Leveraging RF signals for human sensing: Fall detection and localization in human-machine shared workspaces

Safe human-machine interactions promote high flexibility in collaborative workspaces. Fall detection and localization of the operator are major issues in ensuring a safe working environment. However, many proposed solutions are not applicable for deployment in industrial environments due to their performance limitations in practical contexts. In this paper, we propose an integrated framework for both localization and fall detection of operators inside a shared workspace that employs radio-frequency (RF) signal analysis in real-time. Multipath and non-line-of-sight (NLOS) scattering that affect RF signal propagation can be leveraged for human sensing in complex workspaces: the proposed system continuously monitors the fluctuations of the RF field across the space by a dense network of WiFi compliant radio devices operating at 2.4GHz. To increase the accuracy of the localization system, a sensor fusion algorithm using Extended Kalman Filter techniques is employed. The proposed method may be used for integrating measurements from both RF nodes and an additional image-based system. For fall detection, a Hidden Markov Model is applied to discern different postures of the operator and to detect a fall event by tracking the fluctuations of the wireless signal quality. Fall detector performances are validated through experimental measurements. The preliminary results confirm the effectiveness of the proposed approach for different body configurations and pre-impact postures to correctly detect a fall event. Finally, some results about sensor fusion for improved operator localization are presented.