Robust Control of OpenMANIPULATOR-X Using Reinforcement Learning

Abstract

To tackle the complexities of production assembly line optimization, we present a mixed-integer programming formulation for production assembly line optimization, minimizing costs of workstation construction, task-specific equipment, and times buffer while ensuring production demands are met. A key focus of this research is to achieve a balanced workload among worksta-tions, which is complicated by the variability in workstation reliability characterized by Mean Time to Failure (MTTF) and Mean Time to Repair (MTTR).We address this challenge through the development of a mathematical model that incorporates task precedence relationships and actual cycle time constraints. We validate our proposed method through numerical experiments using real-world datasets, demonstrating considerable practical benefits for complex production line. The result demonstrates efficiency improvements and cost reductions, which enhance the overall performance of the assembly line. The proposed model incorporates the impact of machine failures, integrates time buffers, and accommodates the requirements of various tasks, contributing to production optimization. It offers practical guidance for practitioners seeking to improve production efficiency. Furthermore, we introduce the Precedence-Driven Task Grouping (PDTG) method, which enhances the traditional Ranked Positional Weight (RPW) method by offering wider flexibility in task assignment. This approach reduces the number of workstations, improves task allocation efficiency, and minimizes idle time. © 2025 The Authors.