Embedded Files
2.2 Simulation of Constrained Tendon Routing
Paper contribution 2
Paper contribution 2
The paper analyzes and simulates the constrained tendon routing methods.
* Mathematical analysis is done in the main paper.
** The simulation code is located here.
2.2.1 Force Capability of Force-constrained Tendon Routings (FTR)
2.2.1 Force Capability of Force-constrained Tendon Routings (FTR)
The robot with FTR sometimes cannot apply contact force that is located inside the friction cone (Fig. S.2.2.1a)
This is because the robot with FTR has a limited achievable torque manifold (Fig. S.2.2.1b)
Controlling extensors can solve this issue by extending the achievable joint torque manifold (Fig. S.2.2.1c)
We found that the extensor control stabilizes the grasp by forming the user's hand as a pinch grasp posture - Interestingly, the human also uses an extensor to stabilize the grasp [Ref 2.1].
Fig. S.2.2.0 Possible issue in the robot that uses force-constrained tendon routing
(a) shows a possible issue that happens in the robot that uses FCR: when the robot has fewer tendons than the joints, the contact force can be located outside the friction cone. In this case, the robot fails to grasp. (b) shows a simplified achievable torque manifold when the robot has two joints actuated by a single tendon. In this case, the robot can only apply joint torque that is in a straight line. (c) shows how the extra tendon expands the achievable torque manifold. When using two tendons, the torque manifold expands to a 2D plane.
2.2.2 Pinch grasp assistance with extensor control
2.2.2 Pinch grasp assistance with extensor control
When the joint has a limited range of motion, the robot can make both a power grasp (Fig. S2.2.2.a) and a pinch grasp (Fig. S2.2.2.b) by controlling the extensor. Video S2.2.1 and Video S2.2.2 show how the robot makes the postures.
Simulations are conducted to see how finger joints (MCP, PIP, and DIP) move and represented in Fig. S2.2.1. As can be seen in the graphs, the co-contraction makes the finger to only flexes MCP joint while keeping other joints extended. This posture is similar with pulp pinch [Ref 2.2.2.1, Ref 2.2.2.2], and therefore we named our robot as Exo-Glove Pinch.
The pulp pinch motion appears more stable than the power grasp when handling thin objects. This motion aligns the index and middle fingers relatively parallel to the thumb, allowing the user's hand to function like a parallel gripper.
[Ref 2.2.2.1] Ng, Poh Kiat, et al. "A review of different pinch techniques." Theoretical Issues in Ergonomics Science 15.5 (2014): 517-533.
[Ref 2.2.2.2] Hara, A., Yamauchi, Y., Kusunose, K. (1994). Analysis of Thumb and Index Finger Joints During Pinching Motion and Writing a Cross, as Measured by Electrogoniometers. In: Hirasawa, Y., Sledge, C.B., Woo, S.LY. (eds) Clinical Biomechanics and Related Research. Springer, Tokyo.
Fig. S2.2.1. Joint angles simulated using both with/without extensor control. When extensor is not used when making the motion, the robot flexes three joints (MCP, PIP, and DIP joints) simultaneously (left). However, when extensor is used, the robot only flexes the MCP joint (right).
Fig. S2.2.1. Joint angles simulated using both with/without extensor control. When extensor is not used when making the motion, the robot flexes three joints (MCP, PIP, and DIP joints) simultaneously (left). However, when extensor is used, the robot only flexes the MCP joint (right).
Video. S2.2.1
Posture without control of the extensors
Video. S2.2.1
Posture without control of the extensors
Video. S2.2.2
Posture with control of the extensors
Video. S2.2.2
Posture with control of the extensors
2.2.3 Simulation to validate how pinch grasp increases the contact force
2.2.3 Simulation to validate how pinch grasp increases the contact force
We simulated how the extensor control increases the maximum contact force while grasping wide (and narrow) objects.
13 objects (thickness 10mm to 70mm with 5mm increases) are used for the simulation. The object shape is fixed as a box shape since we were more interested in how the pinch grasp helps grasp the thin objects.
The contact force (when using extensor control) is 4.66 times the contact force (without extensor control) as shown below.
The contact force doesn't fall to zero in some cases. It was because the fingers penetrated the objects. Therefore, we used maximum contact force as an indicator of the force capability.
Fig. S2.2.3 Simulation results that show how controls with/without extensor affect the maximum contact force.
2.2.4. Example video showing how the extensor control improves the force capability (Object 4)
2.2.4. Example video showing how the extensor control improves the force capability (Object 4)
Video. S2.2.3 Grasping motion when the robot pulls the flexor tendon while controlling the extensor.
Video. S2.2.4 Grasping motion when the robot only pulls the flexor tendon.
2.2.5. Contact force according to the thickness of the objects
2.2.5. Contact force according to the thickness of the objects
Fig. S2.2.5. Simulation result - Contact force measured in time domain with/without extensor control
Fig. S2.2.5. Simulation result - Contact force measured in time domain with/without extensor control
Table. S2.2.1. Simulation result - Maximum contact force with/without extensor control
Table. S2.2.1. Simulation result - Maximum contact force with/without extensor control
* The slope of the contact force changes in the graph. The reason is because we modulated the slope of tension according to the total contact force to stabilize the simulation. The tension increase change when the total contact force becomes 150N.
Page updated
Google Sites
Report abuse