Multi-objective optimization methods can be criticized for lacking a statistically valid measure of the quality and representativeness of a solution. This stance is especially relevant to metaheuristic optimization approaches but can also apply to other methods that typically might only report a small representative subset of a Pareto frontier. Here we present a method to address this deficiency based on random sampling of a solution space to determine, with a specified level of confidence, the fraction of the solution space that is surpassed by an optimization. The Superiority of Multi-Objective Optimization to Random Sampling, or SMORS method, can evaluate quality and representativeness using dominance or other measures, e.g., a spacing measure for high-dimensional spaces. SMORS has been tested in a combinatorial optimization context using a genetic algorithm but could be useful for other optimization methods.
Upcoming weapon programs require an aggressive increase in Application Specific Integrated Circuit (ASIC) production at Sandia National Laboratories (SNL). SNL has developed unique modeling and optimization tools that have been instrumental in improving ASIC production productivity and efficiency, identifying optimal operational and tactical execution plans under resource constraints, and providing confidence in successful mission execution. With ten products and unprecedented levels of demand, a single set of shared resources, highly variable processes, and the need for external supplier task synchronization, scheduling is an integral part of successful manufacturing. The scheduler uses an iterative multi-objective genetic algorithm and a multi-dimensional performance evaluator. Schedule feasibility is assessed using a discrete event simulation (DES) that incorporates operational uncertainty, variability, and resource availability. The tools provide rapid scenario assessments and responses to variances in the operational environment, and have been used to inform major equipment investments and workforce planning decisions in multiple SNL facilities.
This report summarizes the analytical and experimental efforts for the Laboratory Directed Research and Development (LDRD) project entitled ''Advancements in Sensing and Perception using Structured Lighting Techniques''. There is an ever-increasing need for robust, autonomous ground vehicles for counterterrorism and defense missions. Although there has been nearly 30 years of government-sponsored research, it is undisputed that significant advancements in sensing and perception are necessary. We developed an innovative, advanced sensing technology for national security missions serving the Department of Energy, the Department of Defense, and other government agencies. The principal goal of this project was to develop an eye-safe, robust, low-cost, lightweight, 3D structured lighting sensor for use in broad daylight outdoor applications. The market for this technology is wide open due to the unavailability of such a sensor. Currently available laser scanners are slow, bulky and heavy, expensive, fragile, short-range, sensitive to vibration (highly problematic for moving platforms), and unreliable for outdoor use in bright sunlight conditions. Eye-safety issues are a primary concern for currently available laser-based sensors. Passive, stereo-imaging sensors are available for 3D sensing but suffer from several limitations : computationally intensive, require a lighted environment (natural or man-made light source), and don't work for many scenes or regions lacking texture or with ambiguous texture. Our approach leveraged from the advanced capabilities of modern CCD camera technology and Center 6600's expertise in 3D world modeling, mapping, and analysis, using structured lighting. We have a diverse customer base for indoor mapping applications and this research extends our current technology's lifecycle and opens a new market base for outdoor 3D mapping. Applications include precision mapping, autonomous navigation, dexterous manipulation, surveillance and reconnaissance, part inspection, geometric modeling, laser-based 3D volumetric imaging, simultaneous localization and mapping (SLAM), aiding first responders, and supporting soldiers with helmet-mounted LADAR for 3D mapping in urban-environment scenarios. The technology developed in this LDRD overcomes the limitations of current laser-based 3D sensors and contributes to the realization of intelligent machine systems reducing manpower need.
This report summarizes the analytical and experimental efforts for the Laboratory Directed Research and Development (LDRD) project entitled 'Obstacle Detection for Autonomous Navigation'. The principal goal of this project was to develop a mathematical framework for obstacle detection. The framework provides a basis for solutions to many complex obstacle detection problems critical to successful autonomous navigation. Another goal of this project was to characterize sensing requirements in terms of physical characteristics of obstacles, vehicles, and terrain. For example, a specific vehicle traveling at a specific velocity over a specific terrain requires a sensor with a certain range of detection, resolution, field-of-view, and sufficient sensitivity to specific obstacle characteristics. In some cases, combinations of sensors were required to distinguish between different hazardous obstacles and benign terrain. In our framework, the problem was posed as a multidimensional, multiple-hypothesis, pattern recognition problem. Features were extracted from selected sensors that allow hazardous obstacles to be distinguished from benign terrain and other types of obstacles. Another unique thrust of this project was to characterize different terrain classes with respect to both positive (e.g., rocks, trees, fences) and negative (e.g., holes, ditches, drop-offs) obstacles. The density of various hazards per square kilometer was statistically quantified for different terrain categories (e.g., high desert, ponderosa forest, and prairie). This quantification reflects the scale, or size, and mobility of different types of vehicles. The tradeoffs between obstacle detection, position location, path planning, and vehicle mobility capabilities were also to be characterized.