The panel will discuss diverse topics and issues related to the development of autonomous capabilities, their enabling technologies – including artificial intelligence (AI), and their application to aerospace and robotic systems more generally. Some specific topics include: - Constraints on currently fielded autonomous capabilities; - Roles for contemporary AI in design, development, and operations of robotic vehicle systems; - Limitations on the use of AI in safety-critical applications; - Establishing trust in AI methods; - Issues with the certification of autonomous capabilities using established methods and processes; and, - The evolutionary traits needed for AI methods to render trusted, acceptable judgment in the performance of complex missions by autonomous robotic systems; … to name a few. The audience will be afforded a live question-answer period at the conclusion of panel presentations.
Under the sponsorship of TARDEC, UTRC is developing 5–10 kW Solid Oxide Fuel Cell (SOFC) Auxiliary Power Units (APU) that will be capable of operating on JP-8 with a sulfur concentration of up to the specification’s upper limit of 3000 ppmw. These APUs will be sized to fit within the relatively tight space available on U.S. Army vehicles such as the Abrams, Bradley and Stryker. The objective of the base development program that commenced in August 2010 is a 1000 hour TRL-5 demonstration of an APU in an Abrams configuration by mid-2013. This SOFC system is expected to provide power to the 28 VDC vehicle bus at a net efficiency ≥35%. In addition, the noise level is anticipated to be far below that generated by combustion engine-based APU concepts. UTRC has completed the Preliminary Design of the system and has finalized the overall system configuration and the requirements for each of the components. During the Preliminary Design phase, evaluations of the performance of sub-scale prototypes of the desulfurizer, auto-thermal reformer, and stack were completed. In the ongoing Detailed Design phase of the program, which runs through January 2012, each of the full-scale major components will be fabricated and tested. INTRODUCTION A United Technologies Research Center (UTRC) led team is developing JP-8 fueled Solid Oxide Fuel Cell (SOFC) based vehicle Auxiliary Power Units (APUs). These systems are being designed to provide 28 VDC power efficiently (>35%) and quietly to vehicle electric loads when the main engine is off. The value proposition offered by the SOFC-APUs for the vehicle applications are twofold: 1) a decreased fuel burn relative to the provision of this power by the operation of the main engine in an inefficient near idle condition, and 2) the Proceedings of the 2011 Development of a 5 – 10 kW JP-8 Fueled Solid Oxide Fuel Cell Auxiliary Power Unit for Army Vehicle enablement of Silent Watch missions whose durations are limited only by the vehicle fuel supply. The high valuations placed upon fuel efficiency and silence for the envisioned vehicle application coupled with the requirement to provide the desired power fro a sulfur level as high as 3000 ppmw in a compact space represents an ideal scenario for the application of power dense mobile Solid Oxide Fuel Cell technology. Relative to the heat engine systems considered for the same applications, SOFCs offer large (>10%) advantages, and they are anticipated to offer appreciable acoustic benefits as well. On the other hand, relative to other fuel cell based technology, due to their high (~800°C) operating temperature, SOFC stacks can effectively on H2 and CO rich reformate streams and are more tolerant of reformate “impurities” (e.g. S) in sharp contrast to lower operating temperature stack technologies. A consequence of this increased tolerance, SOFC feature significantly more compact liquid hydrocarbon fuel processors than other technology options. However, while the high temperature operation of SOFCs is beneficial from a fuel processing standpoint, it presents an operational challenge in that the system must be b to this operating temperature before power generation commences. At present, this heat-up process is take approximately 30 minutes, and the UTRC working toward this target. SYSTEM CONCEPT UTRC has been working since 2006 under Research Lab (AFRL), TARDEC, and Defence Science and Technology Agency (DSTA) of Singapore sponsorship on the development of compact and lightweight SOFC systems Figure 1, for both Unmanned Aerial Vehicle (UAV) propulsion and vehicle auxiliary power applications that operate on low sulfur liquid hydrocarbon fuels S-8, Ultra-Low Sulfur Diesel (ULSD), and desulfurized JP 8. The major focus of the development effort has been to drive mass and volume from the SOFC systems while maintaining their efficiency advantage relative to the internal combustion engine competition for the applications. In the interest of minimizing system weight and volume at the cost of a reduced efficiency, these 1–2 kW scale systems have featured Catalytic Partial Oxidation (CPOX) fuel reformers. Under the sponsorship of the Office of the Secretary of Defense (OSD) Energy Security Task Force and with program management provided by the Office of Naval Research (ONR), UTRC migrated its lightweight SOFC system technology to 12 kW Marine Ground Generator applications, where operation on low-sulfur (< 400 ppm JP-8 and JP-5 is required, and the reduced importance of system weight in such ground applications enable incorporation of additional efficiency enhancing features Ground Vehicle Systems Engineering and Technology Symposium Page 2 of 3 m JP-8 with
Convolutional Neural Networks (CNNs) achieve state-of-the-art results in many application areas, including image classification. For some applications it would be useful but impractical to deploy them on mobile devices with limited memory and power. A currently active area of research is the compression of deep networks while maintaining accuracy, with the aim of reducing memory usage, energy consumption and processing time. Several network compression methods have been proposed and have achieved good results, but they usually require the specification of parameters and are computationally expensive. We propose a new fast automated method called Unsupervised PulseNet that uses unsupervised k-means clustering to detect clusters of similar filters, and nodes in fully-connected layers, and prunes those that are redundant. We evaluate it on the CIFAR10, CIFAR100 and Tiny-Imagenet datasets using Alexnet, VGG16 and a 2-layer CNN called CifarNet suggested by the Tensorflow group. Compared to other methods in the literature we achieve the greatest compression, in shorter times, and with negligible loss in classification accuracy. In particular, we reduced Alexnet down to less than 0.7% of its original size, while not losing more than 2% classification accuracy.
