In recent years, humanoid cooperative robots for industrial use have been put into practical use in the field of factory automation (FA). In this context, dual-arm robots have the advantage of being able to directly replace human workers. The general operation of robots is pick-and-place, which involves grasping, lifting, moving, and lowering an object to a target position, but this operation is limited by the payload capacity of the robot. In this study, we focus on grasp-less manipulation using gravity and frictional force, such as pushing and rolling, as well as grasp-less handling that is performed using tools without directly touching the object. This paper describes a study of motion accuracy using a gyroscopic sensor in the control of two-axis plate rotary motion using a dual-arm industrial robot, with the aim of improving grasp-less handling technology in the industrial field. The gyroscopic sensor attached to the plate showed a discrepancy between the teaching point and the actual motion. The data showed a large amount of noise, and frequency analysis using FFT analysis was used to determine the peaks. The natural frequency of the plate itself was also analyzed using FFT analysis, and it was found to be consistent with the noise from the gyroscopic sensor, allowing for the identification of the cause.