Fenceless Collision-Free Avoidance Driven by Visual Computation for an Intelligent Cyber–Physical System Employing Both Single- and Double-S Trajectory

Toward human is always one of the critical missions for any cyber physical system (CPS). Autonomous robots perceive embarrassing behavior when treating humans as conventional obstacles. In the context of flexible manufacturing or intelligent production, smart CPSs have an important role, since they contributes to the achievement of higher productivity and greater efficiency. In this paper, for intelligent human–machine interaction, a novel approach to safety, comfortability, and anti-collision using the four strategies of emergency break (EMB), slow-down break (SLB), slow-down break then emergency break (SLEMB), and emergency break then emergency break (EM2B) based on visual computation is presented. First, a dynamic model of whole CPS consisting of a vehicle’s body and freight was established. To guarantee comfortable motion and social collaboration, an interactive framework that includes human information was integrated in order to adjust the proper actions of the robot. For the actively and highly computational method, fusing data from both sensing modules and digital cameras provides an efficient technique to estimate a human’s pose and launch reasonable behavior for the movement of the robot. Later, for more intelligent and adaptive performance, a switching mechanism considering the human state was introduced. Based on the results of computer-based simulation and experiments, our approach is valuable for socially human-aware navigation in front of humans with respect to actively safe constraints. The contributions of this work are summarized as follows: (i) dynamical analysis, including the smooth profiles, was studied to suppress residual vibration; (ii) vision-based computation of reaching humans due to the socially interactive model was implemented; and (iii) the switching mechanism provides an alternative method for choosing the proper solution.