Jia Xu

Automated target recognition (ATR) is one of the most important potential military applications of the many recent advances in artificial intelligence and machine learning. A key obstacle to creating a successful ATR system with machine learning is the collection of high-quality labeled data sets. The authors investigated whether this obstacle could be sidestepped by training object-detection algorithms on data sets made up of high-resolution, realistic artificial images. The authors generated… Show more

Automated target recognition (ATR) is one of the most important potential military applications of the many recent advances in artificial intelligence and machine learning. A key obstacle to creating a successful ATR system with machine learning is the collection of high-quality labeled data sets. The authors investigated whether this obstacle could be sidestepped by training object-detection algorithms on data sets made up of high-resolution, realistic artificial images. The authors generated large quantities of artificial images of a high-mobility multipurpose wheeled vehicle (HMMWV) and investigated whether models trained on these images could then be used to successfully identify real images of HMMWVs. The authors obtained a clear negative result: Models trained on the artificial images performed very poorly on real images. However, they found that using the artificial images to supplement an existing data set of real images consistently results in a performance boost. Interestingly, the improvement was greatest when only a small number of real images was available. The authors suggest a novel method for boosting the performance of ATR systems in contexts where training data are scarce. Many organizations, including the U.S. government and military, are now interested in using synthetic or simulated data to improve machine learning models for a wide variety of tasks. One of the main motivations is that, in times of conflict, there may be a need to quickly create labeled data sets of adversaries' military assets in previously unencountered environments or contexts.

Show less

The authors of this report demonstrate a prototype of a proof-of-concept artificial intelligence (AI) system to help develop and evaluate new concepts of operations for the air domain. The prototype platform integrates open-source deep learning frameworks, contemporary algorithms, and the Advanced Framework for Simulation, Integration, and Modeling—a U.S. Department of Defense–standard combat simulation tool. The goal is to exploit AI systems' ability to learn through replay at scale… Show more

The authors of this report demonstrate a prototype of a proof-of-concept artificial intelligence (AI) system to help develop and evaluate new concepts of operations for the air domain. The prototype platform integrates open-source deep learning frameworks, contemporary algorithms, and the Advanced Framework for Simulation, Integration, and Modeling—a U.S. Department of Defense–standard combat simulation tool. The goal is to exploit AI systems' ability to learn through replay at scale, generalize from experience, and improve over repetitions to accelerate and enrich operational concept development. Show less

As the demand for commercial deliveries increases within cities, companies face a fundamental limitation in surface road capacity. Drone delivery aims to overcome that limitation by exploiting the vertical dimension above city streets. This report explores the vehicle design aspects of the delivery drone problem, including flight efficiency, energy consumption, noise, and safety, which are central to the viability of delivery drones. Importantly, such design aspects also speak to the potential… Show more

As the demand for commercial deliveries increases within cities, companies face a fundamental limitation in surface road capacity. Drone delivery aims to overcome that limitation by exploiting the vertical dimension above city streets. This report explores the vehicle design aspects of the delivery drone problem, including flight efficiency, energy consumption, noise, and safety, which are central to the viability of delivery drones. Importantly, such design aspects also speak to the potential scalability of the concept.

Show less

Although Unmanned Aerial Systems — or drones — were first employed on the battlefield, they are also being used in commercial applications, such as maintaining large buildings and cell phone towers, and they already deliver mail and medicine to hard-to-reach places. The reach of commercial drones is expanding, with the Federal Aviation Administration (FAA) estimating that nearly 3 million commercial drones will be flying by the year 2020. Google and Amazon are making big investments in drone… Show more

Although Unmanned Aerial Systems — or drones — were first employed on the battlefield, they are also being used in commercial applications, such as maintaining large buildings and cell phone towers, and they already deliver mail and medicine to hard-to-reach places. The reach of commercial drones is expanding, with the Federal Aviation Administration (FAA) estimating that nearly 3 million commercial drones will be flying by the year 2020. Google and Amazon are making big investments in drone technology. And start-ups such as Zipline, Flirty, and Matternet are already in the air. This video highlights issues explored in detail in a series of forthcoming RAND reports on the future of commercial delivery drones. These issues include technical matters — vehicle design, air traffic control, and energy consumption — as well as locations where drone use makes the most sense (e.g., in urban versus rural environments); their impact on emissions, traffic congestion, and noise; privacy, data, and liability concerns; and the need for new regulations to deal with all of the above.
Show less

NASA's Aeronautics Research Mission Directorate (ARMD) is working to expand flight research in order to advance the maturation and demonstrate the application of new aeronautics concepts and technologies over the next ten years. It asked RAND to assess available flight research capabilities and future needs, identify any gaps or excess infrastructure, and develop management options that would facilitate increased and improved flight research. We found that NASA has strong flight research… Show more

NASA's Aeronautics Research Mission Directorate (ARMD) is working to expand flight research in order to advance the maturation and demonstrate the application of new aeronautics concepts and technologies over the next ten years. It asked RAND to assess available flight research capabilities and future needs, identify any gaps or excess infrastructure, and develop management options that would facilitate increased and improved flight research. We found that NASA has strong flight research capabilities in most areas relevant for flight research. The few gaps that we identified could be filled through partnering or acquisition of vehicles from the marketplace when needed. Other gaps exist in sub- and full-scale experimental aircraft, but these cannot be acquired before the specific research projects are planned and funded. ARMD is already pursuing multiple efforts to increase flight research. We recommend that ARMD continue its efforts to enhance long-range planning and project funding certainty so that researchers can better include flight research in their plans and specific infrastructure needs can be identified further in advance. Cost-sharing through partnerships remains a valuable option, although industry positioning for increased intellectual property rights may be a limiting factor. Stewardship of flight research capabilities can be improved by instituting a unified, matrixed management structure across centers that can help align incentives while centralizing and improving utilization, partnering, and external outreach efforts. Finally, access and sharing barriers for researchers can be lowered through a voucher system for simple flight research efforts, streamlined processes for planning and access, and instituting state-of-the-art knowledge management approaches to store flight research data and share it with the aeronautics community.

Show less