Analysis of human arms’ movements using Therbligh principles is now used to study the movement therapy for stroke patients. The study aims to design robot arm as a preliminary result to design artificial shoulder-attached-dual-arm robot to imitate human arms’ movements that will help stroke patients for movement therapy. For modeling the dual human arm movements, a DH-parameter based on forward-kinematics of the arms is used. Human hands hold objects with various weights and volumes. We take into account the unpredictability center of mass of the entire arms as uncertainty. The model must follow human’s base behavior. Therefore, human arms anthropometry is required to determine the movement’s parameters. Movement limitation of stroke patients must also be considered as the limitation. Result simulations are presented and will be used as the model for movement controller. Synchronized and symmetrical arms’ movement is expected to improve the balance of the brain’s control system.