optimal_secure_topology_ws

A framework for simulating multi-robot formation control algorithms using ROS and Gazebo.

WORK IN PROGRESS

Project Description

Due to the absolute dearth of multi-agent simulations frameworks for ROS (as opposed to ROS2, which recently had ChoiRbot added to its arsenal), this repository was created with the intention to create a unifying simulation framework for testing Networked Control System algorithms on multi-robot systems. This was originally made for simulating attacks on multi-agent formation control algorithms of the Secure Topology Design project under Prof. Vaibhav Katewa's guidance.

The key idea is to reduce the workload required in setting up the multi-robot simulation stack minimal and allow for plug-and-play testing of distributed control algorithms. To launch a simulation all one has to do is edit the configuration file, called formation_params.yaml in the formation_simulation package to specify the number of robots of simulate and their initial position in the XY world frame of Gazebo simulator. An example of the formation params file can be seen here:

# Specifying the formation specific parameters excluding the information related 
# to the communication graphs. Remember that the numerical paramaters unless specified
# with a decimal will have to be treated by stringstream. The matrices need to be manually
# translated from XmlRpc to EigenXd.

# Selecting the verbosity levels for nodes.
SELECT_DEBUG_MODE: true

# Initial pose format, row n = [x_n, y_n, z_n]

SPAWN_BOTS_GAZEBO: true
SPAWN_BOTS_RVIZ: false
len_uid: 5

TURTLEBOT3_MODEL: burger

formation_config: {
  USING_CUSTOM_UID: true,
  USING_FORMATION_CENTER: true,
  USING_COMMON_FRAME: false,
  num_bots: 5,
  uid_list: [BOT1, BOT2, BOT3, BOT4, BOT5],
  leader_uid: [],
  #              x    y    z    r    p    y
  initial_pose: [4.0, -1.0, 0.0, 0.0, 0.0, 0.0,
                6.0, -1.0, 0.0, 0.0, 0.0, 0.0,
                6.0, 1.0, 0.0, 0.0, 0.0, 0.0,
                4.0, 1.0, 0.0, 0.0, 0.0, 0.0,
                5.0, 0.0, 0.0, 0.0, 0.0, 0.0],
}

# Type can strictly be either directed or undirected. The self connections
# parameter is required for verification of the adjacency matrix as well as
# initializing one in case the matrix is not declared or the dimensions are
# mismatched.

formation_graph: {
  DIRECTED_GRAPH: false,
  SELF_CONNECTIONS: false,
  A: [0.0,  1.0,  1.0,  1.0,  1.0,
      1.0,  0.0,  1.0,  1.0,  1.0,
      1.0,  1.0,  0.0,  1.0,  1.0,
      1.0,  1.0,  1.0,  0.0,  1.0,
      1.0,  1.0,  1.0,  1.0,  0.0,]
}

robot_state_publisher: {
  USE_STATIC_TF: true,
}

test_motion_planner: {
  num_bots: 5,
  USE_STATIC_TRANSFORM: true,
  uid_list: [BOT1, BOT2, BOT3, BOT4, BOT5],
  planner_center: BOT5,
  #                   x   y 
  static_formation: [0.0, 0.0,
                     2.0, 0.0,
                     2.0, 2.0,
                     0.0, 2.0,
                     1.0, 1.0]
}

The package creates all the unique IDs (UIDs) for the robots (unless UIDs are already provided) and creates topics and services based on the communication graph represented by the adjacency matrices. The variable weights of the graph allow for consensus control based algorithms for robot formation control. This open up a lot of possibilites for testing multi-agent control applications in ROS-1.