Abstract: The determination of the number of industrial robots at each workstation of a production line is the basis for promoting manufacturing intelligence. Based on the principle of reliable redundancy design and the probability theory, this paper first investigates the relationship between the number of parallel operating robots and the number of configured backup robots at each workstation of a production line considering industrial robot failures to obtain a formula for calculating the operating time of industrial robot workstations. On this basis, based on the theory of assembly line balance, the number of parallel operating robots and backup robots at each workstation are regarded as decision variables to establish an optimal quantity configuration model for industrial robots at each workstation of a simple operating robot production line with the objective of minimizing the investment cost of industrial robots and maximizing the balance rate of the production line, under the preset capacity requirements. Genetic algorithm is applied to solve the model in an actual case. After optimization, the balance rate of parallel robots at workstations is increased from 89% to 97.21%, while the design cost of a production line is reduced by 9% according to the standard operation time t considering robot failures.