Abstract: Method design and overall optimization of task assignment for a new type of serial and parallel dual-robot associated processing system are carried out. Two algorithms of ant colony optimization, the approximate nondeterministic tree search (ANTS) and the max-min-ant-system (MMAS) are used to be the optimization methods for task assignment. Local search is adopted in MMAS in order to get better local-best-solution from the path construction process. Simulation results and their comparison with the previous relevant study results show that the iteration convergence speed of MMAS in the optimizing process is faster than ANTS, and the quality of the final optimal solution obtained after a period of time for exploration is also better than ANTS. The combination of MMAS and local search further improves the quality of the final solution than using MMAS only. Evolution curves demonstrate the adaptability and superiority of the algorithms for task assignment and optimization of the system. Experiment results further validate the effect of optimization in comparison with traditional combinatorial optimization method.

Key words: Ant colony optimization, Approximate nondeterministic tree search, Local search, Max-min-ant-system, Serial and parallel dual-robots system, Task assignment