The objective of this project is to design a lower cost version of the Turtlebot3 Autonomous Mapping Rover (AMR). AMRs are an increasingly popular field of robotics focused on navigating and generating maps of unknown spaces. The Turtlebot AMR is a robotics platform designed in collaboration between Robotis and Dynamixel. The system is designed to support numerous upgrades, but it’s base function is autonomous mapping. Given it’s modularity, the Turtlebot3 contains many components which are overly sophisticated for its base mapping function. To successfully reduce the cost of the system, the architecture of the Turtlebot3 was analysed to determine which components could be replaced while maintaining the functionality of SLAM, Simultaneous Localization and Mapping. SLAM is the mapping algorithm used by the original Turtlebot3 to navigate unknown spaces and generate a map of the space for future navigation. SLAM uses two data types to accomplish this task, laser rangefinding (LiDAR) and wheel feedback (Odometry). To design a lower cost version of the Turtlebot which supports SLAM these two data types must remain intact. Given the complexity and stability of the laser data, the main modifications were made to the drive architecture which executes motion and produces the wheel feedback.
The replacement drive architecture has a total cost of 104$ compared to the Turtlebot’s which costs 314$. These replacement components were programmed to be compliant with the Robotics Operating System, ROS, which the Turtlebot3 uses to run the mapping algorithm. ROS is a message passing architecture where stable components pass information to each other. Each hardware component publishes and subscribes to specific data types. If all the lower cost replacement components are programmed to subscribe and publish the same data types as the originals, the system's overall functionality should not change. After the replacement system was constructed and programmed, the accuracy of the individual ROS data types was evaluated. We found a high degree of similarity between the behavior of the low cost replacement system and the original. Subsequently, we can conclude that the lower cost system maintains the ability to conduct mapping with SLAM.