Cancer patients’ activity level and performance is difficult to assess outside the clinical setting. Patients are often not able to communicate subtleties in activity performance that may indicate secondary concerns or true pain levels. There is a need for doctors to monitor patient activity performance at home, especially to identify performance anomalies that may require the doctor’s intervention. Many black box activity classification algorithms lack the specificity and succinctness required to alert doctors of degrading health issues, such as suddenly requiring a cane to walk. Additionally, traditional black box classification systems lack the ability to deliver personalized information on activity performance. For instance, in the previous example a doctor does not need to know if a patient who already uses a cane is using a cane during their exercises. By combining deep learning models and symbolic reasoning we have created the Characterizing Human Activities for Cancer Health Awareness (CHA-CHA) system to classify exercises performed at home and alert the doctor of patient specific anomalies in their performance.

Auto battlers are a recent genre of online deck-building games where players choose and arrange cards that then compete against other players' cards in fully-automated battles. As in other deck-building games, such as trading card games, designers must balance the cards to permit a wide variety of competitive strategies. We present Ludus, a framework that combines automated playtesting with global search to optimize parameters for each card that will assist designers in balancing new content. We develop a sampling-based approximation to reduce the playtesting needed during optimization. To guide the global search, we define metrics characterizing the health of the metagame and explore their impacts on the results of the optimization process. Our research focuses on an auto battler game we designed for AI research, but our approach is applicable to other auto battler games.

Robots that interact with humans in a physical space or application need to think about the person's posture, which typically comes from visual sensors like cameras and infra-red. Artificial intelligence and machine learning algorithms use information from these sensors either directly or after some level of symbolic abstraction, and the latter usually partitions the range of observed values to discretize the continuous signal data. Although these representations have been effective in a variety of algorithms with respect to accuracy and task completion, the underlying models are rarely interpretable, which also makes their outputs more difficult to explain to people who request them. Instead of focusing on the possible sensor values that are familiar to a machine, we introduce a qualitative spatial reasoning approach that describes the human posture in terms that are more familiar to people. This paper explores the derivation of our symbolic representation at two levels of detail and its preliminary use as features for interpretable activity recognition.

Gin Rummy is a popular two-player card game involving choices to draw and discard cards to form sets of matching cards. Unlike other popular games such as Chess, Poker, and Go, there is little formal artificial intelligence research about how to make good decisions when playing Gin Rummy. In this paper, we develop an agent that plays Gin Rummy through a combination of known and expected card values, modeling the opponent to predict their cards of interest, and a conservative approach to assessing when to end the hand. In addition to discussing our observations about Gin Rummy that inspired our agent's design and how the agent works, we evaluate the relative importance of various features employed by our agent by competing agents which implement various subsets of those features.

This paper presents a methodology to combine a rigid-body and non-linear aerodynamic model of a hypersonic entry vehicle and provide bounds on the entry conditions required to ensure the vehicle maintains its attitude within prescribed limits. The non-linear aerodynamic model is based on parametric fits of the vehicle’s aerodynamic coefficients found using the computation tools US3D and MGDS. The vehicle’s equations of motion are separated into a nominal, linear time-invariant portion and a sector-bounded static memoryless non-linearity. Properties of the non-linearity are exploited to quantify the vehicle’s dynamic stability using integral quadratic constraint theory. Numerical results demonstrate that the non-linear aerodynamic model can be accounted for when assessing dynamic stability.

Use of Unmanned Aerial Systems (UASs) has grown steadily since the 1990s. They are becoming a standard tool for the military and are growing more commonplace in civilian applications. UASs of a vast array of sizes, capabilities and functions, are increasingly present in the National Airspace (NAS) and pressure to normalize their operations is taking on increasing urgency. The human interface for such systems will remain a critical component-and may increase in importance as human interactions are increasingly distanced, in time and/or space, from the locus of execution. Of particular complexity and concern are the use of larger UASs in dense traffic areas (such as cities) under high autonomy and with multiple vehicles per human operator. This paper summarizes a recently-completed review of multiple UAS operational concepts and evaluation of a suite of user interaction approaches for them. We have paid special attention to enabling ongoing situation awareness and prompt context switching. We have extensively leveraged our prior work on "Playbooks" for human tasking and delegation to automation. We conclude by presenting three novel human-machine interaction tools or approaches for such multi-UAS, high complexity, "routine" operations in the NAS: (1) a "strip-chart" view for combining vehicle-centric and task-or function-centric information in a timeline view, (2) an Activity Interchange display for facilitating task and vehicle handoffs within a defined task or play grammar, and (3) an approach applying plan recognition to verbal interactions to enable more intelligent and context-aware speech interactions for UAS control to speed and reduce workload in operator, Air Traffic Control and UAS interactions.

This chapter will discuss our approach to enabling effective human control over artificial swarms. This chapter will first give an overview of the Playbook architecture and prior work utilizing it in Section 2, including multiple programs for the Department of Defense. We will then provide a brief summary of prior work on artificial swarms, with a special focus on user interaction and interfaces for controlling swarms in Section 3. Section 4 will comprise the bulk of the chapter and will share how we've adapted Playbook for use with swarms of unmanned vehicles, as opposed to individual assets or smaller multi-agent teams, through the ISHIS project (Interface System for Human Interaction with Swarms). Finally, we will conclude with a summary and discussion of potential avenues for further improvement and expansion of Playbook for swarm control in Section 5.

The U.S. military is researching capabilities for collaborative and highly autonomous unmanned aircraft systems to conduct missions in denied or contested airspace, where multiple teams of aircraft would work together under the supervision of a single operator. One of the key challenges in this paradigm is the need for an effective human-system interface (HSI): one that could provide sufficiently detailed command and control authority and monitoring of plan execution for the entire system without overwhelming the operator. Another interesting challenge is the need to proactively find and approve multiple contingency plans, effectively granting the highly autonomous agents more freedom to operate when communication to the human supervisor is denied. This paper describes our recent efforts to design and develop a combined human-system interface and contingency planning tool that addresses these challenges. We present the design of our SuperC3DE system, provide examples of how certain interface components and contingency planning methods would be used in a notional scenario, and discuss key takeaways from formative evaluations by subject matter experts.

Formal methods tools, whose underlying principles are based on mathematics and formal logic, are considered one of the most effective and rigorous means of verifying system properties and assuring the absence of undesirable system behavior. The use of such tools seem to squarely fit the needs of those aiming to develop and certify avionic software as per the DO-178C standard, the set of objectives laid out by FAA to achieve a high level of confidence on the systems. However, our recent work on a NASA-funded research project revealed that there are practical considerations and additional complexities involved in using formal method tools to provide the level of assurance as exemplified by the DO-178C. In this paper we discuss one of the key concerns with formal tools: its soundness — the characteristic of a tool to never permit the verified system property be declared true when it is actually not true. We explored two major classes of formal methods tools — namely model checkers and static analyzers — and observed several threats to their soundness such as tool fallacies and failure modes that could lead to misplaced confidence in the verified system. We present various strategies to mitigate them, including an assurance case framework to verify that potential risks are all mitigated. The intent of this paper is not to discourage but encourage scrupulous use of formal tools to certify critical avionic software by being wary of the subtle but serious issues that may be overlooked.

Advanced capabilities planned for the next generation of unmanned aircraft will be based on complex new algorithms and non-traditional software elements. These aircraft will incorporate adaptive and intelligent control algorithms that will provide enhanced safety, autonomy, and high-level decision-making functions normally performed by human pilots, as well as robustness in the presence of failures and adverse flight conditions. This paper discusses the characteristics of adaptive algorithms and the challenges they present to certification for operation in the National Airspace System (NAS). We provide mitigation strategies that may make it possible to overcome these challenges.

This paper describes an automated process of active perception for cyber defense. Our approach is informed by theoretical ideas from decision theory and recent research results in neuroscience. Our cognitive agent allocates computational and sensing resources to (approximately) optimize its Value of Information. To do this, it draws on models to direct sensors towards phenomena of greatest interest to inform decisions about cyber defense actions. By identifying critical network assets, the organization's mission measures interest (and value of information). This model enables the system to follow leads from inexpensive, inaccurate alerts with targeted use of expensive, accurate sensors. This allows the deployment of sensors to build structured interpretations of situations. From these, an organization can meet mission-centered decision-making requirements with calibrated responses proportional to the likelihood of true detection and degree of threat.

The Direct Fusion Drive (DFD), a compact, anuetronic fusion engine, will enable more challenging exploration missions in the solar system. The engine proposed here uses a deuterium–helium-3 reaction to produce fusion energy by employing a novel field-reversed configuration (FRC) for magnetic confinement. The FRC has a simple linear solenoid coil geometry yet generates higher plasma pressure, hence higher fusion power density, for a given magnetic field strength than other magnetic-confinement plasma devices. Waste heat generated from the plasma׳s Bremsstrahlung and synchrotron radiation is recycled to maintain the fusion temperature. The charged reaction products, augmented by additional propellant, are exhausted through a magnetic nozzle. A 1MW DFD is presented in the context of a mission to deploy the James Webb Space Telescope (6200kg) from GPS orbit to a Sun–Earth L2 halo orbit in 37 days using just 353kg of propellant and about half a kilogram of 3He. The engine is designed to produce 40N of thrust with an exhaust velocity of 56.5km/s and has a specific power of 0.18kW/kg.

Traditional thinking presumes that planetary defense is only achievable with years of warning, so that a mission can launch in time to deflect the asteroid. However, the danger posed by hard-to-detect small meteors was dramatically demonstrated by the 20 meter Chelyabinsk meteor that exploded over Russia in 2013 with the force of 440 kilotons of TNT. A 5-MW Direct Fusion Drive (DFD) engine allows rockets to rapidly reach threatening asteroids that and deflect them using clean burning deuterium-helium-3 fuel, a compact field-reversed configuration, and heated by odd-parity rotating magnetic fields. This paper presents the latest development of the DFD rocket engine based on recent simulations and experiments as well as calculations for an asteroid deflection envelope and a mission plan to deflect an Apophis-type asteroid given just one year of warning. Armed with this new defense technology, we will finally have achieved a vital capability for ensuring our planet’s safety from impact threats.

Space debris is a growing concern for the sustained operation of our satellites. The population in space is continually increasing, both on a gradual basis as new satellites are placed on orbit and in sudden bursts, as evidenced with the recent collision between the Iridium and inactive Cosmos spacecraft. The problem is most severe in densely populated orbit regimes, where many operational satellites face a sustained presence of close-orbiting objects. In general, the frequent occurrence of potential collisions with debris will have a negative impact on mission performance in two important ways. First, repeated avoidance maneuvers diminish fuel and thus reduce mission life. Second, excursions from the nominal orbit during avoidance maneuvers may violate mission requirements or payload constraints. It is therefore important to consider both fuel minimization and station-keeping objectives in the avoidance planning problem. In this paper, we formulate the avoidance maneuver planning problem as a linear program. Avoidance constraints and orbit station-keeping constraints are expressed as linear functions of the control input. The relative orbit dynamics are modeled as a discrete, linear time-varying system that models both circular and eccentric orbits. The original nonlinear, nonconvex avoidance constraints are transformed into a time-varying sequence of linear constraints, and the navigation uncertainty is applied in a worst-case sense. Finally, the minimum-fuel avoidance maneuver problem is formulated with station-keeping constraints in a way that enables automatic relaxation of certain constraints to ensure feasibility.

The Integrated Communications and Optical Navigation System (ICONS) is a flexible navigation system for spacecraft that does not require global positioning system (GPS) measurements. The navigation solution is computed using an Unscented Kalman Filter (UKF) that can accept any combination of range, range-rate, planet chord width, landmark, and angle measurements using any celestial object. Both absolute and relative orbit determination is supported. The UKF employs a full nonlinear dynamical model of the orbit including gravity models and disturbance models. The ICONS package also includes attitude determination algorithms using the UKF algorithm with the Inertial Measurement Unit (IMU). The IMU is used as the dynamical base for the attitude determination algorithms. This makes the sensor a more capable plug-in replacement for a star tracker, thus reducing the integration and test cost of adding this sensor to a spacecraft. Recent additions include an integrated optical communications system which adds communications, and integrated range and range rate measurement and timing. The paper includes test results from trajectories based on the NASA New Horizons spacecraft.

This paper presents the overall design of a small reusable spacecraft capable of flying to an asteroid from low earth orbit, operating near the surface of the asteroid and returning samples to low earth orbit. The spacecraft is in a 6U CubeSat form factor and designed to visit near asteroids as far as 1.3 AU from the sun. Deep space missions are traditionally large and expensive, requiring considerable manpower for operations, use of the Deep Space network for navigation, and costly but slow rad-hard electronics. Several new technologies make this mission possible and affordable in such a small form factor: a 3 cm ion engine from Busek for the low-thrust spirals, an autonomous optical navigation system, precision miniature reaction wheels, high performance and nontoxic green propellant (HGPG) thrusters, and Honeywell’s new Dependable Multiprocessor technology for radiation tolerance. A complete spacecraft design is considered and the paper includes details of the control and guidance algorithms. Simulation results are presented for an example mission to Apophis, including: the earth outward spiral, sun-centered spiral to rendezvous, and proximity operations with the asteroid.

and horizontal landing allows for safe, rapid access to space. The end-to-end system design and preliminary trajectory optimization is presented. The feasibility of the concept is then demonstrated through a high-delity simulation.