Computer Graphics in Spain: a Selection of Papers from CEIG 2009Haptic rendering of objects with rigid and deformable parts
Graphical Abstract
Research highlights
► The handle of a tool is always rigid due to the grasp. ► Coupling the device to the handle allows rendering contact with complex tools. ► Force feedback is computed using a linear model of this coupling. ► This model is aware of the dynamics and contact state of the tool.
Introduction
Haptic rendering is the computational technology that allows us to interact with virtual worlds through the sense of touch. It relies on an algorithm that simulates a virtual world in a physically based manner and computes interaction forces, and on a robotic device that transmits those interaction forces to the user. Haptics science is a multidisciplinary field that brings together psychophysics research for the understanding of tactile cues and human perception; mechanical engineering for the design of robotic devices; control theory for the analysis of the coupling between the real and virtual worlds; and computer science, in particular computer graphics, for the simulation of the virtual world and the design of the haptic rendering algorithm [29].
In this paper, we focus on the haptic rendering of objects that are composed of both rigid and deformable parts. Many of the objects in the real world, including our own flesh, fit this description. In particular, our haptic rendering algorithm serves as the main building block for model-based computation of direct-hand haptic interaction. The hand is the main tactile sensor used by humans to capture information from the world [25]. It allows us to manipulate the world and provides us with two-way interaction with our environment. Despite the importance of the human hand for haptic interaction, current haptic devices and haptic rendering techniques suffer important limitations that have not allowed hand-based virtual touch to reach its full potential. Haptic rendering is typically carried out either through a pen-like robotic device, or through vibrotactile devices. In addition to hardware limitations, most haptic rendering algorithms are limited to point-based interaction between the user and the objects in the environment. There are some notable exceptions, both devices (e.g., exoskeleton structures [23], [7], [33]) and object-object rendering algorithms, also called 6-Degree-of-Freedom (6-DoF) haptic rendering [31], [39], [6].
We follow a strategy for haptic rendering of tool-based contact where a virtual replica of the tool is simulated in a virtual world, and haptic interaction takes place through a viscoelastic coupling between the haptic device and the tool [11]. Following this strategy, direct-hand touch can be rendered by simulating a virtual hand as the tool (as shown in Fig. 1), and coupling the device to this simulated hand. In Section 3, we summarize the adaptation of a multirate 6-DoF haptic rendering algorithm to model multipoint contact with objects composed of rigid and deformable parts, such as a hand. This algorithm does not suffer from the limitations of traditional single-point contact models for capturing the interaction between soft fingers and the environment.
In Section 4, taking the hand as an example model, we describe how to connect rigid and deformable components and how to actuate them through standard haptic devices. We exploit the position and orientation tracked by a haptic device to guide the rigid components of the virtual hand, and we model the coupling between the rigid and deformable components of the hand using stiff connections and implicit integration. As a first step toward full-hand haptic rendering, we consider the hand to be a shape formed by one rigid component and elastic flesh, and we do not take into account the articulations of the fingers.
A computational algorithm, described in Section 5, allows us to simulate a virtual hand using existing implementations of rigid-body and deformable-body simulations as black boxes. The use of implicit integration couples the degrees of freedom of the rigid and deformable components of the virtual hand, but our algorithm allows us to work around these couplings and still use existing implementations as black boxes.
We show our haptic rendering algorithm applied to the interaction between a hand model and other complex deformable objects with friction (see Fig. 5), with several moving deformable objects (see Fig. 6), and with self-collisions between the fingers (see Fig. 3). In Section 6, we also analyze the impact of various parameters of the virtual scene on the computational cost of our haptic rendering algorithm.
Section snippets
Related work
The rendering of haptic interaction with interesting virtual environments relies to a large extent on the ability to simulate efficiently effects such as deformations and contact among the objects in the virtual environment. We refer the reader to surveys on those topics for more information [18], [35], [28], [47]. The efficiency of the techniques for solving contact and deformation problems is growing steadily (see [5], [45], [20], [22], [43], [40] for some recent examples), but the complexity
Overview of the haptic rendering algorithm
Haptic rendering typically distinguishes interaction through a tool object from direct interaction with the hand. Tool-based rendering algorithms simulate a virtual model of the tool and compute contact between the tool and other objects in the environment using robust contact-modeling algorithms. Direct-hand rendering algorithms, on the other hand, track points on the finger tips and model contact forces based on single-point penalty-based models (see [13] for an example). Here, we present a
Hand model with rigid and deformable parts
To model the elasticity of the hand (our deformable tool), we use a linear co-rotational finite element model [34]. To apply the haptic rendering algorithm described in the previous section, we need to identify a rigid handle for each rigid frame tracked by the haptic device. In our case, using the Phantom Omni as the haptic device, we need one rigid handle. We choose to locate the rigid handle at the palm of the hand model. We need to define the rigid-body properties of the handle, but we do
Implicit solution for rigid and deformable parts
As part of the haptic rendering algorithm, we need to solve a constrained dynamics problem involving the deformable hand, the rigid handle, and possibly other dynamic objects in the virtual environment. Then, the coupling force between the tool-hand and the handle must be linearized w.r.t. the handle state, accounting for contact constraints and inertial effects. Both the constrained dynamics solver and the linearization require an efficient solution to the coupled dynamics of the tool-hand and
Experiments
We executed our experiments on a quad-core 2.4 GHz PC with 3 GB of memory (although we used only two processors, for the visual and haptic loops) and a GeForce 8800 GTS. We manipulated the models using a Phantom Omni haptic device from SensAble Technologies.
We tested our algorithm with some 3D scenarios consisting of one deformable tool (a model of a human hand, in most experiments) and a set of rigid and deformable objects. A video showing some of our experiments can be downloaded from //www.gmrv.es/cgarre/DivX_GarreOtaduyCG.avi
Discussion and future work
We have presented a haptic rendering algorithm for computing force interaction through deformable tools such as a human hand. Our algorithm differs from previous rendering algorithms in that it employs a deformable tool model that couples rigid and deformable components, and we have presented a computational algorithm that efficiently solves implicit integration under such coupling, thus allowing effective haptic rendering.
Our proposed algorithm constitutes a step forward toward full-hand
Acknowledgements
We would like to thank the anonymous reviewers for their helpful comments, and the GMRV group at URJC. This work was funded in part by the URJC - Comunidad de Madrid Project CCG08-URJC/DPI-3647 and by the Spanish Science and Innovation Dept. Projects TIN2009-07942 and PSE-300000-5.
References (49)
- Adachi Y, Kumano T, Ogino K. Intermediate representation for stiff virtual objects. In: Virtual reality annual...
- Astley OR, Hayward V. Multirate haptic simulation achieved by coupling finite element meshes through norton...
- Baraff D, Witkin AP. Large steps in cloth simulation. In: Proceedings of ACM SIGGRAPH,...
- Barbagli F, Prattichizzo D, Salisbury KJ. Dynamic local models for stable multi-contact haptic interaction with...
- Barbič J, James DL. Real-time subspace integration for St. Venant-Kirchhoff deformable models. In: Proceedings of ACM...
- et al.
Six-DoF haptic rendering of contact between geometrically complex reduced deformable models
IEEE Transactions on Haptics
(2008) - et al.
The rutger master ii—new design force-feedback glove
IEEE Transactions on Mechatronics
(2004) - Çavuşoğlu MC, Tendick F. Multirate simulation for high fidelity haptic interaction with deformable objects in virtual...
- Cobos S, Ferre M, Sanchez Uran M, Ortego J, Pena C. Efficient human hand kinematics for manipulation tasks. In:...
- Colgate JE, Brown JM. Factors affecting the z-width of a haptic display. In: IEEE international conference on robotics...
Local model of interaction for realistic manipulation of rigid virtual worlds
International Journal of Robotics Research
Realistic haptic rendering of interacting deformable objects in virtual environments
Proceedings of IEEE TVCG
Matrix computations
6-DOF haptic rendering using spatialized normal cone search
IEEE TVCG
Six-degree-of-freedom haptic rendering using incremental and localized computations
Presence
Cited by (14)
A 6-dof zero-order dynamic deformable tool for haptic interactions of deformable and dynamic objects
2020, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)Multirate haptic rendering using local stiffness matrix for stable and transparent simulation involving interaction with deformable objects
2020, IEEE Transactions on Industrial ElectronicsContinuous Collision Detection for Virtual Proxy Haptic Rendering of Deformable Triangular Mesh Models
2019, IEEE Transactions on HapticsData-driven modeling and rendering of force responses from elastic tool deformation
2018, Sensors (Switzerland)