ROS-Neuro: A common middleware for BMI and robotics. The acquisition and recorder packages

ROS-Neuro: implementation of a closed-loop BMI based on motor imagery

“…2 depicts a schematic representation of the ROS-Neuro framework with the proposed packages and main topics. Since the acquisition and recorder packages have been previously presented [19], herein we describe the other four proposed packages that are necessary to close the BMI loop. Although in this work we provide an example of a SMR BMI implementation, this architecture is designed to be generic and compatible with different BMI paradigms (e.g., P300).…”

“…The processing node -implemented in the rosneuro processing package -has access to the stream of EEG data by subscribing to the /acquisition/neurodata topic. Chunks of new data are received as NeuroFrame messages with a frame rate defined in the acquisition node as a ROS parameter [19]. The processing node is in charge to filter the incoming data in temporal, spectral and/or spatial domains.…”

“…The output of the processing module can be further processed by the rosneuro decisionmaking module, for instance by integrating the predictions over time, filtering the decoder output, and thus, to achieve a more stable and reliable control signal. The output is provided to the other modules by publishing again a NeuroOutput message, while a recognized BMI command is published as a NeuroEvent message [19] on the /events/bus topic. It is worth highlighting that both the processing and the decision making modules are publishing the corresponding predictions on distinct topics through the same type of message (NeuroOutput), but containing different information.…”

“…To overcome the aforementioned limitations, in [19] we proposed ROS-Neuro, a common open-source framework for BMI and robotics research and applications. The aim of ROS-Neuro is to provide a common software framework for the implementation of BMI systems and robotic controllers, boosting their mutual integration.…”

“…In [18] we have presented the general perspective of the ROS-Neuro framework and its advantages, while in [19] we proposed the first release of the acquisition and recording modules: namely the rosneuro acquisition and rosneuro recorder packages. These packages handle the acquisition and the recording of neural data from different commercial acquisition devices, facilitating their use among the research community.…”