Research Topics

Selected papers listed in references.

Development of Network-Based Multi-Modal Surgery Training System

In general, minimally invasive surgery is the most difficult surgery because a field of view of an endoscope is limited and force sensation from surgical tools such as forceps is poor. Especially in early clinical education for medical students, a virtual reality surgical simulator will be an effective tool.
In this research, we proposed a network-based visuohaptic surgery training system for laparoscopical techniques. To realize this system, we developed a novel volume-based haptic communication approach, which allows participants at remote sites on the network to simultaneously interact with the same target object in virtual environments presented by multi-level computer performance systems, by only exchanging a small set of manipulation parameters for the target object and additional packet for synchronization of status of binary tree and deformation of shared volume model. In addition, we developed a multi-modal surgery training approach. In this approach, we recorded a video from the first person perspective and operation information of surgical tools of an instructor. And then, we displayed the recorded video and the guidance force to trainees. We constructed a prototype surgery training system, and effectiveness of our approach was confirmed.


Expression of Cystohepatic Duct Anomaly using Modular Structured Organ Model in a Laparoscopic Surgery Simulator

Cystohepatic duct anomaly is often found in a laparoscopic cholecystectomy and these cause certain risks in this surgical procedure. Although anomalies can be detected by preoperative CT or MRI examinations, it is hard to operate smoothly under the actual surgical view on a 2D monitor. Surgical training for such special cases is one of the advantages of a VR surgical simulator. This is especially true in early clinical education for medical students; the simulator can be an effective tool and teaching resource both visually and haptically.
In this research, we have developed a novel laparoscopic surgical simulator. In this simulator, each part of varied objects is maintained by modularized multi-resolution models, and any models which consist of arbitrary combination of these modules can be produced with low computational cost.


Acceleration of Deformation Simulation by Multi-Rate Online Re-Mesh Approach and GPU

In this research, we proposed an approach to generate high quality force and deformation using a novel automatic space and time adaptive level of detail technique in combination with a parallel computation using a graphics processing unit (GPU). The elastic object is represented by a multi-resolution hierarchy of tetrahedral adaptive volume mesh. The tetrahedral adaptive volume mesh and the time step of the deformation simulation are locally refined by online re-mesh and multi-rate simulation to concentrate the computational load into the regions that deform the most.
In order to compute the online re-mesh and multi-rate deformation simulation on a GPU efficiently, we propose a novel data structure which consists of an extended and transposed connection table, a node list and a separated mass list. This effective computation is achieved by the relocatability of the connection table and optimized memory access at the computation of both deformation and re-meshing.


Adaptive and Embedded Deformation Model for Complex Inhomogeneous Elastic Objects

In this research, we proposed a novel approach to haptic interact with complex inhomogeneous elastic objects like organs. In surgery simulations, representation of complex inhomogeneous elastic objects which have various physical properties and geometries is required. Also, consideration of geometric nonlinearity is required.
To solve this problem, we propose an adaptive and embedded deformation model. This approach uses both an online re-mesh deformation model and a corotational FEM model together. Our approach can update an inverse of a global stiffness matrix of a corotational FEM model quickly during the online re-meshing. In addition, an efficient computation algorithm by GPU is also proposed.


Acquisition of Elastically Deformable Object Model Based on Measurement

In this research, we proposed an approach to generating impulse response deformation model (IRDM) based on the measurement of real object. In this approach, the external force on the surface of the object is applied by air jet, and the deformation of the object is measured using markers and stereo cameras. Through implementation of the measurement system and experiments, it was proved that models that behave similarly to the real objects can be obtained. Also, the evaluation of the model using subjects was performed, and the similarity of the model to the real object from the aspect of perception was confirmed.


Impulse Response Deformation Model

In this research, we proposed an approach to haptic interaction with dynamically deformable object. In this approach, the behavior of an object is defined by a set of temporal deformation patterns after impulse force is applied to each degree of freedom, which we call impulse response deformation model. Deformation resulting from interaction is obtained by computing convolution of the model and the history of interaction force. The time complexity of computing interaction force is independent of the complexity of the model. This feature is advantageous for time-critical applications.


Automultiscopic Display

In this research, we developed two types of automultiscopic display. In the display, different and independent views are presented to different viewpoints. Our first display system was realized by rotating a flat-panel display whose viewing angle is strictly limited, or has sharp directivity. Latter display system was realized by rotating a mirror on two axes and switching images directionally at high speed.


Improvement of Cloth Behavior Simulation Using Evaluation Experiments in Vacuum

In spite of many dynamic cloth simulation methods have been proposed in computer graphics and textile engineering fields, there are no quantitative evaluation method for comparing real and virtual with various dynamic cloth deformations that can be seen our daily life.
In this paper, the evaluation experiments in vacuum is proposed and several results of experiments are evaluated and discussed.


Construction of a Virtual Fashion System

We constructed a virtual fashion system which enables virtual fitting of clothes. To realize this system, we proposed several novel approaches: a fast and stable cloth behavior simulation, fast collision detection approach between the cloth and human, a motion capture system using gyro sensors, a large screen, and a virtual mirror.