John L. Bresina

MAPGEN: mixed-initiative planning and scheduling for the Mars Exploration Rover mission

The Mars Exploration Rover mission is one of NASAs most ambitious science missions to date. Launched in the summer of 2003, each rover carries instruments for conducting remote and in site observations to elucidate the planets past climate, water activity, and habitability. Science is MERs primary driver, so making best use of the scientific instruments, within the available resources, is a crucial aspect of the mission. To address this criticality, the MER project team selected MAPGEN (mixed initiative activity plan generator) as an activity-planning tool. MAPGEN combines two existing systems, each with a strong heritage: the APGEN activity-planning tool from the Jet Propulsion Laboratory and the Europa planning and scheduling system from NASA Ames Research Center. We discuss the issues arising from combining these tools in this missions context. MAPGEN is the first AI-based system to control a space platform on another planets surface.

Total-order and partial-order planning: a comparative analysis

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.

Autonomous rovers for Mars exploration

The Pathfinder mission demonstrated the potential for robotic Mars exploration but at the same time indicated the need for more robust rover autonomy. Future planned missions call for long traverses over unknown terrain, robust navigation and instrument placement, and reliable operations for extended periods of time. Ultimately, missions may visit multiple science sites in a single day and perform opportunistic science data collection, as well as complex scouting, construction, and maintenance tasks in preparation for an eventual human presence. Significant advances in robust autonomous operations are needed to enable these types of missions. Towards this end, we have designed an on-board executive architecture that incorporates robust flexible operation, resource utilization, and failure recovery. In addition, we have designed ground tools to produce and refine contingent schedules that take advantage of the on-board architectures flexible execution characteristics. Together, the on-board executive and the ground tools constitute an integrated rover autonomy architecture.

Mission operations planning: beyond MAPGEN

The MAPGEN system was deployed in the Mars Exploration Rover mission as a mission-critical component of the ground operations system. MAPGEN, which was jointly developed by ARC and JPL, represents a successful mission infusion of planning technology. The MER mission has operated spectacularly for over two years now, and we have learned valuable lessons regarding application of mixed-initiative planning technology to mission operations. These lessons have spawned new research in mixed-initiative planning and have influenced the design of a new ground operations system, called ENSEMBLE, that is base-lined for the Phoenix and Mars Science Laboratory missions. This paper discusses some of the lessons learned from the MER mission infusion experience and presents a preliminary report on these subsequent developments

The entropy reduction engine: integrating planning, scheduling, and control

This paper describes the Entropy Reduction Engine, an architecture for the integration of planning, scheduling, and control. The architecture is motivated, presented, and analyzed in terms of its different components; namely, problem reduction, temporal projection, and situated control rule execution. Experience with this architecture has motivated the recent integration of learning, and this paper also describes the learning methods and their impact on architecture performance.

An information infrastructure for coordinating Earth science observations

Earth scientists require timely, coordinated access to remote sensing resources, either directly by requesting that the resource be targeted at a specific location, or indirectly through access to data that has been, or will be acquired and stored in data archives. The information infrastructure for effective coordinated observing does not currently exist. This paper describes a set of capabilities for enabling model-based observing, the idea of linking scheduling observation resources more directly to science goals. Model-based observing is realized in this paper by an approach based on concepts of distributed planning and scheduling. The problem raises challenging issues related to planning under uncertainty, monitoring and repair of plans, and reasoning about human objectives and preferences

Reactive, Integrated Systems Pose New Problems for Machine Learning

Publisher Summary Most research on machine learning and planning has involved performance systems based on classical problem-solving algorithms (for example, STRIPS-Iike planners). AI problem solving has taken various divergent roads from these classical roots; two common current trends are reactive systems embedded in an environment and integrated multicomponent architectures. As performance engines, these advanced systems give rise to new learning problems—both in the sense of new opportunities and new difficulties. This chapter discusses new problems for machine learning. Classical problem-solving systems are typically consisted of a single component with a limited range of objectives and capabilities. Some current research efforts adopt a more holistic, synergistic approach involving integrated architectures with a broader scope of objectives and capabilities. These architectures integrate multiple performance components or multiple styles of reasoning. New issues arise within the context of integrated architectures, which engender new requirements and opportunities for machine learning.

Planning for control

The problems that lie at the boundary of planning and control are studied from an artificial intelligence (AI) planning perspective. An attempt is made to extend the tools of traditional AI planning to handle more complex domains. The capabilities of current plan generation and execution systems are outlined. A sample planning and control problem is defined and then used to show where traditional AI planning fails. The failure of traditional planning is used to motivate a discussion on aspects of a new approach to the relationship between plan generation and plan execution.<<ETX>>

Flight Operations for the LCROSS Lunar Impactor Mission

The LCROSS (Lunar CRater Observation and Sensing Satellite) mission was conceived as a low-cost means of determining the nature of hydrogen concentrated at the polar regions of the moon. Co-manifested for launch with LRO (Lunar Reconnaissance Orbiter), LCROSS guided its spent Centaur upper stage into the Cabeus crater as a kinetic impactor, and observed the impact flash and resulting debris plume for signs of water and other compounds from a Shepherding Spacecraft. Led by NASA Ames Research Center, LCROSS flight operations spanned 112 days, from June 18 through October 9, 2009. This paper summarizes the experiences from the LCROSS flight, highlights the challenges faced during the mission, and examines the reasons for its ultimate success.

Discovering mathematical operator definitions

ABSTRACT In the context of IL, a discovery system for mathematics, we describe our implementation of a general method, Generate, Prune, and Prove (GPP), for the discovery of mathematical operator definitions. This discovery process is driven by the intended purpose of the created operator. The GPP method is general with respect to the operators definition language, the specific operators, and the specified purpose of the operator. We illustrate GPP with one of our case studies - the discovery of the definition of the multiplicative operator for complex numbers.