In the evolving Industry 5.0 landscape, human-robot collaboration (HRC) is crucial to improve safety and productivity in shared industrial workspaces. This paper presents a cost-effective solution to track the human skeleton and identify the body parts that are most exposed to collision risk during a HRC task, i.e. the head and the hands, using a RGB-D camera. The depth images are processed through artificial intelligence (AI) algorithms, ensuring an accurate estimation of human motion and proximity to the robot. Furthermore, a robot control module has been developed to modulate the robot’s speed based on the risk assessment calculated in real time, implementing a collaborative Speed and Separation Monitoring (SSM) scenario. To improve the efficiency of the workcell, the research proposes a multiplication coefficient to compute the minimum protective separation distance suggested by the ISO 10218-2, which weights the actual risk based on the biomechanical limits of the human part involved, according to the maximum permissible forces provided by the standard. Experimental results in a real robotic work cell demonstrate that the proposed system can achieve safety performance comparable to traditional motion capture systems, such as the OptiTrack; meanwhile it does not require operators to wear additional equipment and offers the advantages of reduced costs and a simpler setup, which is more affordable and adaptable to a wider range of industrial environments.

This paper presents the kinematic characterization of a 3-Cylindrical-Prismatic-Universal (3-CPU) parallel manipulator designed for motions of pure rotation. The machine has been conceived at the Polytechnic University of Marche, and recent studies have shown that its kinematic architecture can be exploited for the realization of reconfigurable machines with different kinds of motions (pure rotational, pure translational, and planar motions among others). The 3-CPU concept has been subject to further investigations for a deeper understanding of this peculiar behavior. After a brief introduction to these concepts, the paper faces the position and the differential kinematics of the 3-CPU spherical manipulator aiming at identifying workspace boundaries and its kinematic manipulability.

The paper describes the architecture of a novel parallel kinematics machine that is characterized by motions of pure translation. It is shown that the Cartesian arrangement of the 3 limbs assures simple kinematic relations and an isotropic workspace, without any rotation or translation singularity. Such good features should allow the design of parallel robots that are characterized by simple controllers and good performances, as an alternative to current mechanical schemes.