A Multi-Robot-Based Architecture and a Trust Model for Intelligent Fault Management and Control Systems

Abstract

One of the most important challenges in robotics is the development of a Multi-Robot-based control system in which the robot can make intelligent decisions in a changing environment. This paper proposes a robot-based control approach for dynamically managing robots in such a widely distributed production system. A Multi-Robot-based control system architecture is presented, and its main features are described. Such architecture facilitates the reconfiguration (either self-reconfiguration ensured by the robot itself or distributed reconfiguration executed by the Multi-Robot-based system). The distributed reconfiguration is facilitated through building a trust model that is based on learning from past interactions between intelligent robots. The Multi-Robot-based control system architecture also addresses other specific requirements for production systems, including fault flexibility. Any out-of-control fault occurring in a production system results in the loss of production time, resources, and money. In these cases, robot trust is critical for successful job completion, especially when the work can only be accomplished by sharing knowledge and resources among robots. This work introduces research on the construction of trust estimation models that experimentally calculate and evaluate the trustworthiness of robots in a Multi-Robot system where the robot can choose to cooperate and collaborate exclusively with other trustworthy robots. We compare our proposed trust model with other models described in the literature in terms of performance based on four criteria, which are time steps analysis, RMSD evaluation, interaction analysis, and variation of total feedback. The contribution of the paper can be summarized as follows: (i) the Multi-Robot-based Control Architecture; (ii) how the control robot handles faults; and (iii) the trust model.