Functional tests aim to compare the functionality of a prosthesis with a human hand. The main objective of this work is to present and evaluate an affordable prosthesis (PrHand) built with soft robotic technologies and novel joints based on compliant mechanisms. Two functional tests have been selected in this work. The first is the AHAP protocol, which evaluates how the prosthesis performs eight different grips; three variables are considered: grasping, maintaining, and grasping ability score (GAS). The results were 69.03% with 57.77% in grasping and 80.28% in maintaining. The second test is the AM-ULA, which evaluates the prosthesis by performing 23 Activities of Daily Living. PrHand prosthesis had a score of 2.5 over 4.0. The functionality of the PrHand prosthesis has similar results to other prostheses evaluated in the literature. The comparison with the human hand was 69%. PrHand presents a promising solution for amputees in developing countries regarding cost and functionality.

No abstract

The principal cause of upper limb amputations is due to traumatism. The prosthesis is an assistive device to help in the activities of daily for the amputee person. However, one of the latest reports shows that in developing countries there are around 30 million people without assistive devices. This work presents the development of two kinds of sensors for the PrHand, an upper limb prosthesis based on compliant mechanism and soft-robotics. The sensors are made with polymeric optical fiber (POF), due to their flexibility and low cost, and the working principle is based on intensity variation. The angle sensors are used for monitoring the interphalangeal joint of the fingers, and for the assessment were made cycles of closing and opening each finger. On the other hand, the force sensors are located at the tip of three fingers to track the force made over the objects. Before encoring the sensors were evaluated making five cycles of compressing and decompressing each sensor. The results show a linear behavior between the angle and the voltage variation, one most remarkable angle sensor result was with a sensibility of 0.0357 V/° and an R2 of 99 % closing and 0.0483 V/° opening. In the case of the force sensor, a polynomial relation was found between the voltage changes and the pressure over the sensor; in some cases, the relation between voltage changes and pressure could be linear but that depends on the construction of the sensor. Regarding the obtained R2 of 99 %, its sensibility was 0.0361 V/N compression and 0.0368 V/N decompression. In conclusion, was successfully developed two kinds of sensors for the instrumentation of PrHand prosthesis. It is expected to use angle and sensor variables as input in algorithms of Machine Learning to improve the detection of objects. One aspect to improve is to control in a better way the sensor construction parameters due to the considerable influence over the sensor behavior.

The development of a fiber optic sensor based on intensity variation for angle measurement of the PIP joint in the hand prosthesis PrHand based on soft-robotics is presented and discussed its viability

This paper presents the development of an intelligent soft-sensor system to add haptic perception to the underactuated hand prosthesis PrHand. Two sensors based on optical fiber were constructed, one for finger joint angles and the other for fingertips’ contact force. Three sensor fabrications were tested for the angle sensor by axially rotating the sensors in four positions. The configuration with the most similar response in the four rotations was chosen. The chosen sensors presented a polynomial response with R2 higher than 92%. The tactile force sensors tracked the force made over the objects. Almost all sensors presented a polynomial response with R2 higher than 94%. The system monitored the prosthesis activity by recognizing grasp types. Six machine learning algorithms were tested: linear regression, k-nearest neighbor, support vector machine, decision tree, k-means clustering, and hierarchical clustering. To validate the algorithms, a k-fold test was used with a k = 10, and the accuracy result for k-nearest neighbor was 98.5%, while that for decision tree was 93.3%, enabling the classification of the eight grip types.