Intelligent collaboration: a predictive neural network for dynamic rescheduling in robotic cells

The integration of collaborative robots (cobots) into industrial environments necessitates advanced task allocation strategies to optimize performance and enhance worker satisfaction. Traditional static task allocation methods often fall short in adapting to dynamic operational conditions and addressing the cognitive load on human operators. This study introduces and evaluates a novel adaptive rescheduling mechanism incorporating a neural network-based predictive model. The aim is to address these limitations. The experimental campaign, involving the assembly and disassembly of a multi-component box, tested the system’s effectiveness in real-world scenarios. Results indicate that the adaptive rescheduling mechanism significantly reduced the makespan compared to static allocation methods. This demonstrates the improved operational efficiency. Additionally, human factors were positively impacted, with a notable reduction in participant frustration as measured by the NASA-TLX questionnaire. These findings highlight the potential of predictive analytics in optimizing task allocation, suggesting that adaptive rescheduling mechanisms not only enhance productivity but also contribute to a more supportive and manageable work environment. This research underscores the value of integrating advanced predictive techniques into human-robot collaboration systems and offers a foundation for further exploration and refinement of such approaches to improve both performance and worker well-being in industrial settings.