Advances in Distributed Operations and Mission Activity Planning for Mars Surface Exploration Jason M. Fox Jet Propulsion ...
Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Opera...
Introduction <ul><li>The Mars Exploration Rover Mission (MER) demonstrated that a  centralized  mission activity planning ...
Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Opera...
 
SAP - The Legacy Tool <ul><li>The Science Activity Planner (SAP) was the primary science operations tool for MER </li></ul...
Remote SAP to the rescue? <ul><li>Remote SAP was developed to enable MER science team members the ability to use SAP at th...
Mixed Results <ul><li>Pros </li></ul><ul><ul><li>Remote SAP enabled MER to transition to distributed operations </li></ul>...
We Learned a Few Lessons <ul><li>Design operations tools to support distributed operations  from the very beginning </li><...
Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Opera...
 
 
Maestro: A Step in the Right Direction <ul><li>Designed from  the beginning  to operate in a distributed environment </li>...
Apache Tomcat JMS Server Maestro Client MySQL Data Catalog Plan DB Target DB MySQL Server OSS Flight Firewall dp-daemon me...
MySQL: Relational Database <ul><li>Maestro interfaces with a JPL institutional MySQL database instance installed in a clus...
Other Servers <ul><li>Web Server: Apache </li></ul><ul><ul><li>Data is first compressed and then cataloged </li></ul></ul>...
Security
Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Opera...
Downlink Assessment <ul><li>Contextual awareness is a requirement for activity planning </li></ul><ul><li>For a mobile sur...
Data Products <ul><li>The Operations Product Generation Subsystem (OPGS) pipeline fuses actual images and science data wit...
Data Catalog <ul><li>The OPGS publishes products to a file system </li></ul><ul><li>Distributed planning requires browsing...
Metadata
 
Compression <ul><li>Image products </li></ul><ul><ul><li>Compressed on the server side to reduce network traffic </li></ul...
Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Opera...
Uplink <ul><li>The uplink planning system must enable collaboration among a geographically distributed network of scientis...
Message Ordering and Protocol <ul><li>JMS does not guarantee message order when messages are sent from multiple sessions (...
Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Opera...
More yet to be done … <ul><li>Additional server dependencies </li></ul><ul><ul><li>Phoenix will use a  headless  APGEN (AP...
Conclusion <ul><li>Centralized mission activity planning system for any long-term mission is not feasible </li></ul><ul><l...
Upcoming SlideShare
Loading in …5
×

Advances in Distributed Operations and Mission Activity ...

320 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
320
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
1
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Advances in Distributed Operations and Mission Activity ...

  1. 1. Advances in Distributed Operations and Mission Activity Planning for Mars Surface Exploration Jason M. Fox Jet Propulsion Laboratory June 21, 2006
  2. 2. Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Operations Enhancements </li></ul><ul><li>Downlink Enhancements </li></ul><ul><li>Uplink Enhancements </li></ul><ul><li>Future Work / Conclusions </li></ul>
  3. 3. Introduction <ul><li>The Mars Exploration Rover Mission (MER) demonstrated that a centralized mission activity planning system for a long-term mission is not feasible </li></ul><ul><ul><li>Budget constraints </li></ul></ul><ul><ul><li>Lifestyle constraints </li></ul></ul><ul><li>What is Mission Activity Planning ? </li></ul><ul><ul><li>High level specification of scientific intent </li></ul></ul><ul><ul><li>Highly reactive planning environment </li></ul></ul><ul><ul><li>The focus is short-term, tactical operations of spacecraft activity </li></ul></ul><ul><li>Assertions </li></ul><ul><ul><li>The Ground Data System (GDS) must promote a secure, distributed mission activity planning system </li></ul></ul><ul><ul><li>MER did not get it right the first time around </li></ul></ul>
  4. 4. Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Operations Enhancements </li></ul><ul><li>Downlink Enhancements </li></ul><ul><li>Uplink Enhancements </li></ul><ul><li>Future Work / Conclusions </li></ul>
  5. 6. SAP - The Legacy Tool <ul><li>The Science Activity Planner (SAP) was the primary science operations tool for MER </li></ul><ul><li>Visualization and planning capabilities enabled mission operations team to direct the activities of both MER rovers </li></ul><ul><li>Budget cuts forced the descope of the original design for distributed operations features </li></ul><ul><li>Good News == The successful landing of both rovers and their unexpected long life </li></ul><ul><li>Bad News == Distributed operations capabilities are now required of SAP </li></ul>
  6. 7. Remote SAP to the rescue? <ul><li>Remote SAP was developed to enable MER science team members the ability to use SAP at their home institutions </li></ul><ul><li>Addresses the following requirements </li></ul><ul><ul><li>Distributed downlink data (e.g. rover images) to remote scientists in a timely fashion </li></ul></ul><ul><ul><li>Share labels assigned to surface features as they are created </li></ul></ul><ul><ul><li>Facilitate the sharing of files containing the planned scientific observations for the following day </li></ul></ul>
  7. 8. Mixed Results <ul><li>Pros </li></ul><ul><ul><li>Remote SAP enabled MER to transition to distributed operations </li></ul></ul><ul><ul><li>Supported hundreds of sols of successful operations </li></ul></ul><ul><ul><li>The downlink distribution system successfully delivered data to users on most sols </li></ul></ul><ul><li>Cons </li></ul><ul><ul><li>The system is very brittle </li></ul></ul><ul><ul><li>Difficult to administer and monitor </li></ul></ul><ul><ul><li>Target sharing is slow and unreliable </li></ul></ul><ul><ul><li>The plan sharing capabilities often fail </li></ul></ul>
  8. 9. We Learned a Few Lessons <ul><li>Design operations tools to support distributed operations from the very beginning </li></ul><ul><li>Ensure that all users, whether local or remote , use the tool in the same way </li></ul><ul><li>Eliminate dependencies on share file systems that will not function across typical firewalls (e.g. NFS) </li></ul><ul><li>Avoid polling-based synchronization in favor of event notification </li></ul><ul><li>Use relational databases to persist all data that supports concurrent editing </li></ul>
  9. 10. Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Operations Enhancements </li></ul><ul><li>Downlink Enhancements </li></ul><ul><li>Uplink Enhancements </li></ul><ul><li>Future Work / Conclusions </li></ul>
  10. 13. Maestro: A Step in the Right Direction <ul><li>Designed from the beginning to operate in a distributed environment </li></ul><ul><li>An authenticated user can accomplish her task using Maestro on any computer that has a network connection </li></ul><ul><li>This radical new approach to distributed operations requires a lot of server side support. These services must be: </li></ul><ul><ul><li>Reliable </li></ul></ul><ul><ul><li>Scalable </li></ul></ul><ul><ul><li>Secure </li></ul></ul><ul><ul><li>Flexible </li></ul></ul><ul><li>And thanks to the open source community, each service Maestro requires is built upon a readily available, capable, and existing project </li></ul>
  11. 14. Apache Tomcat JMS Server Maestro Client MySQL Data Catalog Plan DB Target DB MySQL Server OSS Flight Firewall dp-daemon metadata metadata plans targets data products data products range maps range maps messages, notifications physical machine boundary Key server instance data source Server Architecture Diagram
  12. 15. MySQL: Relational Database <ul><li>Maestro interfaces with a JPL institutional MySQL database instance installed in a clustered environment </li></ul><ul><ul><li>Secure, flexible, and scalable </li></ul></ul><ul><ul><li>Restricts access via IP address </li></ul></ul><ul><li>Auxiliary Data (e.g. authorization information) </li></ul><ul><li>Downlink </li></ul><ul><ul><li>Catalog of downlinked data products </li></ul></ul><ul><ul><li>Rover localization information </li></ul></ul><ul><li>Uplink </li></ul><ul><ul><li>Central planning database </li></ul></ul><ul><ul><li>Targets </li></ul></ul>
  13. 16. Other Servers <ul><li>Web Server: Apache </li></ul><ul><ul><li>Data is first compressed and then cataloged </li></ul></ul><ul><ul><li>Access to cataloged downlinked data via web server </li></ul></ul><ul><ul><li>Pull model superior to Remote SAP system of complete replication </li></ul></ul><ul><li>Messaging Service: Java Message Service (JMS) </li></ul><ul><ul><li>Synchronization required across all clients </li></ul></ul><ul><ul><li>Modifications to targets and plans are announced with JMS </li></ul></ul><ul><ul><li>Eliminates polling requirements in SAP </li></ul></ul><ul><li>Application Server: Tomcat </li></ul><ul><ul><li>Range data is provided via large range map products </li></ul></ul><ul><ul><li>To save bandwidth, do not transfer the entire file but instead retrieve range data per mouse click via a servlet </li></ul></ul>
  14. 17. Security
  15. 18. Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Operations Enhancements </li></ul><ul><li>Downlink Enhancements </li></ul><ul><li>Uplink Enhancements </li></ul><ul><li>Future Work / Conclusions </li></ul>
  16. 19. Downlink Assessment <ul><li>Contextual awareness is a requirement for activity planning </li></ul><ul><li>For a mobile surface-based spacecraft (e.g. MER) the context for planning changes almost daily </li></ul><ul><li>After every traverse or in situ instrument placement: </li></ul><ul><ul><li>Assess the quality of the new science data </li></ul></ul><ul><ul><li>Understand the state of the rover </li></ul></ul><ul><ul><li>Identify and prioritize targets of opportunity in the locale </li></ul></ul><ul><li>Automatic cataloging of data products and the Maestro tool allow scientists to perform the above tasks regardless of geographic location </li></ul>
  17. 20. Data Products <ul><li>The Operations Product Generation Subsystem (OPGS) pipeline fuses actual images and science data with metadata </li></ul><ul><ul><li>Time of acquisition </li></ul></ul><ul><ul><li>Rover position </li></ul></ul><ul><ul><li>Attitude </li></ul></ul><ul><ul><li>State of the instrument, etc </li></ul></ul><ul><li>These data and metadata are combined into an Experiment Data Record (EDR) </li></ul><ul><li>Derived products (e.g. calibrated images, XYZ position maps for stereo image pairs, etc) are called Reduced Data Records (RDR) </li></ul>
  18. 21. Data Catalog <ul><li>The OPGS publishes products to a file system </li></ul><ul><li>Distributed planning requires browsing data via remote access </li></ul><ul><li>The solution </li></ul><ul><ul><li>Add a relational database driven catalog on top of the file system </li></ul></ul><ul><ul><li>Provide a web server to enable remote access to the data products </li></ul></ul>
  19. 22. Metadata
  20. 24. Compression <ul><li>Image products </li></ul><ul><ul><li>Compressed on the server side to reduce network traffic </li></ul></ul><ul><ul><li>JPEG compression with the same PDS-format metadata header reduces product size by a factor of 14 </li></ul></ul><ul><li>Reachability maps </li></ul><ul><ul><li>Compressed (on average) to be three orders of magnitude smaller </li></ul></ul><ul><ul><li>Simplify the map to include only the information scientists require </li></ul></ul><ul><li>Range Maps </li></ul><ul><ul><li>Do not distributed the entire (large) range map product </li></ul></ul><ul><ul><li>Instead, provide answers to range queries on a click-by-click basis using an application server </li></ul></ul>
  21. 25. Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Operations Enhancements </li></ul><ul><li>Downlink Enhancements </li></ul><ul><li>Uplink Enhancements </li></ul><ul><li>Future Work / Conclusions </li></ul>
  22. 26. Uplink <ul><li>The uplink planning system must enable collaboration among a geographically distributed network of scientists </li></ul><ul><li>Two-pronged approach </li></ul><ul><ul><li>A central database to ensure initial synchronization and accessibility for all Maestro clients </li></ul></ul><ul><ul><li>A messaging service to maintain synchronization via event notification </li></ul></ul>
  23. 27. Message Ordering and Protocol <ul><li>JMS does not guarantee message order when messages are sent from multiple sessions (i.e. multiple Maestro clients) </li></ul><ul><li>Maestro implements a message ordering layer through a combination of JMS headers and database tables </li></ul><ul><li>Modifications to the target and/or plan state are robust </li></ul><ul><ul><li>Nominally, all clients receive notifications via JMS </li></ul></ul><ul><ul><li>Safeguards and a recovery protocol are in place to handle client crashes without compromising synchronization </li></ul></ul>
  24. 28. Outline <ul><li>Introduction </li></ul><ul><li>Limitations of Legacy Tools </li></ul><ul><li>Overview of Distributed Operations Enhancements </li></ul><ul><li>Downlink Enhancements </li></ul><ul><li>Uplink Enhancements </li></ul><ul><li>Future Work / Conclusions </li></ul>
  25. 29. More yet to be done … <ul><li>Additional server dependencies </li></ul><ul><ul><li>Phoenix will use a headless APGEN (APcore) for detailed plan modeling </li></ul></ul><ul><ul><li>MSL and Phoenix will both make use of an Ames constraint engine and modeling tool (Europa) </li></ul></ul><ul><li>Offline support </li></ul><ul><li>Better data discovery and catalog generation </li></ul>
  26. 30. Conclusion <ul><li>Centralized mission activity planning system for any long-term mission is not feasible </li></ul><ul><li>Distributed system is more flexible while at the same time more cost effective </li></ul><ul><li>Maestro is a step towards a completely distributed mission operations tool that folded into its design the following lessons from MER </li></ul><ul><ul><li>Design operations tools from the beginning to support distributed operations </li></ul></ul><ul><ul><li>Ensure all users, whether local or remote , use the same tool in the same fashion </li></ul></ul><ul><ul><li>Eliminate dependencies on shared file systems </li></ul></ul><ul><ul><li>Avoid polling-based synchronization in favor of event notification </li></ul></ul><ul><ul><li>Use relational databases to persist all data that needs to support concurrent editing </li></ul></ul>

×