Exact and heuristic methods for solving the Robotic Assembly Line Balancing Problem

• Published in: European journal of operational research Vol. 270; no. 1; pp. 146 - 156
• Main Authors: Borba, Leonardo, Ritt, Marcus, Miralles, Cristóbal
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2018
• Subjects: Beam search; Branch-bound-and-remember; Production; Robotic assembly line balancing; Branch-bound-and-remember; Beam search; Production; Robotic assembly line balancing
• ISSN: 0377-2217; 1872-6860; 1872-6860
• DOI: 10.1016/j.ejor.2018.03.011

•We propose new algorithms for minimizing the cycle time in robotic assembly lines.•The first is a branch-bound-and-remember method with cyclic best-first search.•The second is an iterative beam search.•Both methods use newly proposed lower bounds and dominance rules.•Our methods improve the results from the literature and are faster. In robotic assembly lines, the task times depend on the robots assigned to each station. Robots are considered an unlimited resource and multiple robots of the same type can be assigned to different stations. Thus, the Robotic Assembly Line Balancing Problem (RALBP) consists of assigning a set of tasks and a type of robot to each station, subject to precedence constraints between the tasks. This paper proposes a lower bound, and exact and heuristic algorithms for the RALBP. The lower bound uses chain decomposition to explore the graph dependencies. The exact approaches include a novel linear mixed-integer programming model and a branch-bound-and-remember algorithm with problem-specific dominance rules. The heuristic solution is an iterative beam search with the same rules. To fully explore the different characteristics of the problem, we also propose a new set of instances. The methods and algorithms are extensively tested in computational experiments showing that they are competitive with the current state of the art.