.png){kind=icon}
.png){kind=icon}
Return
Digital twin-driven multi-agent collaborative online optimization of production regulation for smart reconfigurable manufacturing systems with human-robot collaboration
.png){kind=icon}
PRE
AI
M
S
L
W
Y
X
Z
DOI：10.1016/j.rcim.2026.103322.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• This study is the first to explore digital twin-driven online optimization for smart reconfigurable manufacturing systems with human-robot collaboration while jointly considering manufacturing flexibility, position flexibility, and human-robot configuration flexibility. • A novel markov decision process is constructed for smart reconfigurable manufacturing systems with human-robot collaboration, featuring an order size-independent state representation with strong generalization, and a rule-based action set with explicit semantics and computability. A normalized Tchebycheff reward aggregation method is employed to simultaneously optimize enterprise-level objectives (Makespan) and human-centered objectives (human balance index). • A multi-agent collaborative online optimization mechanism is proposed, and its advantages in scalability, fault tolerance, and local disturbance isolation over single-agent approaches are demonstrated. • A multi-objective multi-agent twin delayed deep deterministic policy gradient algorithm (MO-MATD3) is designed, extending dual critics, delayed updates, and target policy smoothing to multi-objective and multi-agent scenarios.
Keywords:
digital twin
multi-agent collaboration
online optimization
human-robot collaboration
smart reconfigurable manufacturing systems
Journal
R
IF:
11.4
Papers:
3.3K
Citations:
1.3W
Organization
Cited Papers