Machine Learning Assisted Visible Light Sensing of the Rotation of a Robotic Arm

With the rise of LED (light-emitting diode)-based luminaires, artificial lighting has become a technology platform, which, besides providing illumination, also provides communication and positioning functionalities. Apart from this, most recently Visible Light Sensing (VLS), in which lighting is used for sensing purposes, emerged as another embodiment of functionalities lighting could take over in the future. Here we show that machine learning assisted VLS has promising potentials to become a meaningful enabler for the Industrial Internet of Things. We show that the rotation of a robotic arm can be accurately monitored by VLS simply by equipping the robotic arm with sequences of colored retroreflective foils. Moreover, we show that the sensing task is compatible with a modulation of the light. This paths the way that sensing and communication tasks can be performed in parallel with one and the same low-complexity infrastructure, that apart from this also could take over the task of the obligatory room lighting. We demonstrate the capability of the approach even if the overall illumination conditions change. Therewith, VLS accentuates as an alternative option for industrial robot monitoring in combination with optical wireless communication.