowly along the road, making them robust against localization errors. A spatio-semantic clustering yields intuitive groups of map features. These groups are then ordered using a robust projection approach, yielding quasi-continuous 1D signals. Such signals can be computed for map and measurement data and their comparison allows detecting road changes. The purposeful design of the signals and their computation only requires lane-level lateral localization and a coarse longitudinal prior, vastly relaxing the requirements on prior localization results compared to the current state of the art. With four example signals, we demonstrate the effectiveness of our approach on a map verification dataset [2], detecting between 49 % and 98 % of all changed features at false alarm rates usually below 15 %. Detecting changes per feature allows to still use unchanged features for AD functions. When omitting this ability and aggregating all features, 98 % of all changed road sections can be detected successfully.successfully.

Detailed high-precision maps have proven to be essential for highly automated driving (HAD). Nevertheless, the issue of validation of such a detailed map has hardly been covered yet. To be a reliable complement for noisy sensors and probabilistic algorithms, a map has to be validated on the fly using data from onboard sensors on a level of detail much higher than previously tried. For detailed high-precision maps, needed for HAD tasks such as localization, the issue of outdated maps has not been examined yet. To estimate its extent, we reason why some road changes matter more than others and use this distinction to analyze almost 80 km of German highways for changes that render a map outdated. Our evaluation reveals almost 200 major road changes, outdating the maps of more than 32 km (41 %). Observed changes include renewals of lane markings, guardrails and whole road surfaces as well as complete reconstructions. Further on, to facilitate the research of map validation concepts, we publish a dataset consisting of major map features of the more than 80 km of German highways which we examined for changes. Finally, we propose the idea of using high-resolution aerial images as easily accessible source for geo-referenced data.

Voraussetzung fur eine zuverlassige Lokalisierung im Schienenverkehr sind geometrische Karten des Trassennetzes. Diese existieren haufig nicht. Manuelle Verfahren zur Kartierung sind mit hohen Kosten verbunden. Abhilfe schafft die Verwendung fahrzeuginterner Sensoren. Deren automatisierte Verarbeitung erfordert simultan zur Kartierung der Umgebung die Lokalisierung des Fahrzeugs. Abhangigkeiten zwischen der Fahrzeugbewegung und der Karte werden auf diese Weise berucksichtigt. Umfang: XII, 146 S. Preis: €40.00 | £37.00 | $70.00

Accurate localization is a fundamental component of driver-assistance systems and autonomous vehicles. For path-constrained motion, a map offers significant information and assists localization with valuable information about the evolution of the kinematic vehicle states. We propose natural parameterized cubic spline curves to approximate true motion constraints, particularly the centerline of individual road lanes or rail tracks. Vehicle kinematics is modeled in 1-D curve coordinates. Since map information is subject to uncertainties, a probabilistic treatment is a prerequisite to obtaining consistent localization results. The proposed probabilistic curvemap (PCM) and the close map-to-vehicle relation enable a straightforward derivation of measurement update equations without additional map-matching steps and offer themselves to classical filter techniques. Incoming sensor measurements are used for simultaneous vehicle localization and local PCM update around the current vehicle position. Thus, every revisit of a location reduces uncertainty in the local PCM. Moreover, when no prior information is provided in the PCM, extrapolation is carried out to handle these situations with incomplete maps. The proposed filter is validated through simulations and real-world railway experiments.

Next generation driver assistance systems demand a precise perception of the vicinity of the vehicle. Sensor readings are usually harnessed to gain knowledge of all moving and stationary obstacles. Commonly two paradigms are followed. Stationary environments are well modeled by non-parametric occupancy grids. Contrary, moving objects require a tracking and are not well suited for grid-based techniques. However, tracking objects requires some prior knowledge and parameterization which is inferior for unstructured cluttered obstacles. Herein we augment a grid-based mapping method designed for static environments with object tracking hence complementing both approaches. To this end we classify regions of stereo images into moving and stationary parts. The stationary part is fused in our static grid whereas moving parts are tracked yielding reliable motion estimates. The classifier used to distinguish moving from stationary parts is based on a Sequential Probability Ratio Test (SPRT), a model selection method which blends well into the tracking architecture. Thereby we achieve real-time operability on modest computing hardware.

Precise localization of rail vehicles is a key element toward the development and deployment of novel train control systems that offer enhanced security and efficiency. Typically, research on train navigation systems approaches this task either by data fusion of an increasing number of onboard sensors or by additional infrastructure installations that are combined with the localization of a global navigation satellite system (GNSS). The former approach is cost intensive and only gradually improves reliability and availability of localization information, whereas the latter approach suffers from the absence of satellite signals in places that are important for railroad applications such as tunnels or railway stations. In contrast, this paper employs a novel single eddy current sensor (ECS) mounted on the rail vehicle that directly pursues observations of the rail on a topological map. The localization task is formulated in a model-based probabilistic framework that enables us to derive signal processing techniques for board-autonomous speed estimation and recognition of particular events such as railroad switches by pattern recognition. In particular, turnouts are detected by Bayesian inference based on hidden Markov models (HMMs). In the final step, position on a topological map is estimated by sequential Monte Carlo sampling that combines speed and event information acquired from the ECS signal. Experiments with simulated and real-world data from an experimental rail vehicle indicate that the proposed system yields position and speed information of high reliability in real time.

This proposal presents the integration of interpolating global cubic splines into a general function regression framework to approximate curved functions or more dimensional curves based on noisy observations. We rearrange the iterative process of spline parameter calculation and obtain a practical linear matrix formulation. Then we employ Bayesian techniques to estimate spline model parameters and to perform model selection. While the number of basis functions is equal to the number of spline supporting points, this regularizer is automatically adapted towards an optimal trade-off between model complexity and model predicting performance, when Bayesian model selection is performed. Finally, we apply the proposed method within a robotic mapping scenario and learn geometric shapes of roads from noisy GPS positions measurements.

Localization of rail vehicles is fundamental for any autonomous systems to perform tasks in logistics or personal transport. This contribution presents a novel onboard localization system, based on an eddy current sensor system (ECS), that is capable of a precise train localization when combined with a simple topological map. In contrary to commonly applied travel distance determination by integrating the estimated velocity, we propose an event triggered counting approach, which makes use of the unique sensor capabilities. Rail switches are chosen as landmarks for a global map association and as reliable start and end points for the counting procedure. They are extracted via a Bayesian filter approach, in particular hidden Markov models are applied for detection and classification. Additional features are modeled in a subsequent step and merged within a topological map employing a naïve Bayesian approach in the spatial domain. This allows for a flexible sensor integration and an easy determination of the most probable vehicle position based on traveled distanced.

In automotive domain localization is typically performed through fusion of observations e.g. GPS positions and roadmaps. A transfer of these strategies to rail vehicle positioning is often impossible, because geometric track maps are not available. Key focus of this contribution is the enlargement of given topological track maps with geometric features, to enable a map-assisted rail vehicle localization based on geometric measurements. Initially we compute the optimal path within the track topology based on local track features that are measured with an eddy current sensor system. Then we process noisy INS positions and estimate the geometric shape of the map segments the measurement train has passed. Finally we augment the corresponding segments with geometric information. The proposed method is validated on real data in a real railway scenario.

Die Positionsbestimmung von Schienenfahrzeugen spielt eine zentrale Rolle in der Sicherheit des Bahnbetriebs, weil diese Information sowohl fur die Fahrwegsicherung als auch fur die Zugbeeinflussung verwendet wird. Die Basis der Eisenbahnsicherung ist die sichere Information uber die topologische Position aller Zuge und die sichere Freimeldung der Gleise. Daher sind sowohl die Gewinnung der Information uber die Zugposition als auch der Vollstandigkeitsuberwachung der Zuge hoch sicherheitsrelevant. Um die betriebliche Sicherheit nachweisen zu konnen, muss die Erfullung hoher Anforderungen an die Genauigkeit, Korrektheit, Zuverlassigkeit und Verfugbarkeit der Ortungsinformationen nach den Sicherheitsstandards der Eisenbahn nachgewiesen werden. Aus dem Bahnbetrieb resultieren weitere Anforderungen wie z. B. die Gleisselektivitat oder zeitliche Anforderungen um die Leistungsfahigkeit der Zugfolge erhalten zu konnen. Um die Eisenbahn auch okonomisch konkurrenzfahig zu machen, muss ein neues innovatives System sowohl in der Beschaffung wie auch im Betrieb kostengunstig sein. Ein Ansatz alle diese Ziele zu erreichen, ist es die gesamte Technologie fur die Ortung an Bord der Zuge zu konzentrieren. Das Ziel des Projektes “Entwicklung eines Demonstrators fur Ortungsaufgaben mit Sicherheitsverantwortung im Schienenguterverkehr – DemoOrt“ ist die Demonstration der Funktionalitat parallel zur detaillierten Untersuchung der Sicherheit wie auch der okonomischen Effizienz eines solchen fahrzeugautonomen Systems. Das hier entwickelte und untersuchte System basiert auf einem Globalen Satellitennavigationssystem (GNSS) – heute das amerikanische GPS, in Zukunft das Europaische GNSS GALILEO – in Kombination mit einer digitalen Karte und einem neuen innovativen Wirbelstromsensor. Das DemoOrt-Konsortium besteht aus dem Institut fur Verkehrssicherheit und Automatisierungstechnik (iVA) der Technischen Universitat Braunschweig (TU-BS), dem Institut fur Mess- und Regelungstechnik (MRT) der Universitat Karlsruhe, Bombardier Transportation Rail Control Solutions (BT-RCS) und dem Institut fur Verkehrsystemtechnik (ITS) des Deutschen Zentrums fur Luft- und Raumfahrt (DLR). Dieser Bericht stellt die Ergebnisse der Phasen I und II der vom Bundesministerium fur Wirtschaft und Technologie geforderten Projektes DemoOrt dar. In der ersten Phase wurde das System spezifiziert und die Anforderungen bezuglich Funktionalitat, Sicherheit und Wirtschaftlich¬keit gesammelt. Ebenso wurden zwei Prototypen aufgebaut und im Labor getestet. In der zweiten Phase wurden diese beiden Prototypen in Zugen eingebaut. Das System wurde auf je einer Strecke von Ettlingen nach Bad Herrenalb in der Nahe von Karlsruhe und von Poprad nach Strbske Pleso in der Hohen Tatra in der Slowakischen Republik getestet und demonstriert. Diese Demonstrationen fanden Ihren Abschluss im Jahr 2009 zusammen mit dem Abschluss der Studien uber Sicherheit und Wirtschaftlichkeit sowie der Auswertung der gesammelten Daten.

Precise train localization is a prerequisite for any efficient security and disposition tasks in modern transportation systems. In contrast to airplanes or ships, the localization of a train is not satisfyingly solvable with satellite systems. Occlusions in urban areas, tunnels and forests enforce the application of diverse sensor principles or cost intensive installations on track side. This contribution proposes an on-board train localization solely relying on a topological map and an eddy current sensor. The sensor principle is based on electromagnetic induction and is capable of estimating the speed of the train as well as detecting and classifying turnouts. These are represented by nodes in the map and allow absolute localization and recalibration of the distance measurement. Bayesian methods, in particular hidden Markov models in combination with the sequential Monte Carlo method, are used to solve the arising problems of a real-time and generalized detection and classification as well as localization within the topological map.

In the domain of rail vehicles a robust localization is a fundamental precondition for reasonable actions within operating systems. In addition to reliable sensor measurements an efficient and precise localization is highly dependent on accurate map information. In that context digital maps are often not available in the required quality. Key focus of this contribution is to substitute established manual map generation work flows through a fully automated mapping procedure. The mapping is realized in an off line estimation process, based on noisy position measurements from a satellite navigation system and local rail track features, that can be extracted from continuous eddy current sensor signals. Result of the mapping is a track specific topological map, enriched with geometrical information of turnout positions and track distances between them. The proposed method is validated on real data in a given railway scenario.

A reliable and precise velocity measurement is an important component of new train location systems. Several ways of measurement, e.g. with an optical sensor, GPS and eddy current sensors have already been under examination. This paper covers the velocity determination via correlation analysis using the measurement data of an eddy current sensor system. The common cross correlation is well suited to measure constant speeds. However, during phases of acceleration it is not sufficient for good accuracy due to a signal warp. For this reason a model based approach for signal rewarp will be presented in this paper, that eliminates or at least minimizes the influence of acceleration. Besides its main application in signal pre-processing for the velocity determination via cross correlation, the results of this algorithm can afterwards be used as an estimation of the acceleration as well.

Die Positionsbestimmung von Schienenfahrzeugen spielt eine zentrale Rolle in der Sicherheit des Bahnbetriebs, weil diese Information sowohl für die Fahrwegsicherung als auch für die Zugbeeinflussung verwendet wird. Die Basis der Eisenbahnsicherung ist die sichere Information über die topologische Position aller Züge und die sichere Freimeldung der Gleise. Daher sind sowohl die Gewinnung der Information über die Zugposition als auch der Vollständigkeitsüberwachung der Züge hoch sicherheitsrelevant. Um die betriebliche Sicherheit nachweisen zu können, muss die Erfüllung hoher Anforderungen an die Genauigkeit, Korrektheit, Zuverlässigkeit und Verfügbarkeit der Ortungsinformationen nach den Sicherheitsstandards der Eisenbahn nachgewiesen werden. Aus dem Bahnbetrieb resultieren weitere Anforderungen wie z. B. die Gleisselektivität oder zeitliche Anforderungen um die Leistungsfähigkeit der Zugfolge erhalten zu können. Um die Eisenbahn auch ökonomisch konkurrenzfähig zu machen, muss ein neues innovatives System sowohl in der Beschaffung wie auch im Betrieb kostengünstig sein. Ein Ansatz alle diese Ziele zu erreichen, ist es die gesamte Technologie für die Ortung an Bord der Züge zu konzentrieren. Das Ziel des Projektes “Entwicklung eines Demonstrators für Ortungsaufgaben mit Sicherheitsverantwortung im Schienengüterverkehr – DemoOrt“ ist die Demonstration der Funktionalität parallel zur detaillierten Untersuchung der Sicherheit wie auch der ökonomischen Effizienz eines solchen fahrzeugautonomen Systems. Das hier entwickelte und untersuchte System basiert auf einem Globalen …

Robust localization is a fundamental component of autonomous vehicles. In that context essential information has to be provided by a sufficient selection of sensor observations. In case of land-based road-constrained motion a map offers significant information and assists localization with valuable information about the evolution of the kinematic vehicle states. Throughout this proposal cubic spline curves are chosen to approximate the true motion constraints, e.g. roads or tracks, and the vehicle kinematics are modelled in one-dimensional curve coordinates. The resulting map-to-vehicle relation enables a straight forward derivation of measurement update equations and offers itself to classical filter techniques. Incoming sensor measurements are used for a simultaneous vehicle localization and a local road map update around the current vehicle position. Moreover map-extrapolation is carried out to handle situations with incomplete maps. The proposed method is validated with simulations and within a real railway scenario.

An exact localization of trains is essential for effective disposition and design of modern train operating systems, allowing a better use of the given infrastructure. In this paper we propose to use turnouts on rail tracks as absolute landmarks and re-calibration points for onboard location systems. The measurements base on an eddy current sensor system, additionally providing speed information through correlating inhomogeneities along the rail track. This paper presents a hidden Markov model approach that offers a robust detection and separation of turnouts. The proposed algorithm makes it possible to process whole train stations with successive turnouts continuously, to perform a first low-level classification and to separate close events that can be accurately cut out of the signal, which is a basis for an advanced classification.

This contribution presents the project DemoOrt which aims to show the technical and economic feasability of the safe localosation of trains using the combination of satelliten navigation, eddy current sensors and a digital map. The project has been funded by the German Ministry of economics and technology and is going into a one year demonstration phase now.