Cooperative Service Robots for the Predictive Maintenance of Process Plants

In this contribution we outline a new project in the field of service robotics. The goal is to develop teams of mobile robots that perform routine inspection, health monitoring, and early fault detection in large chemical process plants or in public utility networks. We envisage a general-purpose assistant for the maintenance engineer to collect, compress and screen large amounts of inspection data. It will provide fast and extensive coverage of large or hard-to-access plant structures by employing non-contact sensors: an infrared camera, a laser scanner (mainly for navigation and mapping purposes), electronic micro noses, and non-destructive test sensors. Objects of primary interest are isolated vessels, pipe ducts, heating or cooling aggregates, valves, flanges, pumps, fans, filters, and electrical equipment. As a mobile network of sensors continuously observing the process from outside the vehicles supplement today's process control systems with stationary networks of in-line sensors. The paper focuses on three aspects: the demand for such a novelty in inspection and maintenance, the overall system concept, and main scientific challenges to solve.