Marker-based optical outside-in tracking is a mature and robust technology used by many AR, VR and motion capture applications. However, in small environments the tracking cameras are often difficult to install. An example scenario are ergonomic studies in car manufacturing, where the motion of a worker needs to be tracked in small spaces such as the trunk of a car. In this paper, we describe how to extend the tracking volume in small, cluttered environments using small and flexible wireless cameras in form of unmodified mobile phones that can quickly be installed. Since those small cameras are not synchronized with the main tracking cameras, we describe several modifications to the tracking algorithms, such as inter-frame interpolation, the replacement of the least-squares adjustment by a Kalman filter and the integration of rolling-shutter compensation. To support the quick setup of additional cameras while the tracking system is running, the system is extended by an on-line calibration technique that determines the extrinsic camera parameters without requiring a dedicated calibration step.

In industrial manufacturing processes, targets for infrared marker based tracking have to be robust and must integrate into the tools without disturbing the work flow. In this paper, we propose cylindrical markers attached directly to the tool. We show that, targets equipped with cylindrical markers can be tracked with about the same precision as targets equipped with spherical markers, if a correction for the reduced symmetry is applied. Additionally, the markers can be placed on different parts of a tool with a flexible connection, which is also considered in this paper.

In many augmented reality applications, in particular in the medical and industrial domains, knowledge about tracking errors is important. Most current approaches characterize tracking errors by 6×6 covariance matrices that describe the uncertainty of a 6DOF pose, where the center of rotational error lies in the origin of a target coordinate system. This origin is assumed to coincide with the geometric centroid of a tracking target. In this paper, we show that, in case of a multi-camera fiducial tracking system, the geometric centroid of a body does not necessarily coincide with the point of minimum error. The latter is not fixed to a particular location, but moves, depending on the individual observations. We describe how to compute this point of minimum error given a covariance matrix and verify the validity of the approach using Monte Carlo simulations on a number of scenarios. Looking at the movement of the point of minimum error, we find that it can be located surprisingly far away from its expected position. This is further validated by an experiment using a real camera system.

Indoor tracking scenarios still face challenges in providing continuous tracking support in wide-area workplaces. This is especially the case in Augmented Reality since such augmentations generally require exact full 6DOF pose measurements in order to continuously display 3D graphics from user-related view points. Many single sensor systems have been explored but only few of them have the capability to track reliably in wide-area environments. We introduce a mobile multi-sensor platform to overcome the shortcomings of single sensor systems. The platform is equipped with a detachable optical camera and a rigidly mounted odometric measurement system providing relative positions and orientations with respect to the ground plane. The camera is used for marker-based as well as for marker-less (feature-based) inside-out tracking as part of a hybrid approach. We explain the principle tracking technologies in our competitive/cooperative fusion approach and show possible enhancements to further developments. This inside-out approach scales well with increasing tracking range, as opposed to stationary outside-in tracking.

We describe a mobile augmented reality application that is based on 3D snapshotting using multiple photographs. Optical square markers provide the anchor for reconstructed virtual objects in the scene. A novel approach based on pixel flow highly improves tracking performance. This dual tracking approach also allows for a new single-button user interface metaphor for moving virtual objects in the scene. The development of the AR viewer was accompanied by user studies confirming the chosen approach.

The task of vision based people tracking is a major research problem in the context of surveillance applications or human behavior estimation, but it has had only minimal impact on (Ubiquitous) Augmented Reality applications thus far. Deploying stationary infrastructural cameras within indoor environments for the purpose of Augmented Reality could provide a users' devices with additional functionality that a small device and mobile sensors cannot provide to its user. Therefore people tracking could be expected to become an ubiquitously available infrastructural element in buildings since surveillance cameras are widely used. The use for scenarios indoors or close to buildings is obvious. We present and discuss several different ways where people tracking in real-time could influence the fields of Augmented Reality and further vision based applications.

Ubiquitous tracking (Ubitrack) setups, consisting of many previously unknown sensors, offer many possibilities to perform sensor fusion in order to increase robustness and accuracy. In particular, the dynamic combination of mobile and stationary trackers enables the creation of new wide-area tracking concepts. In this work, we present a setup in which a gyroscope is dynamically fused with three different mobile and stationary sensors, based on the concepts of spatial relationship graphs (SRGs) and patterns. For this, we contribute new patterns that, based on well-known algorithms, enable the transformation of rotation velocity and the fusion with different absolute trackers. The usefulness of the approach is shown in a system that automatically reconfigures the SRG based on course tracking data, and, depending on the structure of this SRG, automatically selects a suitable fusion algorithm.

It is widely recognized that inertial sensors, in particular gyroscopes, can improve the latency and accuracy of orientation tracking by fusing the inertial measurements with data from other sensors. In our previous work, we introduced the concepts of spatial relationship graphs and spatial relationship patterns to formally model multi-sensor tracking setups and derive valid applications of well-known algorithms in order to infer new spatial relationships for tracking and calibration. In this work, we extend our approach by providing additional spatial relationship patterns that transform incremental rotations and add gyroscope alignment and fusion. The usefulness of the resulting tracking configurations is evaluated in two different scenarios with both inside-out and outside-in tracking.

The paper explores the use of laser projectors as an alternative to head-mounted displays for Augmented Reality. We describe the development of an Augmented Reality Laser Projector and report on experiences setting up AR systems that use laser projectors, reasoning about several design criteria.

Ubiquitous tracking setups, covering large tracking areas with many heterogeneous sensors of varying accuracy, require dedicated middleware to facilitate development of stationary and mobile applications by providing a simple interface and encapsulating the details of sensing, calibration and sensor fusion. In this paper we present a centrally coordinated peer-to-peer architecture for ubiquitous tracking, where a server computes optimal data flow configurations for sensor and application clients, which are directly exchanging tracking data with low latency using a light-weight data flow framework. The server's decisions are inferred from an actively maintained central spatial relationship graph (SRG) using spatial relationship patterns. The system is compared to a previous Ubitrack implementation using the highly distributed DWARF middleware. It exhibits significantly better performance in a reference scenario.

Marker-based optical tracking systems are widely used in augmented reality, medical navigation and industrial applications. We propose a model for the prediction of the target registration error (TRE) in these kinds of tracking systems by estimating the fiducial location error (FLE) from two-dimensional errors on the image plane and propagating that error to a given point of interest. We have designed a set of experiments in order to estimate the actual parameters of the model for any given tracking system. We present the results of a study which we used to demonstrate the effect of different sources of error. The method is applied to real applications to show the usefulness for any kind of augmented reality system. We also present a set of tools that can be used to visualize the accuracy at design time.

With tracking setups becoming increasingly complex, it gets more difficult to find suitable algorithms for tracking, calibration and sensor fusion. A large number of solutions exists in the literature for various combinations of sensors, however, no development methodology is available for systematic analysis of tracking setups. When modeling a system as a spatial relationship graph (SRG), which describes coordinate systems and known transformations, all algorithms used for tracking and calibration correspond to certain patterns in the graph. This paper introduces a formal model for representing such spatial relationship patterns and presents a small catalog of patterns frequently used in augmented reality systems. We also describe an algorithm to identify patterns in SRGs at runtime for automatic construction of data flows networks for tracking and calibration.

Augmented reality (AR) provides a natural interface to the "calm" pervasive technology anticipated in large-scale ubiquitous computing environments. However, the range of classic AR applications has been limited by the scope, range and cost of sensors used for tracking. Hybrid tracking approaches can go some way to extending this range. We propose an approach, called ubiquitous tracking, in which data from widespread and diverse heterogeneous tracking sensors is automatically and dynamically fused, and then transparently provided to applications. A formal model represents spatial relationships between objects as a graph attributed with quality-of-service parameters. This paper presents a software implementation, in which a dynamic data flow network of distributed software components is thereby constructed in response to queries and optimisation criteria specified by applications. This implementation is demonstrated using a small laboratory example, and larger setups modelled in a simulation environment.