MRMM project Multi Robots for Multi Missions MRMM
MRMM project (Multi Robots for Multi Missions)
MRMM Project- Multi Robots for Multi Missions
Full System
Goal
July 2016 demonstration Single patrol mission for a team All robots register in the main server 4 Robots get command to execute a mission After the mission was started command one robot to leave the mission (all the rest of robots should recalculate the mission) Join another robot to the mission Shut down one of the robots Shut down the main server and start it again without harm the mission execution
Challenges Manage large amount of challenges during execution Consider real-world constraints (energy consumption, low cost robots, unreliable communication) Agile adaption of a mission Mission management support join/leave of teammate Balance/priority/comparison of missions
Manage large amount of challenges during execution Robot Malfunctions Communication hangs up Environment hazards Mission info & alerts
What was done System specification document Mission formalization (based on BITE) version 1. 0 Server-clients architecture implementation (distributed services) Mission patrol planner Generic wrapping of algorithms as system services Robot-Mission allocation service based on robot properties and mission
System Specification Document Mission Manager Manage the mission team Synchronization in the team Monitor mission status Manage the mission initialization and ending Mission Execution Engine The commands are received from a ROS Topic Join/Leave mission commands Reconnect to a mission
Mission Formalization Mission flow- a logical flow describing the mission Recipe- a platform specific description how to execute the mission Plan- instantiation of the recipe
Server-clients architecture Robots Services User Console Mission Autonomous system Logic brain Command & Control Goal Drone 1 Navigate to XY User 1 Patrol 1 Drone 2 Mission Allocation User 2 Patrol 2 Vehicle 1 Patrol Planner Vehicle 2 Grip Planner Grip area
Execute mission User 1 Vehicle 1 Detection Vehicle 2 Detection Patrol Planner Navigate Vehicle 3 Detection Navigate
Server-clients messages ● Generic message protocol ● Java & C++ implementation ● Book of messages
Patrol mission planner
Algorithms as system services ●Generic wrapping ●Java implementation Hungarian algorithm - a combinatorial optimization algorithm that solves the assignment problem in polynomial time Christofides algorithm - approximation algorithm for finding a solution to the traveling salesman problem (a cyclic path visiting all given waypoints)
Ongoing Mission execution implementation (integration with Cogni. Tao for management team) System API definitions (ongoing development) Real robot implementation as system’s clients Research- comparing utility between different missions and mission prioritization
Thank you!
- Slides: 18