Gokulnarayanan_sathyanarayanan_intern

Multi agent path planning

This repository contains the code for Multi agent path planning using decentralized approach.

Multi agent planning demo:

1. Clone the repository -master branch.
2. catkin make

Agent server launch

1. Inorder to launch the multi agent server and rviz visualization. Run the command roslaunch multi_agent_planning multiagent.launch . This loads two agents with first agent having id 1,start position(0,0,0) and other agent having id 2 and start postion(5,5,0). This also opens rviz with preset config for visualization . But currently it will be empty. The current position is displayed in the topic /agent_feedback.

2. Start position of the agent are loaded from the config file. To change the start position of agent1 , open config/ agent1_info.yaml. Similarly for agent 2 open config/ agent2_info.yaml

Decentralized planner launch

2. To launch the planer with A*. Run the command rosrun multi_agent_planning planner. This will launch the /get plan service.

Update the goal for the agent

3. Run the command rosrun multi_agent_planning agent1_goal which will update the goal for the agent1 with (5,5,0).
4. Run the command rosrun multi_agent_planning agent2_goal which will update the goal for the agent2 with (10,8,0).

Final Visualizations

Agent 1 Path:(0,0,0) to (5,5,0)

Agent1 and Agent2 path: A1((0,0,0) to (5,5,0)) A2(5,5,0) to (10,8,0)

Changing goal for agent1:

Method1

1. Open the /src/agent1_goal.cpp and modify the g.request.goal_x = 5; g.request.goal_y = 5; g.request.goal_theta =0;

Method 2

2.rosservice call /agent1/agent/update_goal "goal_x: 0 goal_y: 0 goal_theta: 0"

Changing goal for agent2:

Method1

1. Open the /src/agent2_goal.cpp and modify the g.request.goal_x = 5; g.request.goal_y = 5; g.request.goal_theta =0;

Method2

2.rosservice call /agent2/agent/update_goal "goal_x: 0 goal_y: 0 goal_theta: 0"

Note:

The agent will always start the path planning from the start position.

Directory and File structure:

include

1. roadmap.h - creates a 4 connected roadmap of 10x10 grid of nodes using adjacency list representation.

2. node.h - class implementation of a node in the roadmap.

3. path planner.h - A*implementation for shortest path planning.

4. Decentralized_planner.h - class implementation of decentralized planner which routes the plan requests from multiple agents

5. agent.h - class implementation of an agent. '''

src

1. agent.cpp - main function for the agent node

2. Decentralized_planner_main.cpp - main function for planner execution

launch

1. multiagent.launch - Launches the agent node. Launches rviz for visualization.

msg

1. path.msg - custom message to store the path

2. Robot_state.msg - custom message to publish the state of the agent

srv

1. goal.srv - service file requesting and responding to a goal.

2. plan_request.srv - service file requesting a plan.