Digital twin-enabled multi-robot system for collaborative assembly of unorganized parts

Agility and flexibility in production are essential to meet the challenge of the demand surge for diverse personalized products that come in small volumes. Automatic assembly is one critical process at stake due to the inherent complexity of the number of parts, their shapes, and the randomness of their initial orientations and positions upon arrival at the station. This research introduces a digital twin framework with an information integration layer designed to enhance the flexibility and efficiency of multi-robot operating systems by collaborating with operator skills, addressing the demand for personalized production. The digital twin model allows operators to remotely implant their skills to oversee the assembly process and to control and train the multi-robot operations through an immersive virtual reality interface upon the arrival of new orders without disrupting the current operations of the physical system. The virtual world allows simulations of the algorithms for collisionfree movement and task optimization among multiple robots. Besides, image processing techniques are employed to identify parts arriving in random orientations at the station, providing flexibility during the physical assembly phase. According to the investigation, integrating virtual reality and real-time control challenges data processing and requires robust networking and computational resources. This study contributes to multi-robot operations by providing a scalable and adaptable solution by collaborating with operator skills that enhance both the planning and execution phases of complex assembly processes to new product specifications and changes in design. The methodology presented is foreseen to apply to other manufacturing processes.