Cellular neural network based deformation simulation with haptic force feedback

Abstract

Abstract-This paper presents a new methodology for deformable object modelling by drawing an analogy between cellular neural network (CNN) and elastic deformation. The potential energy stored in an elastic body as a result of a deformation caused by an external force is propagated among mass points by the non-linear CNN activity. An improved CNN model is developed for propagating the energy generated by the external force on the object surface in the natural manner of Poisson equation. The proposed methodology models non-linear materials with nonlinear CNN rather than geometric non-linearity in the most existing deformation methods. It can not only deal with large-range deformations, but it can also accommodate isotropic, anisotropic and inhomogeneous materials by simply modifying constitutive constants.

Learning and Prediction of Soft Object Deformation Using Visual Analysis of Robot Interactions

“…Neural networks are well-fitted for such tasks, as they are computationally simple, have the ability to map complex and non-linear data relationships and have the ability to learn and then predict in real-time the displayed behavior. This explains the interest of researchers from both the deformable object modeling [5,6] and the grasping and manipulation research fields [7][8][9][10][11] into such techniques.…”

“…In the area of deformable object models, object deformation is formulated as a dynamic cellular network that propagates the energy generated by an external force among an object's mass points following Poisson equation in [5]. Greminger et al [6] learn the behavior of an elastic object subject to an applied force, by means of a neural network which has as inputs the coordinates of a point over a non-deformed body (obtained by a computer vision tracking algorithm based on boundary-element method that builds on the equations of the elasticity) and the applied load on the body, and as outputs the coordinates of the same point in the deformed body.…”

“…Neural network architectures are chosen in the context of this work for reasons similar to those mentioned above, namely their capability to store (offline) and predict (online) the complex relationship between the deformation of the object and the interaction parameters at each robotic finger. However, unlike the other neural network approaches encountered in the literature, the proposed approach neither imposes certain equations to model the elastic behavior [5,6] or certain dynamic models at the points of contact [7], nor requires a certain representation of the deformable object [9,11]. The proposed solution combines in an original manner neural architectures to identify an object of interest, to track its contour in visual data and to associate and predict its shape under interactions exercised with a robotic hand.…”

Real‐time biomechanics using the finite element method and machine learning: Review and perspective

“…The main reason to use a mesh is that in many cases it is not possible or it is decided not to directly apply the selected phenomena to the anatomical structure being studied. Such decisions are linked to ethical limitations when the subject is alive; the logistics required to apply and record the results of many configu- An exceptional case is noticed in the earliest work considered in this review 30 , where a dynamic cellular NN is used to mimic the deformations experimented by the liver tissue when external forces are applied to it. In this case, the parameters of the network are not trained and their setting is left to the user.…”

A framework for predicting three-dimensional prostate deformation in real time

“…Moreover, Zhong et al . showed that a neural network can also be used to predict material deformation based on the theory of conservation of energy . In their paper, the potential energy stored in the elastic body is propagated through mass points by the cellular neural network activity.…”