In this proposal, I present a research plan for active exploration and modeling of temporal phenomena to support service robots in long-term deployment settings, such as space station assistive robots like Astrobee. Under this framework, an autonomous robot will plan data collection activities around user-scheduled tasks and routine behaviors, which will be used to construct contextual models of temporal phenomena informing successful task execution. Such modeling will be used, for example, to guide efficient search for objects in inventory management tasks, taking into account likely locations across time, as well as inform when sensor readings should be taken for monitoring system health. Over an extended deployment, the robot will plan data collection over insufficiently modeled contexts to learn an increasingly robust model over time. Additionally, information gathering activities will be scheduled alongside normal tasks in a manner that allows for maximal task efficiency.
More »Examples of potential applications include EVA robotic assistants and personal satellite assistants.
More »Organizations Performing Work | Role | Type | Location |
---|---|---|---|
Oregon State University | Lead Organization | Academia | Corvallis, Oregon |
Ames Research Center (ARC) | Supporting Organization | NASA Center | Moffett Field, California |