Page 148 - Human Inspired Dexterity in Robotic Manipulation

144 Human Inspired Dexterity in Robotic Manipulation

mode. In all, the range of graspable objects is wide and includes soft, rigid,
deformable, fragile, small (length <30 mm), large (>80 mm long), thin
(<0.5 mm), and heavy (>3 kg) objects.

7.5 CONCLUSIONS

This chapter presented a novel fluid fingertip based on the hybrid soft and
hard structure of human fingertips. The fluid fingertip was constructed from
a deformable rubber bag filled with incompressible fluid and a rigid part
inside the fluid. The fluid softness reduces contact impact, allows safe inter-
action with humans and surrounding matter, adapts to the shape of target
objects, and has high robustness. Moreover, the fluid incompressibility pro-
vides uniform contact pressure and local passive compliance, which are help-
ful to grasp fragile objects. Because fracture is governed by pressure or stress
domain, the profile of the fluid pressure corresponding to contact pressure
can identify the object state related to fracture. When compressing gel-like
objects, such as tofu, a linear behavior is observed before yielding and frac-
ture. A method to detect the linear compressible behavior for fracture avoid-
ance based on the comparison of fitting accuracy of simple and complex
models was presented. The fluid part is effective for grasping several kinds
of objects. However, its drawback is it cannot compensate for the weight
of a heavy object. Conversely, the rigid part of the fluid fingertip is useful
for handling the weight of a heavy object. Moreover, softness, not only
on surfaces but also at the joints, is also effective. An intentional contact with
an environment can extend the manipulation and grasping ability of robotic
hands, as well as minimizing the number of actuators, thus resulting to a
more simple structure. With this in mind, this chapter also presents a novel
underactuated soft gripper in which fluid fingertips were embedded. Utiliz-
ing contact with an environment, three different grasping modes (parallel
gripping, pinching, and enveloping modes) were realized with one actuator.
The experimental validation showed the novel gripper is capable of handling
a wide variety of objects.

REFERENCES
[1] R. Adachi, Y. Fujihira, T. Watanabe, in: Identification of danger state for grasping del-
icate tofu with fingertips containing viscoelastic fluid, 2015 IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS), IEEE, 2015, pp. 497–503. Avail-
able at: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber¼7353418.