Commonsense Knowledge & Hybrid Artificial Intelligence for Trusted Flexible MAnufacturing 4.0 – CHAIKMAT
Commonsense Knowledge & Hybrid Artificial Intelligence for Trusted Flexible MAnufacturing 4.0
/
Challenges and objectives
In the context of Industry 4.0, flexible production plays an important role in the development of the factory of the future. The CHAIKMAT project aims to propose a new AI-based approach that will improve manufacturing flexibility, increase the transparency of decision-making, and strengthen human-machine trust. The project team therefore proposes a human-centered AI approach that verifies whether machines are capable of executing a new production process and provides human experts with interpretable explanations on how the decision-making process unfolds. To achieve this, CHAIKMAT will orchestrate ontologies, semantic reasoning, and machine learning models through a common-sense knowledge graph in manufacturing. Validation will be ensured via a test platform consisting of small robotic stations and transport systems that will be designed and developed within the project. In the context of Industry 4.0, CHAIKMAT will study the effectiveness of using standards proposed in the digital twin movement, in particular the «Asset Administration Shell« (AAS) proposed by the International Digital Twin Alliance (IDTA), and the developments to be expected. These developments will aim to facilitate the semi-automatic integration of new elements into the production chain.
/
The CHAIKMAT technical platform, comprising 4 robots with heterogeneous technologies, communication modes and functionalities, is largely operational. It allows the research issues planned in the project definition to emerge (equipment capabilities, error decision-making, process modeling and interpretation). These issues are addressed from their semantic aspect, from the decision-making process regarding the operations to be carried out (common sense knowledge) to the description of the equipment (semantic asset administration shell) and the orchestration of the actions of the different elements involved. Ongoing developments concern the remote control and monitoring of the chain (via a website), and the integration of AI equipment (vision kits) into the manufacturing processes that illustrate the current operation of this miniature production line. Current scientific work related to the platform explores the collaboration between the different elements (robots, sensors) within a given task, and more precisely the substitution of subsystems by other subsystems with equivalent functionalities during a technical failure (automatic or semi-automatic reconfiguration based on the semantic description of the actors and tasks).
/
Naqvi, M. R.; et al. Survey on ontology-based explainable AI in manufacturing. Journal of Intelligent Manufacturing. 2024, 1-23.
Naqvi, M. R.; Sarkar, A.; Ameri, F.; Araghi, S. N.; Karray, M. H. RobOntics: Workshop on Ontologies in Autonomous Robotics, August 28, 2023, Seoul, South Korea, Application of MSDL in Modeling Capabilities of Robots. 2023.
Liu, Z.; Sarkar, A.; Araghi, S. N.; Archimede, B.; Karray, M. H. Exploring models' interoperability in digital thread and twin: a proposal for an ontology-driven approach based on intra- and cross-organizational enterprise integration. In 27th International Conf on Production Research, July 2023.
Sarkar, A.; Naqvi, R.; Elmhadhbi, L.; Archimede, B.; Karray, M. H. Common-sense Knowledge & Hybrid Artificial Intelligence for Trusted Flexible Manufacturing: Chaikmat 4.0. In Flexible Automation and Intelligent Manufacturing: The Human-Data-Technology Nexus: Proceedings of FAIM 2022. Springer International Publishing. 2022, 2, 455-465.
In the context of Industry 4.0, flexible manufacturing plays a significant role in developing the factory of the future. CHAIKMAT project aims to propose a novel AI-based approach that will enhance flexibility in manufacturing, increase the transparency of decision-making systems, and improve trust between humans and machines. Accordingly, we propose a human-centric Artificial Intelligence (AI) approach that investigates whether the factory’s machines are capable of performing a new production process and then provides human experts meaningful explanations of how the decision process is conducted. To do so, CHAIKMAT will orchestrate exploiting ontologies, using semantic web reasoning capabilities, and machine learning models through a manufacturing commonsense Knowledge graph. The validation will be ensured through a real test plant that includes small robot stations and transportation systems that will be designed and developed within the project.
Project coordination
Hedi Karray (LABORATOIRE GENIE DE PRODUCTION)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
LGP LABORATOIRE GENIE DE PRODUCTION
Texas State University / INFONEER
University of Southern California / Information Sciences Institute
Clemson University/ Geometric Reasoning and Artificial Intelligence Lab
Help of the ANR 315,952 euros
Beginning and duration of the scientific project:
September 2021
- 48 Months
