Search

Faculty and researchers in the Department of Computer Science at Cornell University have been studying the challenges of information management and high-assurance computing for a variety of demanding settings, including cloud computing and data transmission in complex network environments. The main objective of the project titled Engaging Students via Innovative Militarily Useful Technologies was to expose students to the research and use of these technologies which have potential application in military scenarios and in nationally important critical infrastructure areas. By exposing students to this type of research, we can come closer to realizing the potential application of these tools, and the students themselves benefit from the experience in many ways. Historically, a significant percentage of Cornell students have pursued careers in the military, with military contractors, and in the civilian critical infrastructure areas. During the course of this program students were involved in four main areas: Scalable Landmark Recognition (ePaparazzi), the Software-defined Network Interface Card (SoNIC), Smart Grid Security (GridControl), and additional projects relating to specific aspects of the Isis2 software platform., The original document contains color images.

A hybrid fractal vision system is being developed for landmark detection and recognition in natural scenes. At the current quarter of research, a reconfigurable neural network is being designed to recognize landmarks. The fractal model detected the landmarks for cluttered images, and the neural network would recognize those landmarks. A brief description of the theoretical design of this Reconfigurable Neural Network is given here. Also, some of the initial results obtained by testing the neural network on real image data are included with this report. A new learning method is also being developed and briefly reported here. Automatic recognition systems can be useful in both military and commercial domains. Tasks such as military surveillance, automatic target recognition, automatic vehicle navigation, material handling, inspection, data compression/decompression, autonomous robot navigation, etc are some of the practical issues directly enhanced by automatic and robust vision systems.

Landmarks are useful cues for autonomous mobile robot navigation. Due to the changing imaging conditions, the appearance of the landmarks is varying in outdoor environments. We will develop a novel system to detect and recognize the landmarks. The developed system will be able to overcome changes in scale, lighting, etc. The proposed approach is based on a two-step method, using both fractal based object classifier and neural network based object identifier. Since fractals are inherently scale invariant over a finite range of scales, they make good models for outdoor scene objects. Since neural networks have fast recognition capabilities, they are a good choice in real time mobile robot applications.

The contract made significant progress across a broad front on the problems of computer vision for outdoor mobile robots. New algorithms were built in neural networks, range data analysis, object recognition and road finding. Perception modules were integrated into new systems (including on road and off road, notably on the new Navlab II vehicle); and there were notable programmatic events, ranging from generating two new thesis proposals to playing a major role in the 'Tiger Team', shaping the architecture for the new DARPA program in Unmanned Ground Vehicles. This report begins with a summary of the year's activities and accomplishments, in this chapter. Chapter 2, '3-D Landmark Recognition from Range Image', provides more detail on object recognition from multiple sensor locations. Chapter 3, 'Representation and Recovery of Road Geometry in YARF', discusses geometry issues in YARF, our symbolic road tracking system. The last two chapters discuss systems issues that are important in providing cues and constraints for an active vision approach to robot driving.' A Computational Model of Driving for Autonomous Vehicles', Chapter 4, introduces the complexities of reasoning for driving in traffic. The fifth and final chapter, 'Combining artificial neural networks and symbolic processing for autonomous robot guidance', shows how we combine neural nets with map data in a complete system.

Partial Contents: Vision and visual exploration for the Stanford Mobile robot; AuRA -- An architecture for vision-based robot navigation; Guiding an autonomous land vehicle using knowledge-based landmark recognition; The CMU navigational architecture; Qualitative navigation; Interpreting terrain using hierarchical symbolic grouping for multi-spectral images; Design of a prototype interactive cartographic display and analysis environment; Image understanding architecture; Constructs for cooperative image understanding environments; Interpreting aerial photographs by segmentation and search; Initial hypothesis formation in image understanding using an automatically generated knowledge base; Goal-directed control of low-level processes for image interpretation; Active vision; An integrated system that unifies multiple shape from texture algorithms; DRIVE - Dynamic Reasoning from Integrated Visual Evidence; What is a 'Degenerate' view?; Role and use of color in a general vision system; Using a color reflection model to separate highlights from object color; Recognizing unexpected objects -- A proposed approach; Parallel algorithms for computer vision on the connection machine; Solving the depth interpolation problem on a fine grained, mesh- and tree-connected SIMD machine; Survey of parallel computers; Evidence combination using likelihood generators; Multi-modal segmentation using Markov random fields; Minimization of the quantization error in camera calibration; Uncertainty analysis of image measurements; Results of motion estimation with more than two frames; Tracing finite motions without correspondence., See also Volume 1, ADA221427.