Search

We address the problem of calibrating an embedded depth camera network designed for people tracking purposes. In our system, the nodes of the network are responsible for detecting the people moving in their view, and sending the observations to a centralized server for data fusion and tracking. We employ a plan-view approach where the depth camera views are transformed to top-view height maps where people are observed. As the server transforms the observations to a global plan-view coordinate system, accurate geometric calibration of the sensors has to be performed. Our main contribution is an auto-calibration method for such depth camera networks. In our approach, the sensor network topology and the initial 2D rigid transformations that map the observations to the global frame are determined using observations only. To distribute the errors in the initial calibration, the transformation parameters and the estimated positions of people are refined using a global optimization routine. To overcome inaccurate depth camera parameters, we re-calibrate the sensors using more flexible transformations, and experiment with similarity, affine, homography and thin-plate spline mappings. We evaluate the robustness, accuracy and precision of the approach using several real-life data sets, and compare the results to a checkerboard-based calibration method as well as to the ground truth trajectories collected with a mobile robot.

The unobtrusive methods to monitor and track people behaviour are getting more common. At the same time novel lighting control and the LED technologies allow dynamic real-time controlling of the illumination systems. This progress will support the development of implicit interaction between humans and lighting systems. In this paper, we describe the results of our study on how to create a situation-aware real-time light control system that adapts automatically to person behaviour in ambient space. In future this approach could be used e.g. for guiding visitors in public places or promoting new products. We have implemented a proof-of-concept application -- a Treasure Hunt game - by using a depth sensor based people tracking service, wirelessly controlled RGB LEDs and a light controlling unit where the LED bulbs change their colour according to person position and movement in a room.

Laboratory scale punch through tests on floating rubble consisting of plastic blocks were conducted and simulated with a 3D discrete numerical model. The purpose was to analyse the experimental method and to validate the model. The motivation of using plastic blocks instead of ice was to simplify the interpretation of results as the plastic blocks do not freeze together. The indentor force and the lateral force induced by the rubble on one of the basin walls were recorded as a function of indentor penetration. Further, the experiments were recorded with a video camera and a motion tracking software was used to analyse the rubble deformation. The force records and deformation patterns from the experiments and simulations were in agreement. The evolution of the deformation patterns could be closely linked to the indentor force records, which demonstrates the need for the numerical model to correctly represent the rubble deformation. The experiments and the simulations showed, that the lateral force within the pile increased considerably during a punch through experiment. This makes the interpretation of punch through experiment results for material modelling challenging: the friction angle of the rubble can become overestimated making the punch through test unsuitable for achieving accurate values for friction angle. Consequently, no value for the rubble friction angle was derived here.

In augmented reality, marker-based tracking is the most common method for camera pose estimation. Most of the markers are black and white patterns that are visually obtrusive, but they can be hidden from the video using image inpainting methods. In this paper, we present a computationally efficient approach to achieve this. We use a high-resolution hiding texture, which is captured and generated only once. To capture continuous changes in illumination, reflections and exposure, we also compute a very low-resolution texture at each frame. The coarse and fine textures are combined to obtain a detailed hiding texture which reacts to changing conditions and runs efficiently in mobile phone environments.

We applied a computer vision based method and developed multi-touch technology to be adopted in various display types. In the selected design, multiple cameras are placed on the side of the display with their optical axis parallel to the screen. The display edges opposite the cameras are illuminated and fingers are detected since they block the light in the camera images. The approach is scalable and can be used in a wide variety of displays. Due to self-occlusion of the touchpoints, it is challenging to relate camera measurements to tracked points. In this paper, we present our approach for tracking and managing multiple touchpoints in such camera set-ups. We describe the mathematical background for modeling and calibrating the cameras, the design of the extended Kalman filter for point tracking, and the logic for adding, updating and removing the touchpoints. We analyze the potential accuracy and robustness of the system using several simulations and present two different real-life implementations of the approach.

One of the main challenges of introducing Augmented Reality technology into industrial applications is fast, reliable, low-cost authoring of animated AR content of product data created in 3D CAD systems. In this paper we present a product data information pipeline from industry used commercial CAD systems to a marker-based augmented reality system. The pipeline utilizes ISO 10303 (STEP) data protocol, DFA-Tool (Design For Assembly), and VTT’s proprietary content creation and augmenting software. The pipeline is verified with a real industrial application.

In this paper, we describe the demonstration system in which a wooden 3D puzzle is assembled using an augmented reality system. The 3D puzzle emulated a simplified assembly task in a factory. We use the 3D puzzle as a means to study how to implement an augmented assembly system to a real setting in a factory and explore the advantages and disadvantages of augmented reality techniques. With this demonstration system we also test, demonstrate and evaluate different input modalities for augmented assembly setup. Furthermore, the demonstration system helps us better interact with industry and get invaluable feedback.

In this paper, a multimodal system for augmented reality aided assembly work is designed and implemented. The multimodal interface allows for speech and gestural input. The system emulates a simplified assembly task in a factory. A 3D puzzle is used to study how to implement the augmented assembly system to a real setting in a factory. The system is used as a demonstrator and as a test-bed to evaluate different input modalities for augmented assembly setups. Preliminary system evaluation results are presented, the user experience is discussed, and some directions for future work are given.