Our senior design project focuses on the revival and development of a heterogeneous multi-robot containment and escort system utilizing a Hiwonder SpiderPi hexapod robot as the leader and four TurtleBots as the escorts or followers. This project introduces students to coordinated control, real-time communication, and containment logic across the robot team, and in the future could aim to aid real-world applications like automation, surveillance, other laboratories, and possibly the military.
The primary objective of our project is to integrate the SpiderPi and TurtleBot2 platforms using a ROS2 communication framework, enabling the TurtleBots to hold a convex hull escort formation around the SpiderPi as it navigates its environment. This project also aims to achieve our sponsor’s goals of modernizing older versions of her robotic hardware, extending her lab’s capabilities, and providing future students with a framework to grow with the ever-evolving multi-agent robotic research.
The basis of our system is a leader-follower containment strategy with the leaders implementing collision avoidance and followers keeping the containment hull formation consistent. Position tracking is facilitated through AR tag fiducial markers (essentially a QR code type of sensing) selected through a trade study considering resource availability, cost, efficiency, and the requirements for this project. AR tags became the optimal strategy due to its compatibility with ROS2 and their capabilities to run the course intended.
Simulations play a central role in our system: TurtleSim gives us an avenue to experiment with the convex containment structure and Gazebo allows for simulation expansion into a 3D environment incorporating our sensors, mapping algorithms, and a more advanced control logic.
At the current stage, our project has completed a conceptual design, subsystem trade studies, and preliminary simulation validations. Results from our 2D and 3D simulations through TurtleSim and Gazebo respectively have shown us that our requirements are attainable. In the near future we also look to implementing our simulation and algorithms on incoming hardware and a complete physical system to start applying our progress on a real-time scale. We've come far in simulations and algorithm knowledge, but more research, improvement, and trial and error are still in the works for the future of our project.