-TEMPEST is a planner for long-range planetary navigation that bridges the gap between path planning and classical planning and scheduling. In addition to planning routes, our approach yields the timing and placement of actions to conserve and restore expendable resources and that abide by operational constraints. TEMPEST calls upon the Incremental Search Engine (ISE) to enable heuristic path planning and efficient re-planning under global constraints, over a four dimensional state space. We describe our approach, then demonstrate how the planner operates in a simulated Mars science traverse. Following a brief summary of TEMPEST results from a recent rover field experiment, we evaluate our research progress and describe our current and future work.
INTRODUCTION 1The NASA Mars Technology Program (MTP) [5] currently funds a wide range of technology projects in support of the Mobile Science Laboratory (MSL) mission concept, under investigation for the 2009 launch opportunity [35]. Under this umbrella, the MTP supports a number competitivelyselected robotics research tasks. In contrast with more general research calls, the MTP selects technologies based on specific MSL mission needs, and requires that they meet basic criteria for developmental maturity prior to selection. To ease software evaluation and possible inclusion in future missions, participating teams are expected to integrate their software within a new NASA software architecture, CLARAty (Coupled Layer Architecture for Robotic Autonomy) [34], by the conclusion of the project. Participants in the program include teams at NASA/JPL, NASA Field Centers and universities.As one of the MTP participants, our team is developing software for autonomous global path planning, combining longrange route planning, resource management and enforcement of constraints. Several MSL design features motivate our work. To enable the investigation of specific surface targets, the MSL mission seeks long-range mobility to overcome worst-case landing errors. Furthermore, MSL may involve travel to several distinct sites, interleaving periods of dedicated, localized science data collection with periods of traverse and opportunistic science. To alleviate the power constraints of previous missions, the MSL rover may be powered by RTG (Radioisotope Thermo-electric Generator), and possibly supplemented by solar arrays. However, the power sources may not be sufficient to supply locomotion for long time periods. Batteries will store energy for use in high-power activities. Finally, current plans dictate a primary mission lasting up to 500 sols, entailing daily and seasonal lighting changes. All these factors motivate global path planning.Our work in this area began under, and is supplemented by, other NASA-funded projects, namely the Sun-Synchronous Navigation project [36][37][31] and the Life in the Atacama project [38][32]. In this paper, we describe our progress to date, as a result of all these efforts, but with an emphasis on our work directed towards the MTP. We begin by fu...