In manufacturing environments, human workers interact with increasingly autonomous machinery. To ensure workspace safety and production efficiency during human-robot (HR) cooperation (HRC), continuous and accurate tracking and perception of workers’ activities is the key. Examples are reactive collision-prevention, active recognition of workers’ intention, as well as accurate tracking of human-robot interactions. The RadioSense project intends to move forward the state-of-the-art in advanced sensing and perception for next generation manufacturing workspace. We explore passive radio sensing, or “Radio Vision” technology which aims to track, recognize and analyse human-robot interactions continuously without requiring workers to wear any devices, and without the need for privacy-intrusive video, while ensuring workers’ safety and privacy in industrial environments. RadioSense technology leverages real-time collection and processing of heterogeneous radio signal streams (e.g., those found in 4G/5G and WiFi connections) and multiple Channel State Information (CSI) between different links/antennas. Perturbations induced by moving bodies/objects on the EM wavefield can be processed to extrapolate a 2D/3D image of humans and the environment. The radio data streams from a multitude of wireless devices with different carrier frequencies, bandwidths and spatial resolutions, according to their different technologies and protocols, form a specific type of “big data” and provide a non-intrusive way to monitor and analyse the next-generation workspace. On top of the big data analytic tools, the RadioSense project aims to address three key challenges in smart manufacturing systems, namely: device-free localization of multiple subjects co-present in the workspace, smart actuation or control of machines using intuitive gesture-based interfaces, and safety guarantee for workers in the workspace, by monitoring anomalous human-robot interactions. RadioSense will adopt next generation (5G) high-frequency technologies as well as distributed massive MIMO systems. A cloud-IoT open platform will be also developed to support cloud-based CSI data fusion, analysis and manipulation tools to guarantee a seamless integration with off-the-shelf industrial sensors. Building on all partner's long expertise in device-free CSI-based sensing, the RadioSense project kick starts the development of innovative sensing tools as key enablers in advanced manufacturing and HRC workspaces. The theoretic foundations derived will generate new research domains in radio-based recognition and create an impact both scientifically and industrially appropriate to the technological advancement it represents.
Contact: Dr. Stephan Sigg (coordinator), firstname.lastname@example.org