This proposal seeks a project-based educational cooperative agreement between harvey mudd college (HMC) and the planning and scheduling group's (PSG) overlap with NASA's autonomous systems and…

This proposal seeks a project-based educational cooperative agreement between harvey mudd college (HMC) and the planning and scheduling group's (PSG) overlap with NASA's autonomous systems and operations project (ASO). The collaboration would support research into task-execution and task-replanning/rescheduling. Demanding a new balance between these classically-distinct capabilities are families of resource-limited reasoners, e.g. cubesats in low-earth orbit, next-generation planetary rovers, and robot manipulators with sliding-scale autonomy. Balancing between schedules' execution and cognition planning-and-scheduling executives span a design space with difficult tradeoffs. A key tension arises when uncertainty or unplanned events necessitate a schedule change. On one hand, full-featured replanning/rescheduling systems require time, memory, and communications not available in many systems. When full replanning is impossible or undesirable, a scheduling executive depends crucially on its integration with a rescheduling system. The specifics of individual missions demand expertise across this tradeoff space, prompting recent attention to the continuum between schedule execution and (RE)cognition (frank 2017). NASA ames research center s planning and scheduling group (PSG) has identified a workflow through which students and faculty at HMC will collaborate to deepen understanding of this design space: + scenario-development showing contexts for these (RE)scheduling/execution tradeoffs for specific space missions. + integration and performance analysis on existing systems, e.g., comparing plexil with ROS's smach as the executive, exploration of different rescheduling algorithms, etc. + integration/performance analysis of different schedulers and executives, e.g. schedulers based on simple temporal networks with uncertainty (stnus) as developed by the HMC team (lund ET AL. 2017) + developing and deploying (RE)scheduling-and-execution architectures on different processors (e.g., raspberry PI, arduino, or beaglebone), different battery/power capabilities, different sensors, and different robots, as proxies for space vehicles or missions. This project (1) will pursue this workflow and investigate these tradeoffs at HMC, collaboratively with DR. Frank at NASA ames, (2) will prototype a novel execution/rescheduling system -- on physical robots at HMC, and (3) will deliver the software, documentation, and reporting on that system. The collaboration and deliverables will improve and inform NASA's ability to design the resourceconstrained rescheduling executives required by next-generation of robot-, aero-, and space-systems. A clinical-practice collaboration with harvey mudd college in addition to this core research-and-prototyping collaboration, the proposed partnership with HMC's clinic program would further NASA's educational objectives. As the nation's first and longest-running clinical-practice engineering program, the HMC clinic provides tested-and-proven scaffolding for this collaboration. That scaffolding enables a faculty advisor (DR. JIM boerkoel), the clinic director (DR. Zachary dodds), and a team of 4-5 harvey mudd college students to collaborate on this open-ended project with partnering liaisons at NASA (DR. Jeremy frank). A cornerstone of HMC's academic approach, clinic places an external project into the college's schedule, regularly engages faculty and students with outside liaisons, and specifies deliverables. Because IT is a course for the faculty and a capstone required of all engineers and CS majors, clinic's timetable drives the college's academic rhythms and expectations. The clinic program has been widely adopted and adapted: the ascendence of its approach offers NASA and HMC the opportunity to serve NASA's research and educational goals, to reach a cohort of earnest, talented, and ambitious young scientists, and to grow capillaries into new approaches to academic collaboration.