Throughput maximization by balancing, sequencing and coordinating motions of operations in multi-robot stations

In this paper we present an optimization method for throughput maximization by balancing, sequencing and coordinating motions of operations in multi-robot stations. Maximizing throughput by cutting cycle time in robot stations is an important area within virtual manufacturing. However, the combinatorial problem we solve is general and the method proposed can be used also in other industrial context and sectors. The problem consists of distributing a set of tasks/operations among robots, finding the sequence in which they should perform these tasks in a way to avoid collisions and minimize the time from the start of the first task to the end of the last one (cycle time/makespan). Due to the large number of degrees of freedom, there is a lack of direct or complete method of practical relevance. To resolve this problem, we design a lazy method which iteratively solves a relaxed problem and updates problem information. The optimization step is done by a genetic algorithm wi th a local search method. The method is tested on a stud welding station case with up to four robots and 80 studs, showing that it is possible to achieve good solutions. Furthermore, a comparison with two greedy approaches is also carried out. One of the greedy approaches is a decoupled strategy where the station is optimized in a first step with respect to balancing and sequencing and then the collisions among the robots are resolved by coordination