Alan L. Jennings

Dynamic programming applied to UAV way point path planning in wind

Path planning for small unmanned air vehicles (UAVs) becomes a difficult problem when accounting for wind. Wind can affect the path quality in a nonlinear manner requiring extended segment lengths for accurate measurements. Dynamic programming offers an efficient solution to create way point paths justified when settling lengths are linearly interpolated. A database of good turn progressions supports this method and can be composed off-line. The heuristic inherent in dynamic programming provides near optimal paths in very fast computational time because it restricts the search to a small number of candidates. A solution can be guaranteed by including turns over 180deg.

Texture-Based Photogrammetry Accuracy on Curved Surfaces

Photogrammetry is a preferred technique for noncontact measurement of flexible structures, such as large membranes and flapping-wing vehicles. Traditionally, discrete features are triangulated between photos. New techniques use the local texture of a surface to match random speckle among images. Texture-based photogrammetry allows for higher resolution surfaces, and this paper tests if the precision is similar to traditional photogrammetry. Test surfaces provide known, rounded profiles used to compare surface reconstructions. Many images are taken of an object with a projected texture, and the depth error is used to quantify the accuracy of the results. Results on surfacemeshes from800 to 2500 points show accuracy on the order of 1:2000, or nearly that of one pixel. Camera locations had a surprisingly minor effect on surface quality. There is a slight correlation of more parallel views tomore points on the surface. To show that the technique extends to practical use, results are shown for aflapping cycle of amembranewing.Dense surfacemeshes are important for higherfidelitymodels in computational fluid dynamics and finite element analysis.

Optical flow background subtraction for real-time PTZ camera object tracking

The use of pan/tilt/zoom (PTZ) camera systems with Computer Vision (CV) techniques is a burgeoning field. Utility is most commonly seen with security systems, robotics, navigation and for capturing sports events. This paper seeks to expand the use of PTZs in the area of measurement; specifically, the real-time tracking and measurement of Nano/Micro Unmanned Aircraft Systems (UAS). Empirical methods for developing various Nano/Micro UASs, typically ornithopter-related, show possibilities, but require theoretical development for continued understanding and advancements. The study of Nano/Micro UAS state characteristics would enable empirical development by providing supplementary model information for use in finite element and computational fluid dynamics analyses. One such advancement is to develop a metrology system using CV tracking coupled with videogrammetry techniques. The focus of this work is to provide a unique method for obtaining high-resolution, high frame-rate images of a UAS. A novel approach using a Graphics Processing Unit (GPU)-based pyramidal implementation of the Lucas-Kanade feature tracker (i.e. optical flow) to subtract PTZ movement is used to obtain motion measurements for directing the PTZ cameras.

Unbounded Motion Optimization by Developmental Learning

An algorithm is presented for autonomous motion development with unbounded waveform resolution. Rather than a single optimization in a very large space, memory is built to support incremental improvements; therefore, complexity is balanced by experience. Analogously, human development manages complexity by limiting it during initial learning stages. Motions are represented by cubic spline interpolation; therefore, the development technique applies broadly to function optimization. Adding a node to the splines allows all previous memory samples to transfer to the higher dimension space exactly. The memory-based model, which is a locally weighted regression (LWR), predicts the expected outcome for a motion and provides gradient information for optimizing the motion. Results are compared against bootstrapping a direct optimization (DO) on a mathematical problem. Additionally, the method has been implemented to learn voltage profiles with the lowest peak current for starting a motor. This method shows practical accuracy and scalability.

Pan-Tilt-Zoom Hybrid Camera System For Dynamic Tracking and Measurement

Noncontact measurements of lightweight flexible aerospace structures present several challenges. Objects are usually mounted on a test stand, because current noncontact measurement techniques require that the net motion of the object be zero.However, it is often desirable to takemeasurements of the object under operational conditions, and in the case of miniature aerial vehicles and deploying space structures, the test article will undergo significant translational motion. This paper describes a hybrid noncontact metrology system that will enable measurement of structural kinematics of anobject freelymovingabout a volume.Byusing a real-timevideogrammetry system, a set of pan–tilt–zoomcameras is coordinated to track large-scalenetmotionandproducehigh-speedhigh-quality images for photogrammetric surface reconstruction. The design of the system is presented in detail. Amethod of generating the calibration parameters for the pan–tilt–zoom cameras using curve fits is presented, and it is shown to produce results approximately half the precision of optimized calibration. It is shown how object size and speed determines tracking accuracy and object resolution. Finally, an example of surface reconstruction of a moving object is presented.

Optimal Inverse Functions Created via Population-Based Optimization

Finding optimal inputs for a multiple-input, single-output system is taxing for a system operator. Population-based optimization is used to create sets of functions that produce a locally optimal input based on a desired output. An operator or higher level planner could use one of the functions in real time. For the optimization, each agent in the population uses the cost and output gradients to take steps lowering the cost while maintaining their current output. When an agent reaches an optimal input for its current output, additional agents are generated in the output gradient directions. The new agents then settle to the local optima for the new output values. The set of associated optimal points forms an inverse function, via spline interpolation, from a desired output to an optimal input. In this manner, multiple locally optimal functions can be created. These functions are naturally clustered in input and output spaces allowing for a continuous inverse function. The operator selects the best cluster over the anticipated range of desired outputs and adjusts the set point (desired output) while maintaining optimality. This reduces the demand from controlling multiple inputs, to controlling a single set point with no loss in performance. Results are demonstrated on a sample set of functions and on a robot control problem.

Empirically bounding of space booms with tape spring hinges

Self-deploying structures seek to provide a compact launch package for large, lightweight satellite booms. One self- deploying method is a foldable tape spring. This paper examines the large scale behavior of a boom attached by a tape spring hinge during mock deployments. A boom attached by tape spring to a rigid stand was released and the boom bounced up to 60 ○ before coming to rest (as opposed to snap-through behavior). These large amplitude bounces can cause the boom to collide with sensors, other booms or arrays causing damage or preventing full deployment. Results show the first bounce of deployment is nearly bounded by a four parameter ellipse. The ellipses of similar folds are similar also, suggesting that a model can be developed. Free-fall tests simulating the free-free condition found in microgravity also show similar elliptical motion. Envelopes that bound the extents of the boom motion allow for collisions to be prevented by adjustment of the design.

Periodic and Average Flapping Wing Force Measurement

Research into the flight of flapping winged vehicles (ornithopters) is still uncovering the basic aeroelastic interactions involved in flight at this scale. The complex relationships between structure and fluid are driven by vortex shedding and pushing air, rather than traditional aerodynamics. Passive membranes with synergistic structural stiffness are effective in creating billowing and producing lift. The periodic flapping of the wings involves periodic aerodynamics requiring different testing methods. This work evaluates the effect of wind speed and mount stiffness on performance of an ornithopter compared to in-flight conditions. A commercial-off-the-shelf flapping-wing vehicle is used since it is known to fly. In-flight conditions are ascertained via videogrammetry since it offers accurate, noncontact measurements of pose and speed. Results measure the nominal speed along with the periodic plunge, surge and pitch of the vehicle. Force testing is conducted at various air speeds to show that despite the low flying speed, some air flow is necessary to achieve the lift observed in flight.

Flapping Wing Force and Deformation Analysis

Research into the flight of miniature aerial vehicles (MAV) is still uncovering the basic aeroelastic interactions involved in flight at this scale. The complex relationships between structure and fluid are driven by vortex shedding and pushing air, rather than traditional aerodynamics. Passive membranes are effective in creating billowing and producing lift. A commercial flapping-wing vehicle is used since it is know to fly. This work fuses photogrammetry and force measurements with finite element analysis. Force measurements reveal that rigid body vibrations cannot be ignored and fixed mounts for tests may effect the aerodynamic response. High speed images of the flap cycle are used to recreate path of the beam tip and show the large billowing of the wing. Images are synced with six degree of freedom force and moment measurements. A finite element analysis model show how simple inertial loading can create billowing, but occurs out of phase with aerodynamic billowing. Suggestions for future testing are set forth to address assumptions of each testing method.

Simulation of Locking Space Truss Deployments

There is a growing need in the space industry for large antenna apertures that can be deployed from existing launch vehicle payload fairings in order to increase ground to space communications sensitivity. The solutions presented here involves a sparse antenna array made from self-deploying, folded box trusses which use locking joints and tension cables to create a parabolic reflector. This paper presents method for simulating a largely unconstrained deployment of a single box truss. Results of this simulation establish the framework for future work that will include an entire truss system of eight truss boxes that constitute an entire arm of the sparse antenna array.