Parallel Robot Scheduling with Genetic Algorithms

Abstract

There are some main goals in parallel robot scheduling. Those are total completion time, maximum earliness, and maximum tardiness. According to the theoretical viewpoint, parallel robot scheduling is a generalization of the single robot scheduling and a special study of the flow shop. From the practical viewpoint, solution techniques are useful in the real-world problems. Parallel robot scheduling has to deal with balancing the load in practice. Scheduling parallel robot may be considered as a double-step. First, which jobs are allocated to which Robot. Second, allocated jobs sequence. Also, preemption plays a more important role in parallel robot scheduling. Robots may be identical or not. Jobs have a precedence constraint. For all problem structures may be applied different solution techniques for instance algorithms, search algorithms or artificial intelligence techniques. In this chapter we interest in different solution techniques for parallel robot scheduling. In this chapter, first, a genetic algorithm is used to schedule jobs that have precedence constraints minimizing the total earliness and tardiness cost and maximum flow time on nnumber of job and m-number of identical parallel robots. The second one is without precedence constraint. There are many algorithms and heuristics related to the scheduling problem of parallel machines and robots. In this study, a genetic algorithm has been used to find the job schedule, which minimizes maximum flow time. We know that this problem is in the class of NP-hard combinatorial problem. (Kanjo & Ase, 2003) studied about scheduling in a multi robot welding system. (Sun & Zhu, 2002) applied a genetic algorithm for scheduling dual resources with robots. (Zacharia & Asparagatos, 2005) proposed a method on GAs for optimal robot task scheduling. In this study, the job with n-number of precedence constraints is assign ed minimizing mean tardiness on m-number of parallel robot using genetic algorithms. (Koulamas,1997) developed a heuristic noted hybrid simulated annealing (HAS) based on simulated annealing. (Chen et al.,1997) has developed highes priority job first (HPJF) method, which is based on extension of the WI method extended with various priority rules such as minimum processing time first (priority = 1/processing time), maximum processing time first (priority=processing time), minimum deadline first (priority=1/due date) and maximum deadline first (priority = Due date). (Alidaee & Rosa, 1997) proposed a heuristic which is based on extending the modified due date (MDD) method belonging (Baker &