Page 161 - Human Inspired Dexterity in Robotic Manipulation

158 Human Inspired Dexterity in Robotic Manipulation

disturbance

(A)

disturbance

disturbance
(B)
Fig. 8.8 Comparison of two controllers. (A) Absolute impedance controllers and
(B) blind grasping controllers. Both controllers can stabilize the manipulation.
However, the robustness of the impedance controller is weaker than that of the
blind grasping.

fingers, the other finger followed the disturbed finger to preserve the relative
distance. The hand secured the grasp even if the equilibrium point was chan-
ged. Therefore, the blind-grasping controller is more robust to external dis-
turbances than the joint-impedance controller.

8.3.4 Superposition of Lower Priority Tasks to Stable Grasping

In-hand manipulation is a skill based on grasping. In this section, a method
for in-hand manipulation will be discussed by using the remaining control
DOF after grasping. Fig. 8.7 shows the state space of the hand-object system.
The state of the system moves on a manifold that satisfies the contact con-
straints if the contact force remains within the friction cones, and the
in-hand manipulation constrains the state on the subspace of the stable-grasp
manifold.
Position and orientation control, as lower priority tasks, can be executed
by superposing the virtual potentials in the stable grasping controller (8.6), as
shown in Fig. 8.9. For example, the position controller is given as follows:
T
u pci ¼ J ðx pci x *Þ (8.8)
pci
pci
where x pci and x pci * are current and desired positions of the grasped object
and J pci is the Jacobian matrix of x pci with respect to the joint space of the ith
finger. Likewise, the orientation control can be obtained as follows:

T
u oci ¼ J ðθ oci θ *Þ, (8.9)
oci pci