Towards Dynamic Interoperability Methods for Complex and/or Reconfigurable Robotic Systems in Dynamic Unstructured Industrial Environments

Despite the advances in industrial automation systems and integration approaches, many remain as monolithic, complex, hard-wired, and custom-designed for specialized applications or processes and inflexible to change or updates. This makes it challenging to integrate them with newer and emerging intelligent devices (e.g., robotic manipulators, mobile robots, and IoT-enabled edge devices) due to inherent complexities requiring significant engineering efforts for redesign, testing, tuning, and calibration. The causes include inflexible engineering practices and lack of foresight to design systems for extensibility and interoperability. In contrast, the software industry has progressed beyond monolithic systems and has embraced modular architectures and micro/nano-services based designs, especially with pervasive cloud computing.
Nevertheless, it will be crucial for the future manufacturing and logistics automation systems to offer flexibility, reconfigurability as well as interoperability, as the existing systems in each facility will be expected to work with new devices and/or systems for newer applications and/or to support the production of brand-new products with minimal downtime or changeover time. Furthermore, mobile systems (autonomous mobile robots, manipulator arms or work cells) can introduce significant uncertainty to throughput, due to the stochastic nature of their movements in human-intensive operational environments. Besides, robot failures may require other robots to be quickly summoned and/or repurposed to resume operations.
In this PhD project, the student can research and develop novel methods to establish meaningful interoperability between such complex automation systems and guest entities (e.g., new / repurposed / reconfigured robots and devices), first at a syntactic level, then proceeding to semantic levels and finally progressing towards the level of dynamic interoperability. This may require novel information exchange models, vendor-neutral / self-describing languages and ontologies and compact / efficient knowledge representations. Achieving efficient task execution between entities and minimizing the operational risks (be it the risk of causing operational failures or unsafe situations due to interoperability gaps) are critical. The proposed methods can be verified in simulated / representative industrial environments and may eventually be considered for deployment in production where relevant.