One persistent issue with transform graphs has been their resource use. `ROS TF works through replicated copies of the entire transform tree at every node that uses it, and is implemented through unicast TCP connections between nodes`. In systems with many dynamic parts, this has sometimes been called the `TF firehose`, because of the large stream of incoming messages. micro-ROS will go beyond this state of the art by running the dynamic transform tree in an embedded device, while keeping resource use to a minimum based on an analysis of the spatial and temporal details actually necessary. Further, enabling real-time queries even in the face of concurrent updates through integration will be realized through integration with the micro-ROS real-time executor. It is also planned to integrate the embedded TF will with the node lifecycle to achieve further power-savings
Link: https://micro-ros.github.io//docs/concepts/client_library/embedded_tf/