Paper contribution 3

• Simulation results show that the extensor control enables the pinch motion by improving the force capability.

•  Inspired by the simulation result, we have developed a tendon-driven hand-wearable robot, Exo-Glove Pinch, the next version of Exo-Glove, using two active tendons called flexor and extensor.

• Experiments done by an individual with spinal cord injury (SCI) show that extensors can have two distinct functions: 

1) extending the fingers

2) stabilizing the grasp (Interestingly, the human also stabilizes the grasp using extensor)

3.2. Grasp performance of the Exo-Glove Pinch done by a SCI person

• Fingertip force is measured to see the force capbility of the Exo-Glove Pinch.

• Fig. S3.3 shows the experimental setup to measure the fingertip force. 

• Fig. S3.4 shows the experimental result from a SCI person.

Video. S.3.1 Videos showing how the finger moves when doing the experiment (with the experimental setup shown in Fig. S3.3) when the extensor is not controlled (left) and when the extensor is controlled (right). The experiments are done by a SCI person.

3.3. Qualitative assessment of the Exo-Glove Pinch (done by a SCI person)

• We believe qualitative assessment of the wearable robot (how the user actually feels) is also important. Below is interview video done by a SCI person.

3.4. Benefits of assisting power and pinch grasp with a single robot

• Exo-Glove Pinch assists two grasps (power/pinch grasps) that fall into two main categories according to the grasp taxonomy [R3.3.1]. 

• Assisting one grasp in each category provides benefits in enabling users to grasp various objects firmly.  Our hypothesis could be: 

  • Power grasp is useful when grasping objects with complex shape (e.g., objects used for adaptability tests) with adaptable motion as it makes many contact points. 

  • Pinch grasp is useful when grasping thin objects with cube shape (e.g., objects used for force capability tests) with more precise force 

• To validate above hypothesis, we tested two grasps against 19 objects (used in adaptability and force capability tests) using two performance metrics which are:

1) Resistance to External Impact Perturbation 

• This test evaluates the grasp's ability to restrain an object when subjected to sudden impact forces. The simulation conditions are as follows: 

1) Compute the optimal control input that maximizes contact force during grasping. 

2) Apply impact force to the object for 0.5 seconds.

3) Measure the object's displacement by comparing its position before and after the force is applied. 

4) Quantify grasp performance using the ratio of impulse (force x impact time) to displacement. For instance, if the displacement is 1 mm, the performance is 50 Ns/mm. 

5) Do the above process for five force vectors [100N,0,0], [-100N,0,0], [0,100N,0], [0, -100N,0], and [0,0,100N]. The force [-100N,0,0] is excluded as the ground at z=0 obstructs meaningful grasp assessment. 

2) Ability to lift heavy objects

• • • The maximum weight test assesses the system’s ability to sustain high forces in a static manner. The simulation conditions are as follows: 

1) Compute the optimal control input that maximizes contact force during grasping. 

2) Apply a gradually increasing virtual force in +z direction to the object.

3) When the object’s displacement in the z-direction exceeded half its height, measure the corresponding virtual force. For example, for a 0.5m object, measure the force when displacement > 0.25m.

• Results are as below:

[R3.3.1] Feix, Thomas, et al. "The grasp taxonomy of human grasp types." IEEE Transactions on human-machine systems 46.1 (2015): 66-77. 

• Interestingly, when grasping objects 1-6 (used in adaptability test), power grasp outperformed compared to pinch grasp.

• Conversely, when grasping thin-cube shaped objects (used in force capability test), pinch grasp show better results than power grasp. 

• These findings suggest that assisting both grasps from different categories can be an effective strategy, enhancing versatility in handling various object shapes and tasks by leveraging the strengths of each grasp to optimize both adaptability and force capability.