Prev-Talk | Next-Talk | All-Talks | Talks-Sorted
Talk-06: Sampling-Based Motion Planning and Co-Safe LTL for Coverage Missions Using the MOOS-IvP Framework
James McMahon, US Naval Research Laboratory and
Erion Plaku, Catholic University of America
As the complexity and duration of autonomous underwater vehicle (AUV) missions grow the need for enhanced mission and motion-planning capabilities will increase. Currently, missions are given to AUVs in terms of navigating to a desired destination by following a set of predefined waypoints or covering an area of interest by following a specific motion pattern. This makes it difficult to adapt to unanticipated obstacles or changes that prevent the AUVs from following the prespecified plans. Even more problematically it prevents the AUV to venture into new areas in response to new information gathered during exploration may prevent the discovery of additional mines and other objects of interest.
In order to enhance mission and motion planning capabilites we proposea high-level formalism that makes it possible to specify missions in a structured language resembling English. As an illustration, the mission of inspecting several areas and responding to new information can be expressed as
“(always safe) and (eventually (inspect areas A1, A2, . . . , An and (if obstacle then avoid) and (if elevation change then adapt) and (if indication of mines then explore surrounding area) and (if indication of ship then track until numerous sensor readings from different angles have been taken)) followed by return to the base)” (1)
As missions are characterized by global and local events occurring across a time span, the proposed high-level formalism allows for the combination of these events with logical connectives (and, or, not, if/then) and temporal operators (always, eventually, until, followed by). The proposed high-level formalism is rooted in solid mathematical foundations based on well-established temporal logics, such as Linear Temporal Logic (LTL).
In our talk we will discuss our current progress in the development of mission and sampling-based motion planning software to support such high-level specifications. We will discuss our progress in integrating the planning software with the MOOS-IvP framework and discuss future work which will help improve MOOS-IvP's capabilities, such as creating a MOOS application to update bathymetric maps. Furthermore we will discuss our current plans on implementing the software on an Iver-2 AUV.
Categories:
- Motion Planning
- Autonomous Underwater Vehicles (AUVs)
- Ocean Server, Iver2 AUVs
- MOOS-IvP