
110f1af5dcd11d2330766480f41a3a25.ppt
- Количество слайдов: 15
Mission Planning for Sensorwebs Steve Chien, JPL Steve. A. Chien@jpl. nasa. gov 818 -393 -5320 Presented by Yi Chao, JPL
Outline • Technologies for Planning & Prosecution – CASPER/ASPEN “state & resource based semi-automated and autonomous planning & scheduling” – Operational Earth Observing Sensorweb architecture and systems • Routes to OOI Availability
Ocean Sensing Systems Paradigm Shift Platform-centric Sensing Systems AOSN Net-centric, Distributed Autonomous Sensing Systems Uncertain, Unknown Environment Uncertain Communication Self-navigating Network No maps Cooperative Behavior Adaptive Behavior Acoustic sensing Uncertainty
ASPEN Mission Planning Technology • Domain Independent application framework for – Modeling states and resources – Tracking and analyzing conflicts/consistence in current and projected states and resources – Decision support and autonomous decisionmaking based on above assessments – Framework for application of a range of Artificial Intelligence and Operations Research problem solving techniques (including AI/OR solvers) – CASPER = embedded version of ASPEN engine
ASPEN Deployments • Deployed into a number of applications – Space Mission Operations • • MAMM, 3 CS, EO-1 (3+ yrs 24/7) , OE Dramatic cost, effort reductions (EO-1, OE) Increased reliability/robustness (MAMM, EO-1, OE) Increased flexibility (EO-1, OE) – Also deployed for campaigns to ground communication stations, rovers, sea surface (USSV ‘ 05), subsurface (in progress), aerial (UAVSAR in progress) Jet Propulsion Laboratory Artificial intelligence Group
Mission Planning Application to ORION CI • Mission planning capability – Enables specification of high level goals to be automatically translated into deconflicted, operational, executable plans for execution – Enables network and individual asset resources to be tracked and allocated according to prescribed policies – Enables autonomous response appropriately accounting for systems states, resources, exogenous events, and prioritizations Jet Propulsion Laboratory Artificial intelligence Group
Application to CI Enables scheduled coordination among multiple entities Enables single entity autonomy accounting for future state and resources Cooperative Behavior Adaptive Behavior Jet Propulsion Laboratory Artificial intelligence Group
Earth Observing Sensorweb Components Re-tasking In-situ sensors, satellites, UAV’s, airborne instruments Science Targets: Volcanoes, Wildfires, Floods, Ice/Snow Jet Propulsion Laboratory Artificial intelligence Group
Existing EOS Sensorweb • Today – 24/7 monitoring and control with scores of assets • E. g. , over 1700 remote sensing images acquired by EO-1 through autonomous sensorweb responses – Rapid response (slowest timescale is space element) – Hand-crafted integrated science modeling – Scientist defined campaigns and sensor linkages • Linkages – – Space – EO-1, MODIS, AVHRR, GOES, Quikscat Ground (MEVO, CVO, …) Air (ARC UAS Ikhana, UAVSAR) Sea (GSFC surface) Jet Propulsion Laboratory Artificial intelligence Group
Open Geospatial Consortium (OGC) – Sensorweb Enablement (SWE) • Ongoing Work – Developing SWE services for multiple assets • Sensor planning service (SPS), Sensor observation service (SOS), Sensor alert service (SAS), web processing services (WPS), web map services (WMS), … • Services developed for range of assets – space assets, ground networks, air assets – E. g. “alert me when Eo-1 acquired imagery of Mt St Helens indicates a 2 x increase in thermal activity over any 24 hour period” – E. g. , “whenever a cluster ot greater then 20 RSAM 5 x quakes hits in any 20 minute period” reconfigure to double sensing rate! • On-demand, automatic derived products based on – Support of OGC services for ORION nodes, networks, assets?
Sensorweb Map
Note: similarity to Orion CI architecture! Earth Observing Sensorweb Architecture Instruments Data Distributed or centralized Science Alerts Science Agents Fractal, optionally distributed C&C Architecture Actions: Active data acquisition, changes in data rate Science Models Science Campaign s Scientists Science Event Manager Distributed or centralized Requests Automated Planning & Execution Distributed or centralized
Ocean Science and Engineering Paradigm Shifts • Traditional exploration-driven science being replaced by demand-driven system-oriented research and development • Expeditionary ocean science platforms (ships) being replaced by semi-permanent ocean observation and prediction infrastructure – – Real-time, Interactive oceanography, event capture Long time series Integrated sensing and modeling Deployments duration 24/7/365 require automation! MIT Laboratory for Autonomous Marine Sensing Systems
Making available to CI • Open source provision of toolkits – ASPEN • Demonstration adaptations for – AUV, Glider, Fixed network(s) (virtual field) • Scenario documents • Coordination with coastal, global, regional observatories MIT Laboratory for Autonomous Marine Sensing Systems
Questions? Steve Chien, JPL Steve. A. Chien@jpl. nasa. gov 818 -393 -5320
110f1af5dcd11d2330766480f41a3a25.ppt