Abstract: In the last decade, robot manufacturers have started to produce collaborative industrial robots, that can work while safely sharing the workspace with a human operator. In this way, robot repeatability, combined with human dexterity, can move automated assembly to a new level of flexibility. The aim of this paper is to investigate the conditions at which such systems, called Collaborative Assembly Systems (CAS), can be better performing than the traditional manual or automated assembly systems. Throughput and unit direct production cost are considered for the comparison.
The estimation of such performance figures, which is straightforward
in traditional automated assembly systems, becomes more complex in the case of collaborative systems. In fact, both task allocation between the human and the robot, and the way they collaborate/interfere with each other during assembly, affect the throughput of CAS. With the aim of taking into account such parameters, we introduce a set of system variables and a mathematical model which allow to estimate the real convenience of the implementation of CAS in the industrial scenario.
In the paper, the model is applied to compare CAS to manual assembly and
to non-collaborative automated assembly, both with parameters derived from the literature and in a case study. Finally, a set of implementation conditions is derived, related to the task allocation that maximises CAS performance.