Propagating impulse sounds are sensitive to the varying near-surface atmosphere. This study reports on an experimental assessment of this sensitivity under well-controlled outdoor conditions. The experiment, conducted over a flat terrain, features 14 synchronous acoustic sensors at ranges up to 450 m from reproducible, transient sources. It scanned over the upwind, crosswind, and downwind propagations, and also documents the temporal and spatial coherences of the acoustic field. Concurrent atmospheric measurements documented the near-surface, essentially wind-driven atmosphere, and included turbulence monitoring. The analysis reveals how the environmental propagation processes combine to form the large variety of recorded signatures. The deterministic versus stochastic variations of the signatures are distinguished, and both are shown to affect the time of arrival (wander) and the shape (spread) of the pulses. The study also discusses the potential impacts of these variations on acoustic sensing of transient signals like gun shots and explosions.

Sound propagation is largely affected by the near-surface conditions (ground type, turbulence, and weather). The presentation reports on a joint research effort aimed at improving fundamental understanding of these propagation effects on impulsive sounds. Data collections will be presented, which consist of measuring transient acoustic signals at distance from a reproducible sound source, under a variety of acoustic-atmospheric conditions. Analysis of these data, supported by theoretical studies and numerical simulations, is conducted to reveal how the environmental conditions drive the observed variations of acoustic pulses. Challenges in measurements, simulations, and analysis will be discussed.

Für industrielle Vermessungsaufgaben wird zunehmend Bildverarbeitung eingesetzt. Beispielsweise kann hiermit auch die Poseabweichung zwischen Ist- und Sollwert bei Roboter-Endeffektoren verbessert werden. In dieser Arbeit wird ein System vorgestellt, welches die Pose einer Kamera und zusätzlich die 3D-Koordinaten von bestimmten Objektpunkten schätzt. Als Eingabedaten verwendet das System mehrere Bilder, welche Landmarken bei Betrachtung von verschiedenen Kameraposen enthalten. Die Schätzung der Pose wird mittels nichtlinearer Optimierung durchgeführt. Nach lokaler Linearisierung wird eine Kalman-Schätzung in Kovarianzform durchgeführt. Simulations- und Messergebnisse belegen die Leistungsfähigkeit des Verfahrens.