Segmentation of Bimanual Virtual Object Manipulation Tasks Using Multifinger Haptic Interfaces

This paper describes data processing using a haptic interface for virtual object grasping tasks. The haptic interface known as MasterFinger-2 (MF-2) is used. This haptic device enables the simulation of object manipulation using the thumb and index finger, which are inserted within thimbles located at the far end of the haptic interface. The experiments were conducted using a setup that provides the operator with two MF-2s. This setup enables bimanual virtual object manipulation. The haptic device provides data about the angular position of all joints and data from four contact sensors located within the thimbles. The values obtained from the angular positions of the joints enable a calculation of the position and orientation of the thimbles, whereas the values obtained from the contact sensors are used to formulate an approximate calculation of the user's force during object manipulation. This information regarding forces enables the segmentation of manipulation tasks and is essentially based on the rising edges produced by the contact sensor signals. After performing task segmentation, the state of the task in progress is detected. This facilitates improvements in user interaction because restrictions can be established in the relationships between the objects that change dynamically depending on the stage of manipulation. This information allows for an improvement in the realism of the simulation because the forces reflected take into account how the relationships between objects vary over the course of the task. Task segmentation is also used to ascertain objects' physical properties. These features are then used to simulate virtual objects with the same properties.