Copyright is held by the author / owner(s).
SIGGRAPH 2008, Los Angeles, California, August 11–15, 2008.
MeisterGRIP: Cylindrical Interface for Intuitional Robot Manipulation
Shuji Komeiji*1 Katsunari Sato*2 Kouta Minamizawa*3
Naoki Kawakami*5 Susumu Tachi*6
The University of Tokyo
In recent times, robot manipulations have been used to perform
operations in extreme environments, communicate with people in
remote locations, and realize entertainment systems that control
virtual robots. In all these activities, an interface for robotic hand
and arm manipulations is important to interact with the external
environment [Bar-Cohen, 2000]. Conventional interfaces consist
of a few levers and switches; unfortunately, such interfaces are
difficult to control, and the users must have prior experience as far
as the operation of these interfaces is concerned. The other type of
interfaces reflects the postures of the user’s hand and arm directly
in the robotic hand and arm so that the robots are capable of
intuitive manipulation. However, these devices have a trouble of
wearing and restrict the hand size of the user.
We believe that an interface for robotic hand manipulations
should be developed in such a manner that anyone can use it
intuitively and easily. Therefore, we developed a novel interface
called “MeisterGRIP.” This device measures the user’s grasping
conditions and reflects this information in the robotic hand and
arm control. The user is simply required to grasp the device;
therefore, the complexity of the setup is reduced and there are
fewer restrictions on the user. Furthermore, by reflecting the
grasping force of the user directly in the robotic hand and arm,
intuitive manipulation is possible. MeisterGRIP allows robot
manipulations to be easily conducted, and it can be widely used in
general households. In the future, it will be possible to travel and
interact with objects at remote locations by using our proposed
device, even if the user is sitting on a sofa in his/her living room.
In order to measure the user’s grasping conditions, the force
vector distribution caused by the user’s grasping is used. When a
user grasps our proposed device, the device recognizes the five
fingers and the palm of the user as six input positions from the
pattern of the force vector distribution. Then, the three-axis force
vectors are measured at each input point and relayed to each
finger and wrist of the robotic hand . The input points could be set
at any position on the device surface; this allows the user to grasp
the device at any position in any posture. The individual
differences among the hand sizes are also cleared by this feature.
In order to measure the force vector distribution, we applied our
vision-based tactile-sensing technology [Kamiyama, et al., 2004].
The force vector distribution is calculated by capturing the
movement of colored markers on the surface of an elastic body
with a camera. MeisterGRIP has a cylindrical shape to support
any grasping posture, as shown in Figure 1. Two CCD cameras
and two spherical mirrors are attached at the end and the center of
the cylindrical body, respectively. The mirrors enable us to capture
all the markers on the inner surface of the cylindrical body using a
Figure 1: The figure on the left represents MeisterGRIP grasped
by a hand. The figure on the right represents the inner
configuration of the device.
Figure 2: Manipulation of a robot system using MeisterGRIP.
3 Summary and Application
We propose a novel cylindrical interface called MeisterGRIP that
treats the force vector distribution. This elastic device allows
institutional and dexterous robot manipulation based on
vision-based tactile-sensing technology. Furthermore, it provides
universal manipulation that can tolerate the personal differences in
hand sizes and grasping postures of the users.
This device is proposed for use in a cockpit to manipulate a robot
in both real and virtual environments (Figure 2). We attached a
stick to the device and fixed it to the ground. By using the force
information measured from MeisterGRIP, the user can manipulate
not just the robotic hand, but the robotic arm. To manipulate the
robotic arm, we use the six-axis force information caluculated
from force vector distribution. Using this device, the users would
feel as if they become a robot in remote environment by only the
information of user’s grasping.
BAR-COHEN, Y. 1999. Haptic Interfaces, In Automation, Miniature
Robotics and Sensors for Nondestructive Evaluation and Testing,
Volume 4 of the Topics on NDE Series, ASNT, Columbus.
KAMIYAMA, K., KAJIMOTO, H., KAWAKAMI, N., AND TACHI, S. 2004.
Evaluation of a Vision-based Tactile Sensor, In Proceedings of
International Conference on Robotics and Automation (ICRA2004).