Search

This paper considers mission design strategies for mobile robots whose task is to perform spatial sampling of a static environmental field, in the framework of compressive sensing. According to this theory, we can reconstruct compressible fields using O(log n) nonadaptive measurements (where n is the number of sites of the spatial domain), in a basis that is "in coherent" to the representation basis [1]; random uncorrelated measurements satisfy this incoherence requirement. Because an autonomous vehicle is kinematically constrained and has finite energy and communication resources, it is an open question how to best design missions for CS reconstruction. We compare a two-dimensional random walk, a TSP approximation to pass through random points, and a randomized boustrophedon (lawnmower) strategy. Not unexpectedly, all three approaches can yield comparable reconstruction performance if the planning horizons are long enough; if planning occurs only over short time scales, the random walk will have an advantage.

A large portion of work on compressive sampling and sensing has focused on reconstructions from a given measurement set. When the individual samples are expensive and optional, as is the case with autonomous agents operating in a physical domain and under specific energy limits, the CS problem takes on a new aspect because the projection is column-sparse, and the number of samples is not necessarily large. As a result, random sampling may no longer be the best tactic. The underlying incoherence properties in l0 reconstruction, however, can still motivate the purposeful design of samples in planning for CS with one or more agents; we develop here a greedy and computationally tractable sampling rule that will improve errors relative to random points. Several example cases illustrate that the approach is effective and robust., United States. Office of Naval Research (Grant N00014-09-1-0700)

This presentation discusses challenges to the rapid execution of acquisition programs that would introduce affordable, effective, and adaptable systems into widespread use in a timely fashion. It suggests a revamping of the engineering process that would address these challenges, and describes the technical enablers that make that revamping feasible. Doing so will enable engineers and program decision makers to collaboratively focus on building the right things to provide utility in a wide range of joint operations, and across many potential alternative futures. Increased computational power and availability allow us to exploit data and apply services in much more flexible ways. This creates an opportunity to consider capabilities and mission utility more deeply, rather than getting locked into requirements and key performance parameters. That is a critical enabler to engineering for adaptability and maximizing the value of the system to the warfighter. We believe it is time to demonstrate a new engineering ecosystem that combines automated tools and stakeholder participation to generate more counters to potential surprise. Among the key contributing technology concepts are the following: (1) co-evolution of systems and missions via information sharing and decision aids, (2) option-preserving tradespace exploration, (3) analysis and evaluation with respect to life cycle issues, (4) informing requirements refinement, and (5) accelerated design and testing via rapidly composable modeling and analysis tools and risk-sensitive engineering planning aids. The presentation includes 9 briefing charts., Presented at the Naval Postgraduate School Annual Acquisition Research Symposium (9th) held in Monterey, CA on 16-17 May 2012. Published in the Proceedings of the Naval Postgraduate School Annual Acquisition Research Symposium (9th), v2 p3-12, 2012.

This thesis focuses on the development of infrastructure for research with large-scale autonomous marine vehicle fleets and the design of sampling trajectories for compressive sensing (CS). The newly developed infrastructure includes a bare-bones acoustic modem and two types of low cost and scalable vehicles. One vehicle is a holonomic raft designed for station keeping and precise maneuvering, and the other is streamlined kayak for travelling longer distances. The acoustic modem, like the vehicles, is inexpensive and scalable, providing the capability of a large-scale, low cost underwater acoustic network. With these vehicles and modems we utilize compressive sensing, a recently developed framework for sampling sparse signals that offers dramatic reductions in the number of samples required for high fidelity reconstruction of a field. Our novel CS sampling techniques introduce engineering constraints including movement and measurement costs to a better apply CS to sampling with mobile agents. The vehicles and modems, along with compressive sensing, strengthen the movement towards large scale autonomy in the ocean environment.

The primary objective of this project was to provide the Surrogate Semiautonomous Vehicle (SSV) with a Demonstration II capability of performing automatic target recognition/identification (ATR/I). Detected objects of interest would be imaged with FLIR and/or LADAR sensors so the ATR/I algorithms must be compatible with either sensor, as well as exploit the synergy of processing both sensors simultaneously. Our approach does not rely upon the precise coregistration of multiple sensors, but rather performs geometric hashing on the individual FLIR and LADAR images. Hashing represents an object by a collection of points, which are then matched to similarly constructed models. The matching is accomplished by iteratively selecting pairs of points, placing them in a Euclidean geometry coordinate system, concurrently translating and rotating all other object points to the same geometry, and then counting the number of occurrences of object and model points in the same cell. The geometric hashing software, originally developed for 2D SAR and FLIR imagery has been extended to also accommodate 3D Ladar range and intensity imagery. The 2D hashing software was modified to allow up to ten dimensions. Currently, a 4D scheme is being used which represents (x,y) position, range, and intensity features., Prepared in cooperation with Lockheed Martin Vought Systems, Dallas, TX, and New York Univ., NY.