MIMAMP
a Mixed-Initiative Multi-Agent Mission Planner

MIMAMP is a system that intelligently supports the human operator in creating multi-UAV planning problems,
performs automated planning and displays the resulting flight and sensor plans in a user-friendly way.
MIMAMP prevents the user to know anything about PDDL language nor automated planning algorithms and
provides support in creating targets, specifying observation requests, defining missions and understanding the final mission plans.

-

MIMAMP has been developed by D. Dell'Anna and M. Roberti, with the supervision of Prof. Pietro Torasso
@ Department of Computer Science, University of Turin, 2016.

The source code is available here.


Simple mission definition

Draw point or linear targets on the map and define observation requests. Easily specify mission constraints and planning settings.

Interactive interface

Mission definition and planning process are entirely supported by continuous checks of the user's choices and informative messages or suggestions.

Google maps plans visualization

Multi-agent flight plans are displayed on both a map and a tree representation, as well as sensor plans. GMapsFX library allows Google Maps Java integration.

PostgreSQL

The open source object-relational database system PostgreSQL is used to store data and make the application statefull.

Action-based planning

Missions are automatically encoded into PDDL problems. Numerical and temporal planning, as well as different planning domains can be chosen by the user, leveraging the PDDL 2.1 and 2.2 features.

General-purpose planners

Many general-purpose domain independent automated planners (e.g. COLIN, Metric-FF, LPG, TFD, POPF2) are integrated and can be dynamically chosen by the user before planning.


Easy data managing

The interface allow to easily draw on the map new point or linear targets and to define new observation requests, specifying the sensors to use and the temporal constraints.

A simple modular planning architecture

The planning architecture is very simple and modular and allows to dynamically change both the planning model and the planners during execution, without impacting on the input nor the output of the system.