No abstract

No abstract

Estimating surface normals is an important task in computer vision, e.g. in surface reconstruction, registration and object detection. In stereo vision, the error of depth reconstruction increases quadratically with distance. This makes estimation of surface normals an especially demanding task.In this paper, we analyze how error propagates from noisy disparity data to the orientation of the estimated surface normal. Firstly, we derive a transformation for normals between disparity space and world coordinates. Afterwards, the propagation of disparity noise is analyzed by means of a Monte Carlo method. Normal reconstruction at a pixel position requires to consider a certain neighborhood of the pixel. The extent of this neighborhood affects the reconstruction error. Our method allows to determine the optimal neighborhood size required to achieve a pre specified deviation of the angular reconstruction error, defined by a confidence interval. We show that the reconstruction error only depends on the distance of the surface point to the camera, the pixel distance to the principal point in the image plane and the angle at which the viewing ray intersects the surface.

ABSTRACT:In this paper we are presenting work done within the joint development project ANKommEn. It deals with the development of a highly automated robotic system for fast data acquisition in civil disaster scenarios. One of the main requirements is a versatile system, hence the concept embraces a machine cluster consisting of multiple fundamentally different robotic platforms. To cover a large variety of potential deployment scenarios, neither the absolute amount of participants, nor the precise individual layout of each platform shall be restricted within the conceptual design. Thus leading to a variety of special requirements, like onboard and online data processing capabilities for each individual participant and efficient data exchange structures, allowing reliable random data exchange between individual robots. We are demonstrating the functionality and performance by means of a distributed mapping system evaluated with real world data in a challenging urban and rural indoor / outdoor scenarios.

ABSTRACT:The usage of unmanned systems for exploring disaster scenarios has become more and more important in recent times as a supporting system for action forces. These systems have to offer a well-balanced relationship between the quality of support and additional workload. Therefore within the joint research project ANKommEn -german acronym for Automated Navigation and Communication for Exploration -a system for exploration of disaster scenarios is build-up using multiple UAV und UGV controlled via a central ground station. The ground station serves as user interface for defining missions and tasks conducted by the unmanned systems, equipped with different environmental sensors like cameras -RGB as well as IR -or LiDAR. Depending on the exploration task results, in form of pictures, 2D stitched orthophoto or LiDAR point clouds will be transmitted via datalinks and displayed online at the ground station or will be processed in short-term after a mission, e.g. 3D photogrammetry. For mission planning and its execution, UAV/UGV monitoring and georeferencing of environmental sensor data, reliable positioning and attitude information is required. This is gathered using an integrated GNSS/IMU positioning system. In order to increase availability of positioning information in GNSS challenging scenarios, a GNSS-Multiconstellation based approach is used, amongst others. The present paper focuses on the overall system design including the ground station and sensor setups on the UAVs and UGVs, the underlying positioning techniques as well as 2D and 3D exploration based on a RGB camera mounted on board the UAV and its evaluation based on real world field tests.