Mechanical impedance between a soft pneumatic actuator and a human finger
A rehabilitation glove that moves a patient's fingers needs to know how much the finger opposes the motion. This work proposes measuring that through mechanical impedance, the relation between the torque applied by the actuator and the angular velocity with which the finger responds. A soft pneumatic actuator for the index finger was fabricated, a simplified model was formulated, and a test bench was built. The methodology was established; experimental validation was not achieved, and the work documents why.
Context
After a stroke, hand recovery depends on repeating movements many times. Soft exoskeleton gloves make that possible without a therapist constantly present, but their control usually relies on two kinds of signal with known limitations: kinematic variables do not reflect the patient's intention, and electromyography is hard to acquire precisely in this population.
Mechanical impedance describes the physical interaction itself. A finger that cooperates and one that resists the motion, because of pain or spasticity, should produce different impedances under the same stimulus.
The actuator
The actuator is a silicone chamber with bellows on one face, which curls when pressurized. It was cast in 3D-printed molds, and much of the work went into stabilizing the process: geometries, two- and three-part molds, PLA and resin prints, two elastomers and several curing conditions were compared. The thesis records the failures found (pores from closed molds, incomplete cure caused by release agent or temperature, wall ruptures) and the conditions that avoid them.
Two- and three-part molds for casting the actuator
The final configuration produces progressive bending between 8 and 25 psi.
Air-line control setup: compressor (1), regulation (2) and valves (3)
Model and test bench
The model treats the finger as three links (proximal, middle and distal phalanges) and computes the torque at each joint from the contact force on each phalanx. Impedance is obtained as the ratio between that torque and the corresponding angular velocity.
Torque generated by the contact force on the proximal phalanx
Forces were measured with force-sensitive resistors (FSR) placed between the actuator and each phalanx, and angles by tracking markers on video. Tests were run at 10, 15 and 20 psi and at several motion frequencies.
Simplified experiment setup
Results
The impedance values obtained were inconsistent, and even negative in some cases, so reliable ranges for a control strategy could not be established.
Impedance ranges per frequency for three inflation pressures
The work identifies three causes:
- The FSRs showed hysteresis and drift under cyclic loading, confirmed in later calibrations.
- The initial visual tracking was imprecise. After redesigning it with markers of different colors and a region of interest per marker, the error in the amplitude of the three joint angles against a reference trajectory from the literature dropped from 48.9%, 46.6% and 149.1% to 8.9%, 25% and 5.7%.
- Manual execution of the movement introduced variability between repetitions.
Joint angles obtained with the refined tracking system
What is missing
The next step is instrumentation: replace the FSRs with miniature load cells, calibrate the vision system in real angular units, and repeat the experiment. Only with reliable measurements does it make sense to assess whether impedance distinguishes a cooperating finger from a resisting one, and to test that with people in rehabilitation.
How it fits in Robiolab
The project brings together two of the lab's lines, soft actuators and wearable rehabilitation devices, around a question of physical human-robot interaction. The negative result is informative: in soft devices contact is distributed and deformable, and measuring forces at that interface is an open problem that also shows up in the elbow orthosis and the soft tourniquet.