pybela: a Python library to interface scientific and physical computing

Workflows to obtain, examine and prototype with sensor data often involve a back and forth between environments, platforms and programming languages. Usually, sensors are connected to physical computing platforms, and solutions to transmit data to the computer often rely on low-bandwidth communicating channels. It is not obvious how to interface physical computing platforms with data science environments, which also operate under distinct constraints and programming styles. We introduce pybela, a Python library that facilitates real-time, high-bandwidth, bidirectional data streaming between the Bela embedded computing platform and Python, bridging the gap between physical computing environments and data-driven workflows. In this paper, we outline its design, implementation and applications, including deep learning examples.