Large deformation object modeling using finite element method and proper orthogonal decomposition for haptic applications

Haptic Rendering has a wide range of application areas from entertainment to sur-gical simulations, from CAD/CAM to rehabilitation purposes. In each of these applications the main problem is the real-time performance of the overall system. The balance between realistic force reflecting object models and system performance is an important factor in hap-tic simulations. In general, two methods are used in haptic rendering: particle based and fi-nite element method based models. Although particle based methods are easier to implement and faster in computational manner, they have a limited capability to reflect the underlying mechanics. In order to obtain physically realistic models, finite element methods are widely used. Since finite element methods are computationally expensive further decomposition methods are applied to increase the real-time performance. In this work, two large deforma-tion object models are developed for force reflecting purposes and in order to increase the real-time performance, the proper orthogonal decomposition (POD) method is applied to the nonlinear models and performance increase in the overall system is investigated. By the use of finite element methods, proper orthogonal decomposition method is applied to the solutions of the large deflection problem of a beam. Even though the original structure behaves in a nonlinear manner, the POD method resulted in very accurate and fast solutions in real time. Experiments are carried out by using the haptic robot arm of Sensable Technologies. The numerical solutions and animations for haptic arm application will be presented.