Claus B. Madsen

Optimal landmark selection for triangulation of robot position

Abstract A mobile robot can identify its own position relative to a global environment model by using triangulation based on three landmarks in the environment. It is shown that this procedure may be very sensitive to noise depending on spatial landmark configuration, and relative position between robot and landmarks. A general analysis is presented which permits prediction of the uncertainty in the triangulated position. In addition an algorithm is presented for automatic selection of optimal landmarks. This algorithm enables a robot to continuously base its position computation on the set of available landmarks, which provides the least noise sensitive position estimate. It is demonstrated that using this algorithm can result in more than one order of magnitude reduction in uncertainty.

Landmark-based navigation strategies

A mobile robot can identify its own position relative to a global environment model using triangulation based on measuring angular separation between three landmarks in the environment. Multiple views from different locations of a smaller set of landmarks can also be used though. Alternatively the current position estimates can be updated using heading and distance measurements to a single landmark. Using these different strategies 8 position estimating techniques have been designed, analyzed and compared. These are based on viewing 1, 2 and 3 landmarks from one or two different viewpoints. It is shown that these procedures may be very sensitive to noise depending on the spatial landmark configuration, and relative position between robot and landmarks. A general analysis is presented which permits prediction of the uncertainty in the triangulated position. The uncertainty measure can be used to determine which of the light different techniques is the most suitable in specific situations. The entire analysis is based on a basic statistical approach, and verified experimentally. In addition to the evaluation of the individual techniques, an algorithm is presented for automatic selection of optimal landmarks. This algorithm enables a robot to continuously estimate its current position from the set of landmarks which provides the most stable solution. It is demonstrated that using this algorithm can result in more than one order of magnitude reduction in position uncertainty.

A viewpoint planning strategy for determining true angles on polyhedral objects by camera alignment

The paper presents a viewpoint planning strategy which automatically guides a movable camera from an arbitrary position to a position where the optical axis is perpendicular to a plane spanned by any two intersecting edges on a polyhedral object, i.e., a junction. In related work it is common to use the changing length of edge segments to control such alignment, but we demonstrate the use of the apparent angle between the edge segments of the junction. Based on the control on the apparent angle we achieve robustness as well as independence of distance to the object and focal length. The strategy is able to determine the true angle of a junction with an accuracy of approximately 1/spl deg/, and align with an accuracy of approximately 6/spl deg/.

Navigation using range images on a mobile robot

Abstract We describe an integrated navigation system for an autonomous mobile robot using a laser range camera to obtain knowledge about the environment. The implemented system maintains a 2D world model (floor map) by integrating knowledge obtained from several range images acquired as the robot moves around in its attempt to find a path to the goal position. A path planner uses the floor map to generate collision-free paths consisting of sequences of configurations. Car-like kinematic constraints ensure smooth paths that can be send directly to a wheel controller. The system was implemented on the HERMIES-III robot, a vehicle with 3 degrees of freedom, and tested in both laboratory and simulated environments. These tests showed that a simple integration of the environment modeler and the path planner provides the robot with basic explorative and navigational capabilities. In particular, the system is capable of performing total re-planning in cases where the initial path to the goal point turns out to be blocked.

A comparative study of the robustness of two pose estimation techniques

Abstract.The paper presents an analysis of the stability of pose estimation. Stability is defined as sensitivity of the pose parameters towards noise in image features used for estimating pose. The specific emphasis of the analysis is on determining {how the stability varies with viewpoint} relative to an object and to understand the relationships between object geometry, viewpoint, and pose stability. Two pose estimation techniques are investigated. One uses a numerical scheme for finding pose parameters; the other is based on closed form solutions. Both are “pose from trihedral vertices” techniques, which provide the rotation part of object pose based on orientations of three edge segments. The analysis is based on generalized sensitivity analysis propagating the uncertainty in edge segment orientations to the resulting effect on the pose parameters. It is shown that there is a precomputable, generic relationship between viewpoint and pose stability, and that there is a drastic difference in stability over the range of viewpoints. This viewpoint variation is shared by the two investigated techniques. Additionally, the paper offers an explicit way to determine the most robust viewpoints directly for any given vertex model. Experiments on real images show that the results of the work can be used to compute the variance in pose parameters for any given pose. For the predicted {instable} viewpoints the variance in pose parameters is on the order of 20 (degrees squared), whereas the variance for robust viewpoints is on the order of 0.05 (degrees squared), i.e., two orders of magnitude difference.

Estimation of Dynamic Light Changes in Outdoor Scenes Without the use of Calibration Objects

The work presented in this paper explores how dynamical light parameters in an outdoor environment can be estimated for use in a real-time augmented reality (AR) system. A method using existing inverse rendering techniques is used to acquire diffuse surface reflectances in an offline procedure. The reflectances are used in an on-line procedure to estimate the illumination parameters of an outdoor scene. The method presented reduces the light estimation problem of outdoor scenes to a modified Phong shading model with two unknown parameters, which can be determined through the use of a linear equations system. The work presented provides an elegant method for estimating dynamically changing illumination parameters without the need for a calibration object in the scene

Shadow Detection in Dynamic Scenes Using Dense Stereo Information and an Outdoor Illumination Model

We present a system for detecting shadows in dynamic outdoor scenes. The technique is based on fusing background subtraction operations performed on both color and disparity data, respectively. A simple geometrical analysis results in an ability to classify pixels into foreground, shadow candidate, and background. The shadow candidates are further refined by analyzing displacements in log chromaticity space to find the shadow hue shift with the strongest data support and ruling out other displacements. This makes the shadow detection robust towards false positives from rain, for example. The techniques employed allow for 3Hz operation on commodity hardware using a commercially available dense stereo camera solution.

Segmentation of soft shadows based on a daylight- and penumbra model

This paper introduces a new concept within shadow segmentation for usage in shadow removal and augmentation through construction of an alpha overlay shadow model. Previously, an image was considered to consist of shadow and non-shadow regions. We construct a model that accounts for sunlit, umbra and penumbra regions. The model is based on theories about color constancy, daylight, and the geometry that causes penumbra. The behavior of the model is analyzed and a graph cut energy minimization is applied to estimate the alpha parameter. The approach is demonstrated on natural complex image situations. The results are convincing, but the alpha gradient in penumbra must be improved.

Temporal Coherence Strategies for Augmented Reality Labeling

Temporal coherence of annotations is an important factor in augmented reality user interfaces and for information visualization. In this paper, we empirically evaluate four different techniques for annotation. Based on these findings, we follow up with subjective evaluations in a second experiment. Results show that presenting annotations in object space or image space leads to a significant difference in task performance. Furthermore, there is a significant interaction between rendering space and update frequency of annotations. Participants improve significantly in locating annotations, when annotations are presented in object space, and view management update rate is limited. In a follow-up experiment, participants appear to be more satisfied with limited update rate in comparison to a continuous update rate of the view management system.

Graph cut based segmentation of soft shadows for seamless removal and augmentation

This paper introduces a new concept within shadow segmentation for usage in shadow removal and augmentation through construction of a multiplicity alpha overlay shadow model. Previously, an image was considered to consist of shadow and non-shadow regions. This makes it difficult to seamlessly remove shadows and insert augmented shadows that overlap real shadows. We construct a model that accounts for sunlit, umbra and penumbra regions by estimating the degree of shadow. The model is based on theories about color constancy, daylight, and the geometry that causes penumbra. A graph cut energy minimization is applied to estimate the alpha parameter. Overlapping shadow augmentation and removal is also demonstrated. The approach is demonstrated on natural complex image situations. The results are convincing, and the quality of augmented shadows overlapping real shadows and removed shadows depends on the quality of the estimated alpha gradient in penumbra.