Performance of a cognitive personal assistant, RADAR, consisting of multiple machine learning components, natural language processing, and optimization was examined with a test explicitly developed to measure the impact of integrated machine learning when used by a human user in a real world setting. Three conditions (conventional tools, Radar without learning, and Radar with learning) were evaluated in a large-scale, between-subjects study. The study revealed that integrated machine learning does produce a positive impact on overall performance. This paper also discusses how specific machine learning components contributed to human-system performance.

To date, manned space flight has maintained the locus of control for the mission on the ground. Mission control performs tasks such as activity planning, system health management, resource allocation, and astronaut health monitoring. Future exploration missions require the locus of control to shift to on-board due light speed constraints and potential loss of communication. The lunar campaign must begin to utilize a shared control approach to validate and understand the limitations of the technology allowing astronauts to oversee and direct aspects of operation that require timely decision making. Crew-centered Operations require a system-level approach that integrates multiple technologies together to allow a crew-prime concept of operations. This paper will provide an overview of the driving mission requirements, highlighting the limitations of existing approaches to mission operations and identifying the critical technologies necessary to enable a crew-centered mode of operations. The paper will focus on the requirements, trade spaces, and concepts for fulfillment of this capability. The paper will provide a broad overview of relevant technologies including: Activity Planning and Scheduling; System Monitoring; Repair and Recovery; Crew Work Practices.

This extended abstract describes efforts to infuse new planning and scheduling technologies into the Mars Science Laboratory Mission (MSL), a NASA Mars rover mission planned for launch in 2009. Beginning in 2003, we engaged the MSL mission and the developers of the Mission Data System (MDS). MDS is a software system that at the time was the MSL software baseline for both the ground and flight system. We briefly describe the tools that we integrated with MDS, the analysis or experience on previous missions that suggested each tool, and our successes in integrating these tools into a proof-of-concept uplink system we demonstrated in late 2004. In 2004, MSL decided to fall back on the very successful Mars Exploration Rover (MER) mission’s software in order to save development cost, which has resulted in some re-direction of our on-going activities. We briefly describe our new work to enhance the MER-based MSL software.

In this paper, we present a fully operational system that performs closed-loop observation scheduling and execution and has been in daily use on an automatic photoelectric telescope at the Fairborn Observatory on Mt. Hopkins in Arizona. The scheduler is part of the Associate Principal Astronomer system that also provides tools that enable the effective management of automatic telescopes. The paper describes the flexible schedule representation used and describes how the uncertainties and errors inherent in the domain are handled.

This paper presents the Expected Solution Quality (ESQ) method for statistically characterizing scheduling problems and the performance of schedulers. The ESQ method is demonstrated by applying it to a practical telescope scheduling problem. The method addresses the important and difficult issue of how to meaningfully evaluate the performance of a scheduler on a constrained optimization problem for which an optimal solution is not known. At the heart of ESQ is a Monte Carlo algor ithm that estimates a problem's probability density function with respect to solution quality This "quality density function" provides a useful characterization of a scheduling problem, and it also provides a background against which scheduler performance can be meaningfully evaluated. ESQ provides a unitless measure that combines both schedule quality and the amount of time to generate a schedule.

It is commonly acknowledged that there is a tradeoff between schedule quality and schedule robustness. In general terms, schedules that are of high quality tend to be of low robustness, and schedules that are robust tend to be of low quality. To better manage the robustness/quality tradeoff, we have developed an algorithm that implements what we call Just-In-Case scheduling; this algorithm explicitly considers the way in which scheduled actions might fail and how such failures can impact the executability of a schedule. Just-In-Case scheduling is able to build schedules that are robust and of high quality. The Just-In-Case algorithm is motivated in this paper by a specific telescope scheduling problem, and the paper presents the results of an experiment, carried out using real telescope scheduling data, that illustrates the performance improvement one can expect from using it.

For many years, the intuitions underlying partial-order planning were largely taken for granted. Only in the past few years has there been renewed interest in the fundamental principles underlying this paradigm. In this paper, we present a rigorous comparative analysis of partial-order and total-order planning by focusing on two specific planners that can be directly compared. We show that there are some subtle assumptions that underly the wide-spread intuitions regarding the supposed efficiency of partial-order planning. For instance, the superiority of partial-order planning can depend critically upon the search strategy and the structure of the search space. Understanding the underlying assumptions is crucial for constructing efficient planners.

This paper presents a technique for building robust telescope schedules that tend not to break. The technique is called Just-In-Case (JIC) scheduling and it implements the common sense idea of being prepared for likely errors, just in case they should occur. The JIC algorithm analyzes a given schedule, determines where it is likely to break, reinvokes a scheduler to generate a contingent schedule for each highly probable break case, and produces a 'multiply contingent' schedule. The technique was developed for an automatic telescope scheduling problem, and the paper presents empirical results showing that Just-In-Case scheduling performs extremely well for this problem.

Some applications involve automatic generation and execution of schedules that contain actions with uncertain durations. Such uncertainty can cause schedules to break during execution. This paper presents a technique, called Just-In-Case scheduling or JIC, for building robust schedules that tend not to break. The technique implements the common sense idea of being prepared for likely errors, just in case they should occur. The JIC algorithm analyzes a given schedule, determines where it is likely to break, reinvokes the scheduler to generate a schedule for each highly probable break case, and produces a multiply contingent schedule. The technique was developed for a real telescope scheduling problem, and the paper presents empirical results showing that Just-In-Case scheduling performs extremely well for this problem.

The ultimate imaging resolution in the UV and photometric precision achievable with a small (less than 1-meter) telescope located on the Moon is considered. The imaging resolution and photometric precision that might be practically achieved when the effects of the Lunar environment and equipment limitations are accounted for is then suggested. Finally, the practicality of soft landing such a telescope on the moon is considered, along with suggestions of how it might be directly controlled by using astronomers without any significant permanent staff.

This paper outlines a new telescope operations model that is intended to achieve low operating costs with high operating efficiency and high scientific productivity. The model is based on the existing Principal Astronomer approach used in conjunction with ATIS, a language for commanding remotely located automatic telescopes. This paper introduces the notion of an Associate Principal Astronomer, or APA. At the heart of the APA is automatic observation loading and scheduling software, and it is this software that is expected to help achieve efficient and productive telescope operations. The purpose of the APA system is to make it possible for astronomers to submit observation requests to and obtain resulting data from remote automatic telescopes, via the Internet, in a highly-automated way that minimizes human interaction with the system and maximizes the scientific return from observing time.

This paper presents a technique for statistically characterizing a search space and demonstrates the use of this technique within a practical telescope scheduling application. The characterization provides the following: (i) an estimate of the search space size, (ii) a scaling technique for multi-attribute objective functions and search heuristics, (iii) a quality density function for schedules in a search space, (iv) a measure of a scheduler's performance, and (v) support for constructing and tuning search heuristics. This paper describes the random sampling algorithm used to construct this characterization and explains how it can be used to produce this information. As an example, we include a comparative analysis of an heuristic dispatch scheduler and a look-ahead scheduler that performs greedy search.

This paper describes our project on advanced planning and scheduling for remote, fully automatic telescopes. In addition to a more advanced automatic scheduler for nightly observations, our project is also building automated tools to address the entire life-cycle of an observation request, from original receipt to the return of raw data and preliminary data reduction. Our focus is on providing software tools to help a telescope manager who represents a community of participating astronomers; however, the increased automation also improves the way in which the astronomers interact with this manager. Our goal is to make it possible for participating astronomers to submit observation requests and obtain results from a remotely located telescope, via electronic networks, without the necessity of human intervention.

This paper presents an algorithm, called Just-In-Case Scheduling, for building robust schedules that tend not to break. The algorithm implements the common sense idea of being prepared for likely errors, just in case they should occur. The Just-In-Case algorithm analyzes a given nominal schedule, determines the most likely break, and reinvokes a scheduler to generate a schedule to cover that break. After a number of iterations, the Just-In-Case algorithm produces a multiply contingent schedule that is more robust than the original nominal schedule. The algorithm has been developed for a real telescope scheduling domain in order to proactively manage schedule breaks that are due to an inherent uncertainty in observation durations. The paper presents empirical results showing that the algorithm performs extremely well on a representative problem from this domain.

NASA TileWorld (NTW) computer program formulated to further research on planning, scheduling, and control problems. Designed to focus on three particular attributes of real-world problems: exogenous events, uncertain outcomes of actions, and metric time. Written specifically for use by NASA, NTW modified easily to act as software base for other simulated environments. Written in Allegro Common Lisp for Sun-3-(TM) and Sun-4-series(TM) computers running SunOS(TM).