Pioneer MRS Experimental Platform

New Features:

• Apply actionlib and reconstruct the communication framework
• Support Python scientific computing libraries by rewriting the algorithm node

Future Plan:

• Work on rqt visualization, combining plot, console, dynamic reconfigure, etc.
• Explore MATLAB ROS toolbox and try to connect it with this platform
• Explore some new features in ROS 2.0

Goal:

• Make this experimental platform flexible, efficient, and user-friendly.

Distributed Multi-Robot Formation

Main Features:

• Be able to keep formation while moving (by optimal-point algorithm)
• Support MobileSim and Gazebo simulators
• Support Vicon motion capture system
• Outdoor experiments depend on odometry (by init five robots on one line with 1m interval)

Future Plan:

• Simplify launch files that avoid launch sequence problem
• Apply actionlib to re-construct the communication mechanism
• Optimize framework and reduce computational cost
• Provide algorithm interface to Python and MATLAB

V1.2 MRS Control Framework

New Features:

• MRS algorithm demo works well
• Be able to initialize all five robots at the same time on one line with 1m interval
• Support client-server mechanism for requesting neighbor position
• Support both MobileSim and Gazebo simulators

Future Plan:

• Dynamic reconfigure and GUI
• Re-write nodes with Python
• Solve the collision problem

V1.1 MRS Control Framework

Capabilities:

• Control the (x,y) velocity of the hand point of the robot
• Friendly interface to control part of or all the robots at the same time
• Set parameters when launching or running
• Localization depends on either odometry or VICON motion capture system
• Task 1: multi-robot trajectory tracking (circular shape)
• Task 2: distributed adaptive gradient optimization algorithm

Drawbacks:

• Algorithm node has problems which need to be fixed
• Algorithm and trajectory nodes are not independent from extra facility (VICON)

Future Plan:

• Fix the logical bugs in algorithm node
• Re-write some nodes with Python
• Solve the collision problem
• Dynamic reconfigure and GUI

A basic multi-robot control framework

Capabilities:

• Control the hand point of a robot (as a linear system) via feedback linearization
• Distributed launch for 1-5 robots from your own laptop as a master
• Keyboard teleoperation for 1-5 robots at the same time
• Each robot can track a desired trajectory

Drawbacks:

• Localization only depend on odometry
• Parameters are hard codes

Future Plan:

• Localization by VICON motion capture system
• Set parameters in launch files or by dynamic reconfigure
• Develop sonar sensors to avoid collision
• Implement some other multi-agent algorithms
• Add more details into tutorials