ABSTRACT The transition toward Industry 5.0 necessitates safe human‐machine interaction (HMI) in unstructured open spaces, where Internet of Things (IoT) infrastructures serve as the primary backbone for real‐time spatial perception. However, current HMI systems frequently decouple network‐layer operations from physical‐layer control, rendering robotic actuation highly vulnerable to IoT network degradation, including communication latency and stochastic packet loss. This paper proposes a Resilient Collaborative Control Framework (RCCF) that tightly couples network quality‐of‐service (QoS) metrics with physical actuation strategies. The methodology integrates an edge‐deployed adaptive Extended Kalman Filter (EKF) for multi‐sensor fusion under compromised IoT channels, alongside a network‐aware dynamic impedance controller that modulates robotic stiffness and damping in real time based on measured packet drop rates. Experimental evaluations on the publicly available SiT (Spatial Interaction Trajectories) dataset demonstrate that, under a simulated 15% packet loss scenario, the proposed RCCF achieves a tracking root mean square error (RMSE) of 12.4 ± 1.2 mm, an end‐to‐end latency of 31.8 ± 2.5 ms, and a collision avoidance rate of 98.5% ± 0.8%, yielding statistically significant improvements over static baseline controllers (one‐way ANOVA, p < 0.05). The framework effectively mitigates physical safety risks induced by communication degradation, providing a robust cyber‐physical control architecture for dynamic HMI in complex IoT environments.
Yao Ruan (Fri,) studied this question.