Dissertationen zum Thema „Engineering, Aerospace|Physics, Astrophysics|Computer Science“

A computer vision-based navigation feasibility study consisting of two navigation algorithms is presented to determine whether computer vision can be used to safely navigate a small semi-autonomous inspection satellite in proximity to the International Space Station. Using stereoscopic image-sensors and computer vision, the relative attitude determination and the relative distance determination algorithms estimate the inspection satellite's relative position in relation to its host spacecraft. An algorithm needed to calibrate the stereo camera system is presented, and this calibration method is discussed. These relative navigation algorithms are tested in NASA Johnson Space Center's simulation software, Engineering Dynamic On-board Ubiquitous Graphics (DOUG) Graphics for Exploration (EDGE), using a rendered model of the International Space Station to serve as the host spacecraft. Both vision-based algorithms proved to attain successful results, and the recommended future work is discussed.

This work proposes a new method for using reduced order models in lieu of high fidelity analysis during the sensitivity analysis step of gradient based design optimization. The method offers a reduction in the computational cost of finite difference based sensitivity analysis in that context.

The method relies on interpolating reduced order models which are based on proper orthogonal decomposition. The interpolation process is performed using radial basis functions and Grassmann manifold projection. It does not require additional high fidelity analyses to interpolate a reduced order model for new points in the design space. The interpolated models are used specifically for points in the finite difference stencil during sensitivity analysis.

The proposed method is applied to an airfoil shape optimization (ASO) problem and a transport wing optimization (TWO) problem. The errors associated with the reduced order models themselves as well as the gradients calculated from them are evaluated. The effects of the method on the overall optimization path, computation times, and function counts are also examined.

The ASO results indicate that the proposed scheme is a viable method for reducing the computational cost of these optimizations. They also indicate that the adaptive step is an effective method of improving interpolated gradient accuracy. The TWO results indicate that the interpolation accuracy can have a strong impact on optimization search direction.

The study of canonical flows, such as channels, pipes, or boundary layers, is essential for a deeper understanding of the physical mechanisms present in wall-bounded turbulence. Of particular importance in flows delimited by solid walls is the near-wall region where a large fraction of the drag stems from velocity fluctuations in a thin boundary layer adjacent to surfaces. In that context it is interesting to recognize that globally about 10% of all energy is used to overcome turbulent drag in one way or another. The goal of this study is to clarify our understanding in these areas by combining computations and experiments of turbulent duct flows and boundary layers. Oil film interferometry (OFI) and static pressure measurements were carried out over the range 200 < Re_τ < 800 (where Re_τ is the friction Reynolds number, based on duct half-height h and friction velocity u _τ) in an adjustable-geometry duct flow facility. Three-dimensional effects were studied by considering different aspect ratio (AR) configurations. Contrary to the accepted understanding in the field of turbulence research, we found that an aspect ratio of at least 24 is required in order to obtain "high-AR duct conditions" , and a development length of around 200 duct full-heights (for low and intermediate Reynolds numbers) is necessary for appropriate flow development.

The three-dimensional effects present in the flow, i.e., side-wall boundary layers and secondary motions, are also studied by means of direct numerical simulations (DNSs). The spectral element code Nek5000, developed by Fischer et al. (2008), is used to compute turbulent duct flows with aspect ratios from 1 to 10 in streamwise-periodic boxes of length 25h (long enough to capture the longest streamwise structures) and Reynolds numbers Re_τ,c = 180 and 330. While preparing the duct simulations, we also considered the necessary averaging times for converged statistics in simulations of wall turbulence; as a result, a set of guidelines regarding sampling times and intervals is also given. We find that the conditions often computed in z-periodic channels cannot be reproduced experimentally, even at very high aspect ratios such as 48, and therefore conclude that "computational channels" and "experimental high-AR ducts" are two different flows. The implications of these findings on wind tunnel experiments (with aspect ratios typically ranging from 3 to 16), and the large volume of available "two-dimensional" zero pressure gradient boundary layer data, are also assessed in this study. We therefore recommend the computational and experimental study of turbulent pipe flows, since this is the only case where matching canonical conditions can be obtained both in DNS computations and experimental facilities.

In addition, we re-analyze currently available Pitot tube corrections for ZPG turbulent boundary layer measurements, and propose new forms with coefficients dependent on inner-scaled Pitot tube diameter, [special characters omitted]. Reynolds number and probe size effects are both introduced in these coefficients, yielding excellent collapse of data over a much wider range of Pitot tube diameters (from 0.2 to 12.82 mm), and very good agreement with reference hot-wire and PIV data. We developed a new correcting scheme, called κ B—Musker, which is able to provide the highest possible accuracy in probe position when applied to profile measurements of wall-bounded flows.

A software prototype for autonomous 3D scanning of uncooperatively rotating orbital debris using a point cloud sensor is designed and tested. The software successfully generated 3D models under conditions that simulate some on-orbit orbit challenges including relative motion between observer and target, inconsistent target visibility and a target with more than one plane of symmetry. The model scanning software performed well against an irregular object with one plane of symmetry but was weak against objects with 2 planes of symmetry.

The suitability of point cloud sensors and algorithms for space is examined. Terrestrial Graph SLAM is adapted for an uncooperatively rotating orbital debris scanning scenario. A joint EKF attitude estimate and shape similiarity loop closure heuristic for orbital debris is derived and experimentally tested. The binary Extended Fast Point Feature Histogram (EFPFH) is defined and analyzed as a binary quantization of the floating point EFPFH. Both the binary and floating point EPFH are experimentally tested and compared as part of the joint loop closure heuristic.

This dissertation develops and assesses a cascaded method for designing optimal periodic trajectories and link schedules for an unmanned aircraft to ferry data between stationary ground nodes. This results in a fast solution method without the need to artificially constrain system dynamics. Focusing on a fundamental ferrying problem that involves one source and one destination, but includes complex vehicle and Radio-Frequency (RF) dynamics, a cascaded structure to the system dynamics is uncovered. This structure is exploited by reformulating the nonlinear optimization problem into one that reduces the independent control to the vehicle's motion, while the link scheduling control is folded into the objective function and implemented as an optimal policy that depends on candidate motion control. This formulation is proven to maintain optimality while reducing computation time in comparison to traditional ferry optimization methods.

The discrete link scheduling problem takes the form of a combinatorial optimization problem that is known to be NP-Hard. A derived necessary condition for optimality guides the development of several heuristic algorithms, specifically the Most-Data-First Algorithm and the Knapsack Adaptation. These heuristics are extended to larger ferrying scenarios, and assessed analytically and through Monte Carlo simulation, showing better throughput performance in the same order of magnitude of computation time in comparison to other common link scheduling policies. The cascaded optimization method is implemented with a novel embedded software system on a small, unmanned aircraft to validate the simulation results with field experiments. To address the sensitivity of results on trajectory tracking performance, a system that combines motion and link control with waypoint-based navigation is developed and assessed through field experiments. The data ferrying algorithms are further extended by incorporating a Gaussian process to opportunistically learn the RF environment. By continuously improving RF models, the cascaded planner can continually improve the ferrying system's overall performance.

During an extravehicular activity (EVA), the role of an astronaut involves a multitude of complex tasks. Whether that task is a science experiment aboard the International Space Station, or traversing extraterrestrial terrain – attention, communication, and instruction are essential. As an aid, augmented reality (AR) can portray suit informatics and procedures within line-of-sight while minimizing attentional loss. Currently, there exists little research highlighting the human systems considerations to qualify AR systems for space suit applications. This study quantifies user interface (UI) and human performance measures for an AR prototype on the Mark III space suit. For user testing, 21 military pilots and personnel (11 men, 10 women) evaluated UI search tasks and completed a series of AR-instructed EVA dexterity tasks in an elevated luminosity, background clutter, and workload scenario. UI results suggest correlations for readability and usability; whereas, human performance results provide situational awareness, workload, and task performance data.

The main objective of the current dissertation is to develop a "Plug and Play" autopilot. We present a systematic approach to decouple controller and filter design from hardware selection. This approach also addresses the performance decay due to hardware limitations. The outcome of this work is an integrated environment to develop, validate, and test algorithms to be used on flight controllers.

The dynamic model of an off-the-shelf radio controlled airplane is derived. A controller is developed for the aircraft and it is implemented within the proposed framework. The framework and the controller developed are validated using hardware-in-the-loop simulation. A physical experiment with an autonomous ground vehicle is performed to test the autopilot and algorithms before test flights are performed. Finally a test flight is performed using a Great Planes Funster.

A novel microsatellite attitude controller is presented. This controller is also developed and implemented using the approach presented in this dissertation. A satellite attitude hybrid simulator is presented and its design is discussed. An experimental apparatus to test the controller in a microgravity environment is constructed and discussed.

Finally a set of experiments that demonstrate how the framework can be integrated into commercially available autopilots and vehicles is presented. First an off-the-shelf quad-rotor that integrates a look-down camera is used to perform visual navigation and landing. Second, a rover and a quad-rotor are used on a cooperative schema. The rover follows a route and the quad-rotor escorts it. The third experiment presents a popular autopilot on a surveillance mission. For this mission the airplane is equipped not only with the autopilot and radio link, but also with a video system. The video system consists of a camera and a radio link. These experimental results demonstrate the utility of the proposed framework in enhancing the capabilities of off-the-shelf autopilots and vehicles while simultaneously simplifying mission preparation and execution.

In this thesis, we have investigated the Recursive Projection Method, RPM, as an accelerator for computations of both steady and unsteady flows, and as a stabilizer in a bifurcation analysis.

The criterion of basis extraction is discussed. It can be interpreted as a tolerance for the accuracy of the eigenspace spanned by the identified basis, alternatively it can be viewed as a criterion when the approximative Krylov sequence becomes numerically rank deficient.

Steady state calculations were performed on two different turbulent test-cases; a 2D supersonic nozzle flow with the Spalart-Allmaras 1-equation model and a 2D sub-sonic airfoil simulation using the κ - ε model. RPM accelerated the test-cases with a factor between 2 and 5.

In multi-scale problems, it is often of interest to model the macro-scale behavior, still retaining the essential features of the full systems. The ``coarse time stepper'' is a heuristic approach for circumventing the analytical derivation of models. The system studied here is a linear lattice of non-linear reaction sites coupled by diffusion. After reformulation of the time-evolution equation as a fixed-point scheme, RPM coupled with arc-length continuation is used to calculate the bifurcation diagrams of the effective (but analytically unavailable) equation.

Within the frame-work of dual time-stepping, a common approach in unsteady CFD-simulation, RPM is used to accelerate the convergence. Two test-cases were investigated; the von Karman vortex-street behind a cylinder at Re=100, and the periodic shock oscillation of a symmetric airfoil at M ∞ = 0.76 with a Reynolds number Re=11 x 10^6.

It was believed that once a basis had been identified, it could be retained for several steps. The simulations usually showed that the basis could only be retained for one step.

The need for updating the basis motivates the use of Krylov methods. The most common method is the (Block-) Arnoldi algorithm. As the iteration proceeds, Krylov methods become increasingly expensive and restart is required. Two different restart algorithm were tested. The first is that of Lehoucq and Maschhoff, which uses a shifted QR iteration, the second is a block extension of the single-vector Arnoldi method due to Stewart. A flexible hybrid algorithm is derived combining the best features of the two.