ROS2-based Architecture for MAV Data Sensing

Abstract

Micro Aerial Vehicles (MAV) are a flexible and low-cost platform for performing airborne sensor measurements. Their small size and high maneuverability in tri-dimensional space makes them particularly efficient in indoor scenarios. However, integration of said devices in larger, more complex Industrial Internet of Things (IIoT) architectures can cause what can be deemed as an "interoperability challenge" due to the different communication protocols used in these systems and the critical resource-constrained nature of MAVs. In this dissertation, we propose an end-to-end sensing architecture based on the Robot Operating System 2 (ROS2) framework to solve this challenge. In this regard, as the MAVs are not capable of running the desktop version of ROS2, we use micro-ROS, a stripped-down version of ROS2 which was specifically designed to run on resource-constrained devices by using eProsima Micro XRCE-DDS as its middleware. The usage of this middleware allows the MAVs to be seamlessy integrated in the ROS2/DDS world and therefore communicate and cooperate with other ROS2-enabled devices including a ground station, other sensors and even cloud-based nodes. Furthermore, a custom Micro XRCE-DDS agent was developed to use CRTP, the proprietary transport protocol of the MAVs used, the Crazyflies 2.1 quadcopters. As an example use-case we have integrated an Indoor Air Quality (IAQ) sensor on-board the MAVs, to perform measurements in user-provided waypoints which can be used for both monitoring and localization of airborne contaminants. Additionally, we have developed a missionplanning algorithm based on the Multiple Depot Multiple Travelling Salesmen problem that can optimize the total mission time by taking into account not only the distance between sensing waypoints but also the time needed to perform the measurement. Validation of the architecture was done by performing real-world experiments in a "test arena" which showed that the MAVs were capable of successfully performing sensing missions and interact with other devices by leveraging the ROS2 framework. Furthermore, we have performed a preliminary evaluation of the communication performance which showed that although the usage of micro-ROS and our custom agent can cause a considerable latency overhead, the framework can still be viable for sensing applications and the interoperability and flexibility it offers make it a powerful framework for embedded robotic sensing systems.