1. Dual-Tendon Routing for
hand wearable robots

1.1 Issue of previous under-actuated tendon routings in robot applications

Previous under-actuated tendon routings can be categorized as below:

1.1.1 Under-acutated tendon routing
that uses movable pulley
(UTR-M)

Fig. R1.1.1 Schematic of UTR-M

(Pros) Low friction
(Cons) Bulky

1.1.2 Under-acutated tendon routing
that uses fixed pulleys
(UTR-F)

Fig. R1.1.2 Schematic of UTR-F

(Pros 1) Compact
(Pros 2) Cause less elongation at the tendon
(Cons) High friction due to curved tendon path

1.1.3 Under-acutated tendon routing
that uses specific actuator
(UTR-S)

Fig. R1.1.3 Schematic of UTR-S

(Pros) Compact and low-friction
(Cons1) Cause high elongation at the tendon
(Cons2) Actuator size highly depends on the tendon excursion length.

In summary

UTR-M is bulky. It has NOT been used in hand-wearable robots yet.
UTR-F causes high friction at the tendon.
UTR-S suffers from high elongation at the tendon.

-> UTRs that can solve friction and elongation issues are required. Why friction and elongation are not preferred is in 1.2 and 1.3 of this page.

1.2 Friction is not preferred in tendon-driven robots (Issue of UTR-F)

Video. R1.2.1 Video showing how the finger moves when the robot pulls and releases the flexion tendon under different conditions.

Left: the case with high friction
Right: the case with low friction
The finger does not recover to its initial position when the friction is too large.

Video. R1.2.2 Video showing the possible problem when the friction at the flexor is large

Extensor requires a large amount of tension to overcome the friction.
(Sometimes) this phenomenon fails to assist in extension. This is particularly common when the user has joint spasticity, leading to a clenched fist.

1.3 Tendon elongation is not preferred in tendon-driven robots (Issue of UTR-S)

*Elongation is not negligible because it is proportional to the tendon length; the tendon length in Exo-Glove is about 1m because it locates the motor far from the wearing part.

Elongation makes it difficult to estimate the joint position. Exo-Glove does not have joint angle sensors so it`s clitical.
Elongation also increases the actuator size because it requires the robot to pull more tendons to compensate for the elongation.
In other words, higher elongation means lower actuation stiffness. The drawback of low-actuation stiffness is well explained in [R1.1]
[R1.1]Kim, Y. J. (2017). Anthropomorphic low-inertia high-stiffness manipulator for high-speed safe interaction. IEEE Transactions on Robotics, 33(6), 1358-1374.

1.4 How the paper solves the issue: Dual-tendon routing (DTR)

Find the whole possible under-actuated tendon routing by combining UTR-F and UTR-S; we named them as Dual-Tendon Routing.
*UTR-M is excluded because it necessarily increases the end-effector size.
Combining UTR-F and UTR-S gives 2^(n+1)-1 number of possible routings for n finger application.
The combination gives us a chance to find the tendon routing with the lowest friction and elongation at the tendon.

1.5 Derivation of possible number of DTRs

Fig R.1.5 Schematic to derive possible DTRs

In DTR, the tendon passes the finger twice (Fig. R.1.5 (a)) similar to UTR-F.
This is because it inherently provides higher actuation stiffness and causes less elongation.
From points P2 to P2N-1, a fixed pulley (Fig. R.1.5 (c)) or movable pulley at the remote actuator (Fig. R.1.5 (d)) can be used.
The tendon ends (points P0 and P2N) can be attached to the actuator or the end-effector.
Therefore 2^(n+1) number of under-actuated tendon routings are possible for n finger application.
However, we have to exclude one case where the tendon cannot transmit the force to the end-effector.
It occurs when only fixed pulleys are used at points P2 to P2n-1, and the tendon ends are fixed at the end-effector.
Accordingly 2^(n+1)-1 number of DTRs are possible for n finger application.

Page updated

Google Sites

Report abuse

1. Dual-Tendon Routing for hand wearable robots