Objectives:
- Designed and implemented a multiagent system using CrewAI framework and ROS to facilitate campus navigation for visitors.
- Developed inter-agent communication protocols for coordination between Campus Incharge (CI) Agents, Building Incharge (BI) Agents, and Visitor Agents, ensuring smooth navigation and visitor management.
- Simulated system interactions with varying visitor arrivals, hosts, and meeting durations, visualizing navigation paths as graphs and logging inter-agent communication.
- Implemented penalty-based performance metrics to evaluate agents’ efficiency, including OOS (Out of Service) violations and adherence to committed durations.
Commands:
To build the ROS2 environment:
mkdir -p ~/ros2_ws/src
cd ~/ros2_ws/src
To create the main implementation code folder:
ros2 pkg create --build-type ament_python campus_tour
To create the communication code folder:
ros2 pkg create --build-type ament_python campus_tour_interfaces
Initiate the ROS environment:
cd ~/ros2_ws
colcon build
source ~/ros2_ws/install/setup.bash
Run the agent nodes and performance analyzer:
ros2 run campus_tour ci_agent_node
ros2 run campus_tour bi_agent_node
ros2 run campus_tour visitor_node
ros2 run campus_tour performance_analyzer_node
ros2 run campus_tour visualization_node
Run the virtual simulation:
ros2 launch campus_tour camp_sim_launch.py
#or
ros2 run campus_tour campus_map.py
Example Simulation:
- Visitor 1 is escorted to the CSE Department, where the BI Agent provides navigation assistance after verifying their authorization.
- Visitor 2 is escorted to the AI Department, and the BI Agent assists with navigation after verifying their authorization.
- Visitor 3 is escorted to the Cafeteria, where the BI Agent goes OOS for 2 minutes to host after verifying their authorization.
- Visitor 4 waits for 2 minutes while Visitor 2 meets the host, then is escorted to the AI Department after Visitor 2 leaves.
- Visitor 5 is denied access to the Director's Office by the BI Agent due to lack of proper authorization.
