The invention discloses a four-axis iterative learning control industrial robot design method. A three-dimensional part that needs pre-forming is converted into a homogenous curve cluster trajectory,and then curves are sequentially extracted as desired trajectories of system tracking, so that the iterative learning control is separately applicable and therefore is effectively applicable to such trajectory groups so as to complete the machining of the three-dimensional part. Through the effective inheritance of the control information obtained from learning the previous desired trajectory, sothat the initial iterative control information when the current desired trajectory is tracked is no longer learned from scratch, the learning speed of the system is greatly increased, the number of iterations is reduced, and thus the work efficiency is substantially improved. Moreover, the industrial robot system applying the design method can machine the three-dimensional parts whose base trajectory cannot be machined by a conventional rotary body device, greatly improves the diversity of industrial parts, and is very suitable for the production of individualized parts in the industry. .