Control of Cooperating Robots/Robot Arms
FAIR 1 Archive Notice
This page describes research activities and results from the FAIR 2019-2023 program period.
The content is retained for reference purposes and is no longer actively maintained. Information on this page, including research outputs, team members, contact details, collaborations and links, may be incomplete or outdated.
This use case has been succeeded by the Autonomous systems use case within the FAIR 2024-2028 program.
Use Case Description
The use case demonstrates AI can make Smart Manufacturing more intelligent by making the Industry 4.0 assets more intelligent to make low-volume-high-mix production.
Results & Achievements
In 2021, we demonstrated how Reinforcement Learning (RL) models can be used to optimize scheduling of multi-agents in an assembly production line. We showed that Q-Learning RL model can optimize light switches assembly by increasing productivity by 5% and generating scheduling scenarios 10x faster to adapt to unforeseen situations.
In 2022, we extended the scheduling problem to kitting applications where multi-parts should be brought from different locations to different assembly stations in an optimal time and taking into account availability of assembly stations. As a first step, simulation models were developed and generate different scheduling scenarios. RL model were trained with these data and proved feasibility to deal with multi agents scheduling optimization while benchmark methods (such as PDDL scheduler) does not quickly convergence when more than 3 agents are used.
In 2023, we continue making the kitting application more tangible. We developed a ‘world model’ architecture around our InfraFlex set-up that allows different Industry 4.0 assets (cobots, AGVs, smart sensors, etc.) being connected to one system that allows controlling them in a configurable way. We succeeded to validate this architectures on real / virtual cobots and AGVs. The problem formulation for multiple assets scheduling was then better formulated and more relevant data become possible to generate.
Outreach
Different internal & external events
Uptake
In prospect
Basic Use Case Info
Period
2019-2023
Contact
Bey Abdel
Research Groups
Flanders Make, VUB AI, VUB R&MM