Prev-Talk | Next-Talk | All-Talks | Talks-Sorted | MOOS-DAWG'22

Talk-11: MOOS-IvP Release 22.8 - What's New and What is Upcoming

Michael Benjamin, MIT

This talk provides an overview new features and improvements contained MOOS-IvP Release 22.8, as well as discussions about upcoming project efforts. New release features can be grouped into the broad categories of (a) Core autonomy, (b) Simulation, (c) Mission Monitoring, (d) Automated Testing, (e) Mission Analysis

Core autonomy improvements can be found in the IvP Helm itself, IvP Behavior superclass features, and the core support modules of the Contact and Obstacle Managers. The helm includes new support for discerning between goal-directed and constraint-based behaviors to ensure there is always at least a single goal-directed behavior. The IvPBehavior and IvPConctactBehavior superclasses contain several new components to ease behavior development and provide more run-time configuration options through event flags and macros. The new contact manager, pContactMgrV20 replaces the previous contact manager with several new features. The obstacle manager has also been substantially augmented in terms of memory management and obstacle generation.

Simulation improvements include a new replacement for uSimMarine, uSimMarineV22 which supports very high time warp simulations of up to 400x realtime in headless modes, and up to 100x in GUI based sims. Mods to pMarineViewer and other apps have been made to handle the higher time warp simulations. An obstacle simulator, uFldObstacleSim, has been substantially improved to support obstacle avoidance automated simulations and testing.

Substantial mission monitoring improvements have been introduced through a new regime called RealmCasting, similar to AppCasting in earlier releases. Nodes (vehicles and shoreside) are able to produce RealmCasting reports holding the recent publications and subscriptions for a given node and application. Watch reports are also possible to monitor one or more variables across all vehicles, without any coding changes. The pMarineViewer app has also been improved for rendering efficiency in the higher time warp regime, and to support RealmCasting, and to allow for a more customized GUI button layout.

Automated testing improvements have been made to auto-launch missions, using modified versions of uPokeDB and uQueryDB, and through a single customizable script for launching headless missions supervised for completion based on customizable conditions. A number of testing MOOS apps have been written or improved for detecting vehicle or obstacle collisions or near misses, as well as detecting anomalies like trajectory wrap-around.

Mission analysis tools such as alogview have been improved to more efficiently handle missions with large numbers of vehicles, MOOS variables or both. A number of new command line utilities have been added or improved to quickly analyze log files for events of interest such as obstacle or vehicle encounters, or to follow helm mode progression and behavior spawning and deletion.

Upcoming features currently under development will be discussed. This includes further support for automated testing, automated grading of code submissions used in training or education situations, support for moving dynamic datums for missions over large bodies of water, better support for integration into other architectures such as UMAA, and migration of swarm autonomy tools into the public tree.

Categories:

  • MOOS-IvP
  • Simulation
  • Mission Monitoring