This paper presents a design and control method of 2-DOF (degrees of freedom) SCARA-type robot arm utilized for physical therapy. Fuzzy algorithms are applied to control the stiffness of each joint. Load cells are utilized as force sensors for feedback signals of joints. Practitioners will guide trajectory of the end-effector of the arm robot based on the exercise physiotherapists with predetermined trajectories. The results of
controlling the stiffness show that the end-effector of the robot is tracking to the desired trajectory with the approximate accuracy of 99.8%.