applied sciences

Article Concept, Design, Initial Tests and Prototype of Customized Upper Limb Prosthesis

1 2 3 2 Corina Radu (Frent, ) , Maria Magdalena Ros, u , Lucian Matei , Liviu Marian Ungureanu and Mihaiela Iliescu 1,*

1 Institute of Solid Mechanics, Romanian Academy, Constantin Mille 15, 010141 Bucharest, Romania; [email protected] 2 Faculty of Industrial Engineering and Robotics, University POLITEHNICA of Bucharest, Splaiul Independentei 313, 060042 Bucharest, Romania; [email protected] (M.M.R.); [email protected] (L.M.U.) 3 Faculty of Mechanics, University of Craiova, Alexandru Ioan Cuza 13, 200585 Craiova, Romania; [email protected] * Correspondence: [email protected]

Abstract: This paper presents aspects of the concept and design of prostheses for the upper limb. The objective of this research is that of prototyping a customized prosthesis, with EMG signals that initiate the motion. The prosthesis’ fingers’ motions (as well as that of its hand and forearm parts) are driven by micro-motors, and assisted by the individualized command and control system. The software and hardware tandem concept of this mechatronic system enables complex motion (in the horizontal and vertical plane) with accurate trajectory and different set rules (gripping



 pressure, object temperature, acceleration towards the object). One important idea is regarding customization via reverse engineering techniques. Due to this, the dimensions and appearance Citation: Radu (Frent,), C.; Ros, u, (geometric characteristics) of the prosthesis would look like the human hand itself. The trajectories M.M.; Matei, L.; Ungureanu, L.M.; and motions of the fingers, thumbs, and joints have been studied by kinematic analysis with the Iliescu, M. Concept, Design, Initial matrix–vector method aided by Matlab. The concept and design of the mechanical parts allow Tests and Prototype of Customized for complex finger motion—rotational motion in two planes. The command and control system is Upper Limb Prosthesis. Appl. Sci. 2021, 11, 3077. https://doi.org/ embedded, and data received from the sensors are processed by a micro-controller for managing 10.3390/app11073077 micro-motor control. Preliminary testing of the sensors and micro-motors on a small platform, Arduino, was performed. Prototyping of the mechanical components has been a challenge because Academic Editors: Narayan Roger of the high accuracy needed for the geometric precision of the parts. Several techniques of rapid and Governi Lapo prototyping were considered, but only DLP (digital light processing) proved to be the right one.

Received: 6 February 2021 Keywords: prosthesis; reverse engineering; 3D prototyping; embedded command and control Accepted: 27 March 2021 system; sensors Published: 30 March 2021

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in 1. Introduction published maps and institutional affil- iations. The upper limb is an essential organ for any person due to its ability to complete a lot of actions with a higher level of dexterity. This is how, by using their upper limbs, people can achieve tasks such as grasping, holding, picking, and lifting more efficiently than other living creatures can. According to the World Health Organization report on disability [1], about “15% of the global population—over a billion people—lives with some Copyright: © 2021 by the authors. form of disability, of whom 2–4% experience significant difficulties in functioning”. Licensee MDPI, Basel, Switzerland. According to Kim et al. [2], cited by [3], “the ratio of upper-extremity to lower- This article is an open access article extremity amputations is 1:2.2” and the “persons with upper-extremity amputations suffer distributed under the terms and from frustration and difficulty during the rehabilitation process. The main factors causing conditions of the Creative Commons Attribution (CC BY) license (https:// loss of the upper limbs are accidents followed by general diseases and injuries, and also in creativecommons.org/licenses/by/ some cases for diseases and tumours, the amputation is a way of stopping the spread of 4.0/). the disease to the rest of the body [3]”.

Appl. Sci. 2021, 11, 3077. https://doi.org/10.3390/app11073077 https://www.mdpi.com/journal/applsci Appl. Sci. 2021, 11, 3077 2 of 29

Many of these people “suffer from frustration and difficulty during the rehabilitation process and experience paresthesia due to the loss of delicate movement by the hands, com- plicated tactile senses of the upper extremities, and proprioceptive sensory functions [3]” and all of these people require assistive technologies such as prostheses, wheelchairs, etc. Considering the crucial functions of hands, the life of people who are missing hands is complicated, and, for this reason, it is very difficult to adapt to situations in which something that used to be done naturally is now done within limits imposed by this disability. Upper limb prostheses (ULP) are generally classified into two categories based on their functionality: passive prostheses and active prostheses [4], cited by [5]. A good solution for artificial hands is myoelectric prostheses, which are currently being developed. With these types of prostheses, the myoelectric signals of muscle contractions are ultimately transformed into functions desired by the users through the use of surface electrodes [3]. It is a ‘myoelectric control system’ for controlling the myoelectric prosthesis [6], cited by [3]. However, the design, the control structure, and the manufacturing process are difficult, and moreover, the cost is a very important element to consider in all of these phases [3,5]. An adaptable hand with flexibility, dexterity, and load carrying capacity analogous to the human hand seems to be the ideal objective for prosthetic application [7]. Prosthetics research has contributed to the development of very complicated hand prostheses but their structures, with programmable control systems, are not always user- friendly. This makes them very difficult in operation, and the people at whom they are aimed do not use them in the usual way [7–10]. Thus, it is necessary to develop intelligent prostheses that can be easily adapted to the requirements of each individual they are intended for. For the development of such products, researchers must consider the following basic systems: mechanical, actuators and motors, sensors, and command and control [10]. According to [11], for designing a hand prosthesis, the project team needs to un- derstand the mechanics of mechanisms such as gears, levers, and points of mechanical advantage, and electromechanical design such as switches, DC motors, and electronics. Mechatronics is the new word used to describe this “marriage” of mechanical and electronic engineering. However, for applying the concepts of mechatronics in bioengineering, it is necessary to take into account the principles of how the human body functions. Electromyography (EMG) [12] is used to evaluate and record the electrical activity produced by the muscles of human body. Recently, it has been used in the rehabilitation of patients with amputations in the form of robotic prostheses, it is a tool used to evaluate and classify the different movements of body, so that the robotic mechanism can effectively imitate the motions of human limbs. In [13], mathematical algorithms tested in prototypes of intracortical neuromotor prostheses have been developed. These closed-loop algo- rithms [13] offer visual access (neurally derived cursor trajectories), and mechanical access (as in the stimulation of muscles via implanted electrodes) to any number of output devices. The upper limb prostheses should be customized, complex, high-performance, and able to ensure a good life for the person wearing it. There are many hand prostheses available on this special “market”, but efficient prostheses are expensive and are not affordable for all of the people who might need them. This paper presents aspects of the concept and design of a customized prosthesis for the upper limb. The software and hardware tandem concept of this mechatronic system enables complex motion (in the horizontal and vertical planes) with accurate trajectory and different set rules (gripping pressure, object temperature, acceleration toward the object). The command and control system is embedded, and the data received from the sensors are processed by a micro-controller for managing the control of the micro-motors. The customized prosthesis is intended to be light and user-friendly, is estimated to have medium manufacturing costs (a few thousand Euros), and would be reliable, with little maintenance required. Appl. Sci. 2021, 11, x FOR PEER REVIEW 3 of 33 Appl. Sci. 2021, 11, 3077 3 of 29

2. Materials and Methods 2. MaterialsOne of the andfirst Methodssteps in the conceptualization of the customized upper limb prosthe- sis is thatOne of ofthe the kinematic first steps analysis in the conceptualization of its finger mechanism. of the customized In order to upperperform limb this, prosthesis some basicis that aspects of the of kinematicthe mechanism analysis theory of its ha fingerve been mechanism. considered Inas orderfollows to [14–22]: perform this, some • basicLink—is aspects a rigid of the body. mechanism theory have been considered as follows [14–22]: • • Joint—isLink—is a permanent a rigid body. contact between two links. • • PlanarJoint—is motion—is a permanent the motion contact “in between which all two points links. belonging to a link move in a plane • knownPlanar as motion—is the plane of the motion, motion while “in which simultan all pointseously belonging the link tois afree link to move rotate in about a plane anknown axis perpendicular as the plane ofto motion,the plane while of motion”. simultaneously the link is free to rotate about an • Degreeaxis perpendicular of freedom of toa rigid the plane body—is of motion”. the number of independent movements it has. • Degree of freedom of a rigid body—is the number of independent movements it has. Kinematic constraints are “constraints between rigid bodies that result in the de- crease ofKinematic the degrees constraints of freedo arem “constraintsof the rigid body between system”. rigid bodies that result in the decrease ofAn the Assur degrees group of freedom is a set of of the kinematic rigid body chains system”. developed by Leonid Assur whose de- gree of Anfreedom Assur is groupzero. A is complex a set of structure kinematic can chains be constructed developed by by extending Leonid Assur the Assur whose kinematicdegree of chain. freedom The is complex zero. A complexlinkage mechan structureism can can be constructedbe regardedby as extendingoriginating the from Assur somekinematic Assur kinematic chain. The chains complex linkage mechanism can be regarded as originating from someKinematic Assur kinematic analysis is chains the motion analysis aimed at determining the positions, speed, and accelerationKinematic distributions analysis is the of motion each element, analysis aimedpreviously at determining knowing their the positions, constructive speed, characteristicsand acceleration and the distributions relative movement of each element,of the active previously link elements. knowing their constructive characteristicsA dyad is an and Assur the group relative made movement of two parts, of the three active joints, link elements. and two outputs (FigureA 1a,b). dyad is an Assur group made of two parts, three joints, and two outputs (Figure1 a,b). A triadA triad is an is anAssur Assur group group made made of four of four parts, parts, six sixjoints, joints, and and three three outputs outputs (Figure (Figure 1c).1c).

(a)

(b)

Figure 1. Cont. Appl. Sci. 2021, 11, 3077 4 of 29 Appl. Sci. 2021, 11, x FOR PEER REVIEW 4 of 33

Appl. Sci. 2021, 11, x FOR PEER REVIEW 4 of 33

(c) (c) FigureFigure 1. Assur 1. Assur group group types—dyad types—dyad and triad: and triad:(a) kinematic (a) kinematic scheme schemefor the dyad for the (RRR) dyad [21], (RRR) (b) [21], kinetostaticFigure 1. Assur scheme group for thetypes—dyad dyad (RRR) and [21], triad: and (a) (kinematicc) kinematic scheme and kinetostaticfor the dyad schemes(RRR) [21], for ( bthe) kinetostatic(b) kinetostatic scheme scheme for the for dyad the dyad(RRR) (RRR)[21], and [21 (],c) and kinematic (c) kinematic and kinetostatic and kinetostatic schemes schemesfor the for the triad (RR-RR-RR) [22] where: R is rotation, 𝐹 is the force vector (components along axes), and 𝑅 triadtriad (RR-RR-RR) (RR-RR-RR) [22] [22 where:] where: R isR rotation,is rotation, 𝐹 isF theis the force force vector vector (components (components along along axes), axes), and 𝑅 and R is is the joint reaction force vector (components along axes). isthe the joint joint reaction reaction force force vector vector (components(components along axes). axes).

ForForFor the the the mechanism mechanism mechanism of of of thethe the designeddesigned designed prosthes prosthesis,is,is, thethethe kinematic kinematic kinematic analysis analysis analysis was was wasdone done done by by by thethethe matrix–vector matrix–vector matrix–vector method method method [23,24], [23,24], [23,24 whichwhich], which is simp is simple,le,le, intuitive,intuitive, intuitive, and and fits fits and for for open fits open for spatial spatial open con- spatialcon- tours.tours.contours. For For its Forits index index its index finger finger finger mechanism,mechanism, mechanism, the kinematickinematic the kinematic schemescheme scheme is is shown shown is shown in in Figure Figure in Figure 2. 2.The2 The. The final Hartenberg–Denavit trihedral, T1, T2, ..., Ti+1, is associated with the fingertips. The finalfinal Hartenberg–Denavit Hartenberg–Denavit trihedral, trihedral, T1T, T1,2,T ...,2, ...,Ti+1T, iis+1 ,associated is associated with with the thefingertips. fingertips. The The trihedral versos are 𝑛 , 𝑒̅ ×𝑛 . trihedraltrihedral versos versos are are 𝑛n,i ,𝑒e̅ i+1 ×𝑛×ni..

trihedraltrihedral scheme scheme e1 e1 e n4 51e n4 51 T H 1 e61 H e T1 e61 e 61 θ n θ 61 θ43 3n 54θ 43 3 54 e51 H e e51 H 1 e21 n5 e1 n n n4 e21 5 5 n4 n n5 n 4 G 4 G n3 e e41 G 41 n3 θ e α e41 G e 31 41 1 1 θ e O′ α 31 21 11 1 e1 ′ n3 e O11 β e31 F 21 n3 n β ′ e31 n1 1 F O1 e n n1 31 O1 O′ 1 1 e1 e n2 O E, F 31 1 n2 e E, F e21 1 e21 n2 n1 e1 n2 E e21 e21 θ n1 e1O e E21 × 31 θ e1 n0 θ n2 10 O e 21 O × 31 θ n0 e1 n0 n2 e 10 1O e21 n 0 e1 e n0 21

(a) n0 (b)

FigureFigure 2. 2. SchemesSchemes(a) of the of designed the designed index indexfinger fingermechanism: mechanism: (a) kinematic(b) (a) kinematic scheme, (b scheme,) versorial ( b) verso- scheme. Figurerial scheme.2. Schemes of the designed index finger mechanism: (a) kinematic scheme, (b) versorial scheme. Appl. Sci. 2021, 11, 3077 5 of 29

The main steps of the calculus required for the kinematic analysis of the index finger mechanism are presented next. The simplified structure of the mechanism is a serial open kinematic chain made of dyads. The base trihedral, n0, e1 × n0, e1, overlaps the orthogonal reference trihedral Oxyz with the i, j, k versors. The position vector of the T1 customized

point, rT1 , is given by relation (1): = + + + + + + + + + + + rT1 s1e1 a1n1 s2e2 a2n2 s3e3 a3n3 s4e4 a4n4 s5e5 a5n5 s6e6 a6n6 (1)

where:

- s1 represents the distance between normal versors, ni−1 and ni (axial displacement), with a positive sign along the ei axis; - ai is the distance between the ei and ei+1 axes, positive sign; - θi, i−1 is the angle between the ni−1ni−1 and ni versors, positive sign along ei axis; - αi is the angle between the ei and ei+1 versors (crossing angle), positive sign along ni axis. In order to simplify the transformation matrix elements’ notation, the symbols are: θ1 = θ10, θ2 = θ20, and θ3 = θ30. Coordinates transformation of a point, from Ti trihedral to Ti+1 trihedral, is:

Ti+1 = Ai·Ti (2)

where:   cos θi sin θi 0 Ai =  − sin θi· cos αi cos θi· cos αi sin αi  (3) sin θi· sin αi − cos αi· sin αi cos αi is the transformation matrix. Consequently, there are evidenced the relations that follows:

Ti+1 = Ai·Ti = Ai(Ai−1·Ti−1) = Ai Ai−1(Ai−2·Ti−2) = Ai Ai−1 ... A2 A1T1 (4)

and ni = Din0 + Eie1 × n0 + Fie1 (5)

ei+1 = Ai+1n0 + Bi+1e1 × n0 + Ci+1e1 (6)

where Ai+1, Bi+1, C1+1, Di, Ei, Fi, Ki, Li, Mi represent the component elements of the transformation matrix,   Di Ei Fi Ai = Ai Ai−1·····A2 A1 =  Ki Li Mi  (7) Ai+1 Bi+1 Ci+1

Based on all the above, the projections of the position vector, rT1, on the Oxyz trihe- dral are:

XT = a1·D1 + a2·D2 + ··· + a5·D5 + s2·A2 + s3·A3 + ··· + s6·A6 YT = a1·E1 + a2·E2 + ··· + a5·E5 + s2·B2 + s3·B3 + ··· + s6·B6 (8) ZT = a2·F2 + a3·F3 + ··· + a5·F5 + s1 + s2·C2 + s3·C3 + ··· + s6·C6

or, written as a matrix, it turns into:

 T XT YT ZT = P·AS (9)

where:   D1 D2 D3 D4 D5 0 A2 A3 A4 A5 A6 P =  E1 E2 E3 E4 E5 0 B2 B3 B4 B5 B6  (10) 0 F2 F3 F4 F5 1 C2 C3 C4 C5 C6 Appl. Sci. 2021, 11, x FOR PEER REVIEW 6 of 33

Appl. Sci. 2021, 11, 3077 6 of 29

𝐴𝑆 = 𝑎 𝑎 𝑎 𝑎 𝑎 𝑠 𝑠 𝑠 𝑠 𝑠 𝑠 (11) For the designed index finger mechanism (see Figure 2), there are some customized valuesand that equal zero, as follows: T   AS = a1 a2 a3 a4 a5 s1 s2 s3 s4 s5 s6 (11) - the axial displacements: 𝑠 =0; 𝑠 =0; 𝑠 =0 - theFor distances the designed between index axes finger versors: mechanism 𝑎 =0; (see𝑎 =0 Figure; 𝑎2=0), there. are some customized valuesResults that equal of the zero, kinematic as follows: analysis are further presented in Section 3. The mechatronic

system- the for axial the displacements:upper limb has sbeen3 = 0;designeds4 = 0; sso5 =that0 it enables finger and forearm motion like- thatthe distancesof a real limb between as much axes as versors: possible.a1 =Each0; a finger2 = 0; isa5 driven= 0. by three micromotors, and its mechanism allows for complex motion, in both the horizontal and vertical planes. Results of the kinematic analysis are further presented in Section3. The mechatronic The hand and forearm are to be driven by two motors each (and, if required, by assigned system for the upper limb has been designed so that it enables finger and forearm motion gearboxes), and their motion will be possible in two planes (horizontal and vertical). like that of a real limb as much as possible. Each finger is driven by three micromotors, From the point of view of the system’s appearance and its components’ dimensions, and its mechanism allows for complex motion, in both the horizontal and vertical planes. it has been envisaged to look similar to a real hand. The reverse engineering technique The hand and forearm are to be driven by two motors each (and, if required, by assigned has been used to scan a real hand and to obtain its surface dimensions. The technique used gearboxes), and their motion will be possible in two planes (horizontal and vertical). was laser scanning, the specific equipment was a MetraSCAN 3D scanner, and the scan- From the point of view of the system’s appearance and its components’ dimensions, it ning process did not last more than 3 min. Still, if any slight movement of the real hand has been envisaged to look similar to a real hand. The reverse engineering technique has occurs while scanning, the acquired data (point cloud) generate a contour surface with been used to scan a real hand and to obtain its surface dimensions. The technique used was errors. This is why the process had to be repeated few times till a correct and complete laser scanning, the specific equipment was a MetraSCAN 3D scanner, and the scanning surface had been generated. process did not last more than 3 min. Still, if any slight movement of the real hand occurs The scanning process was performed using a 0.1 mm mesh resolution at a measure- while scanning, the acquired data (point cloud) generate a contour surface with errors. ment rate of 800,000 measurements/s. The finished scanned body was achieved by align- This is why the process had to be repeated few times till a correct and complete surface ing 3 scans, with each scan containing overlapping areas. The resulting body can be seen had been generated. in Figure 3b,c. The scanning process was performed using a 0.1 mm mesh resolution at a measure- After each scan, the concerned body and also the table body on which the hand is ment rate of 800,000 measurements/s. The finished scanned body was achieved by aligning resting (see Figure 3a) were cleaned by different surplus meshes by using the proprietary 3 scans, with each scan containing overlapping areas. The resulting body can be seen in software (VXmodel) and different types of commands and functions within the software. Figure3b,c. The images taken while scanning are presented in Figure 3.

(b)

(a) (c)

FigureFigure 3.3. ReverseReverse engineeringengineering technique for for the the real real upper upper limb: limb: (a (a) )scanning scanning process, process, (b (b) )3D 3D scan scan surface—with surface—with errors, errors, ((cc)) 3D 3D scanscan surface—accurate.surface—accurate.

After each scan, the concerned body and also the table body on which the hand is resting (see Figure3a) were cleaned by different surplus meshes by using the proprietary Appl. Sci. 2021, 11, 3077 7 of 29

Appl. Sci. 2021, 11, x FOR PEER REVIEW 7 of 33 software (VXmodel) and different types of commands and functions within the software. The images taken while scanning are presented in Figure3. Once the surface of the real upper limb was obtained (by combining and aligning 3 differentOnce scans the surface of the same of the hand), real upper further limb processing was obtained to determine (by combining the hand and aligning and fingers’ 3 dimensionsdifferent scans was of performed. the same hand), Information further about processing the dimensions to determine and the contour hand and characteristics fingers’ (shapedimensions and size) was ofperformed. the hand Information were obtained about by the using dimensions reverse and engineering contour characteristics software (Geo- magic(shape DesignX and size) and of Solidworks)the hand were and obtained plane generation, by using reverse as can engineering be seen in Figure software4. For (Ge- this process,omagic weDesignX used 43and planes Solidworks) from the and tips plane of the generation, fingers to theas can back be of seen the in hand.. Figure The 4. planesFor werethis generatedprocess, we using used the 43 offsetplanes function, from the andtips theof the distance fingers between to the back two of consecutive the hand.. planesThe wasplanes set towere 10 mm.generated using the offset function, and the distance between two consecu- tive planes was set to 10 mm.

FigureFigure 4. 4.Reverse Reverse engineeredengineered realreal upper limb surface— surface—transversaltransversal plane plane contour contour characteristics characteristics

IfIf the the interest interest onon dimensionsdimensions referred to to th thee horizontal horizontal plane, plane, the the scanning scanning process, process, usingusing the the proprietary proprietary softwaresoftware (VXmodel), gene generatedrated aa triangle triangle mesh mesh in inreal real time time through through thethe alignment alignment or or registration registration method.method. This This proc processess can can be be seen seen in in Figure Figure 5a.5a. After After this this is is done,done, the the dimensions dimensions and and dimension dimension parameters parameters were were obtained obtained by by using using reverse reverse engineer- engi- ingneering software software (Geomagic (Geomagic DesignX) DesignX) and specificand specif tools,ic tools, such such as in as Figure in Figure5b. Then, 5.b. Then, a sketch a wassketch generated was generated and parametrized, and parametrized, as in Figure as in Figure5c. 5c. The command and control system for the upper limb prosthesis is embedded and designed for customized applied mechatronics application. Available applications of this prothesis are in the social–medical fields for people with upper limb amputations (hand, forearm or, arm) who would be closer to “normal” everyday life. Another application would be in the field of CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosives) security, for taking samples and/or the manipu- lation of harmful products in dangerous environments (radiations, chemicals). Still, this presented research is focused on applications for improving the quality of life of people with upper limb disabilities via prosthesis. This designed system uses data acquisition and control, the data being received from sensors (pressure, tilt, temperature, etc.) and further processed by a micro-controller to manage the micro-motors’ control. Depending on the missing part of the human body, the Appl. Sci. 2021, 11, 3077 8 of 29

Appl. Sci. 2021, 11, x FOR PEER REVIEWsystem’s design and structure varies from the point of view of the sensors’ type, number,8 of 33 and position; micro-motors; micro-controllers; electric circuit pattern, etc.

(a)

(b) (c)

FigureFigure 5. Reverse5. Reverse engineered engineered real real upper upper limb limb surface—horizontal surface—horizontal planeplane contourcontour characteristics: ( (aa)) triangle triangle mesh, mesh, (b (b) )di- di- mensions obtained, (c) parametrization. mensions obtained, (c) parametrization.

TheThe algorithm command scheme and control of the system upper limbfor the prosthesis upper limb is shown prosthesis in Figure is embedded6. The novelty and ofdesigned this concept for customized is that, for eachapplied of the mechatronics five fingers, application. there are three micro-motors to generate motionsAvailable (not servomotors, applications as of with this mostprothesis of the are available in the social–medical myoelectric prostheses), fields for people so that thewith combination upper limb of amputations software and (hand, hardware forearm (including or, arm) thewho mechatronic would be closer system) to “normal” is specific toeveryday this prosthesis. life. Another The fingers’ application control would is accurate be in the and field complex of CBRNE due (Chemical, to the complexity Biological, of motionsRadiological, (in the Nuclear, horizontal and and Explosives) vertical planes). security, for taking samples and/or the manipula- tionBasically, of harmful the products main components in dangerous of the environments system are: (radiations, chemicals). Still, this presented research is focused on applications for improving the quality of life of people - sensors—for peripheral control (pressure sensors for the gripping force of an object, with upper limb disabilities via prosthesis. temperatureThis designed sensors system for uses estimating data acquisition the temperature and control, of thethe objectdata being to be received gripped, from and sensorstilt sensors (pressure, for thetilt, inclinationtemperature, angle etc.) of an thed further fingers’ processed phalanges); by a micro-controller to - managedrivers the for micro-motors’ the micro-motors; control. Depending on the missing part of the human body, - the micro-motorssystem’s design (for and the structure fingers’ varies motion) from and the motors point (forof view the handof the andsensors’ forearm type, motion) num- ber,with and anposition; assigned micro-motors; demultiplication micro-controllers; gearbox, if required;electric circuit pattern, etc. - microcontroller—aThe algorithm scheme microcomputer of the upper limb with prosthesis digital and is shown analogic in Figure inputs 6. andThe novelty a GPIO of this(General-Purpose concept is that, for Input/Output) each of the five port. fingers, there are three micro-motors to generate motions (not servomotors, as with most of the available myoelectric prostheses), so that the combination of software and hardware (including the mechatronic system) is specific Appl. Sci. 2021, 11, x FOR PEER REVIEW 9 of 33

Appl. Sci. 2021, 11, 3077 9 of 29 to this prosthesis. The fingers’ control is accurate and complex due to the complexity of motions (in the horizontal and vertical planes).

FigureFigure 6. 6.The The algorithm algorithm schemescheme of the upper upper limb. limb.

TheBasically, whole the system main iscomponents conceived of as the an system open hardware are: platform with high versatility. A sound or vibration warning system could be added if visually impaired people would need the prosthesis. The main aspects of how the customized prosthesis is designed to work are presented in Figure7. The first step is that of checking the battery status. If it is necessary, then the battery should be plugged in via the USB port. This port also enables writing to the Appl. Sci. 2021, 11, x FOR PEER REVIEW 10 of 33

- sensors—for peripheral control (pressure sensors for the gripping force of an object, temperature sensors for estimating the temperature of the object to be gripped, and tilt sensors for the inclination angle of the fingers’ phalanges); - drivers for the micro-motors; - micro-motors (for the fingers’ motion) and motors (for the hand and forearm motion) with an assigned demultiplication gearbox, if required; - microcontroller—a microcomputer with digital and analogic inputs and a GPIO (General-Purpose Input/Output) port. The whole system is conceived as an open hardware platform with high versatility. A sound or vibration warning system could be added if visually impaired people would Appl. Sci. 2021, 11, 3077 need the prosthesis. 10 of 29 The main aspects of how the customized prosthesis is designed to work are presented in Figure 7. The first step is that of checking the battery status. If it is necessary, then the flashbattery memory, should operational be plugged maintenance,in via the USB adjustment/repairsport. This port also enables the system, writing and to updating the flash and memory, operational maintenance, adjustment/repairs the system, and updating and im- improving the prosthesis’ system. proving the prosthesis’ system.

FigureFigure 7.7. TheThe block scheme for for the the designed designed prosthesis. prosthesis.

TheThe temperaturetemperature sensorsensor is an infrared active digital sensor sensor with with an an I2C I2C communica- communication protocol,tion protocol, and isand aimed is aimed at determining at determining the temperature the temperature of the of objectthe object going going to be to gripped. be Thegripped. gyroscope–accelerometer The gyroscope–accelerometer sensors aresensor aimeds are at aimed determining at determining the angular the angular position po- of the prosthesissition of the and prosthesis its motion and acceleration. its motion accele Theration. pressure The andpressure tilt sensors and tilt aresensors resistive, are resis- and for highertive, and conversion for higher accuracyconversion they accuracy are connected they are connected to an ADC to (Analog-to-Digitalan ADC (Analog-to-Digital Converter) byConverter) an instrumentation by an instrumentation amplifier. amplifier. If there is If not there enough is not digitalenough inputs, digital theninputs, different then dif- ADC componentsferent ADC components will be used will (in be the used case (in of the multi-channel case of multi-channel EMG). EMG). TheThe EMG sensor sensor (MyoWare) (MyoWare) is aimed is aimed at initia at initiatingting the motion. the motion. It could Itbe couldmulti-chan- be multi- channelnel EMG EMG for higher for higher sensitivity. sensitivity. Micro-motorMicro-motor drivers get get commands commands from from the the controller controller via via a GPIO a GPIO interface. interface. The The feedbackfeedback looploop is aimed at at controlling controlling the the energy energy consumption consumption by bymeasures measures on each on each of the of the micro-motors. The software–hardware tandem ensures motion control by experimenting with the conditions for the correct gripping of objects that are going to be set. By utilizing software limits, p = psoft, Acc = Accsoft, ◦C < XTemp, α < αsoft, the system continues with the motion or stops it. In AppendixA, the customized scheme for the EMG sensor (Figure A1) and the customized schemes for the resistive sensors (pressure, tilt) and for the gyroscope sensor (Figure A2) are presented.

3. Results 3.1. Results of Kinematic and Kinetosatic Analyses of Index Finger Mechanism The kinematic scheme of the designed prosthesis’ fingers is shown in Figure8 (where T, T1, T2, T3, T4 stands for each fingertip point). The designed mechanism enables two independent rotational motions for each finger and one rotation for each of the phalanges. Simulation of the fingertips’ trajectories, as a result of kinematic analyses (with Matlab), is evidenced in Figure9. Appl. Sci. 2021, 11, x FOR PEER REVIEW 11 of 33

micro-motors. The software–hardware tandem ensures motion control by experimenting with the conditions for the correct gripping of objects that are going to be set. By utilizing software limits, p = psoft, Acc = Accsoft, °C < XTemp, α < αsoft, the system continues with the motion or stops it. In Appendix A, the customized scheme for the EMG sensor (Figure A1) and the cus- tomized schemes for the resistive sensors (pressure, tilt) and for the gyroscope sensor (Figure A2) are presented.

3. Results 3.1. Results of Kinematic and Kinetosatic Analyses of Index Finger Mechanism Appl. Sci. 2021, 11, 3077 11 of 29 The kinematic scheme of the designed prosthesis’ fingers is shown in Figure 8 (where 𝑇, 𝑇 ,𝑇,𝑇,𝑇 stands for each fingertip point).

Appl. Sci. 2021, 11, x FOR PEER REVIEW 12 of 33

FigureFigure 8. 8. KinematicKinematic scheme scheme of ofthe the ha hand’snd’s fingers’ fingers’ mechanisms. mechanisms.

The designed mechanism enables two independent rotational motions for each finger and0.14 one rotation for each of the phalanges. Simulation of the fingertips’ trajectories, as a result of kinematic analyses (with Matlab), is evidenced in Figure 9. 0.12

0.1

0.08

0.06

0.04

0.02

0

-0.02 -0.05 O Y 0 X 0.05 -0.05 0 0.05 0.1

Figure 9. Simulation of fingertip trajectories. Figure 9. Simulation of fingertip trajectories. The kinetostatic analysis for the customized finger mechanisms was performed. For The kinetostatic analysis for the customized finger mechanisms was performed. For example, for the index finger mechanism (see Figure 10), there are three phalanges (P1, P2, example, for the index finger mechanism (see Figure 10), there are three phalanges (P1, P3) and the mechanism that enables the rotational motion of the tip and middle phalanx P2, P3) and the mechanism that enables the rotational motion of the tip and middle phal- anx(P mechanism). (P mechanism).

Figure 10. Index finger mechanism.

The mechanism is of triad structure, with B, E, and G being the input joints, and C, D, and F being the interior joints (Figure 11). For the index finger model, the kinetostatic calculation was done with Matlab software, assuming that each of the loading forces, 𝑄 (i = 1, 2, 3), are about 20 N (estimated value based on experiments with grasping relatively small objects). The simulation results are presented in Figure 12. Appl. Sci. 2021, 11, x FOR PEER REVIEW 12 of 33

0.14

0.12

0.1

0.08

0.06

0.04

0.02

0

-0.02 -0.05 O Y 0 X 0.05 -0.05 0 0.05 0.1

Figure 9. Simulation of fingertip trajectories.

The kinetostatic analysis for the customized finger mechanisms was performed. For Appl. Sci. 2021, 11, 3077 example, for the index finger mechanism (see Figure 10), there are three phalanges (P1, 12 of 29 P2, P3) and the mechanism that enables the rotational motion of the tip and middle phal- anx (P mechanism).

Figure 10. Index finger mechanism. Figure 10. Index finger mechanism. The mechanism is of triad structure, with B, E, and G being the input joints, and C, The mechanism is of triad structure, with B, E, and G being the input joints, and C, D, and F being the interior joints (Figure 11). For the index finger model, the kinetostatic D, and F being the interior joints (Figure 11). For the index finger model, the kinetostatic calculation was done with Matlab software, assuming that each of the loading forces, 𝑄 Q (icalculation = 1, 2, 3), are was about done 20 N with (estimated Matlab value software, based on assuming experiments that with each grasping of the loading relatively forces, i Appl.Appl. Sci. Sci. 2021 2021, 11, 11, x, FORx FOR PEER PEER REVIEW REVIEW( i = 1, 2, 3), are about 20 N (estimated value based on experiments with grasping1313 of of33 33 relatively small objects). The simulation results are presented in Figure 12. small objects). The simulation results are presented in Figure 12.

TriadTriad (Assur (Assur group) group)

FigureFigureFigure 11. 11. Kinetostatic 11. KinetostaticKinetostatic scheme scheme scheme of of th the ofindexe index the finger index finger mechanism. finger mechanism. mechanism.

FigureFigureFigure 12. 12. Kinetostatic 12. KinetostaticKinetostatic analysis analysis analysis of of the the ofindex index the fi indexnger—final finger—final finger—final results results for results for active active for torque. torque. active torque.

ThereThere were were two two methods methods (kinetostatic (kinetostatic and and virtual virtual power) power) used used for for determining determining the the drivingdriving torque torque values values (in (in N·m), N·m), both both resulting resulting in in identical identical values. values. This This proves proves the the correct correct calculationcalculation results. results.

3.2.3.2. Results Results of ofDesign Design and and Model Model of ofthe the Mechatronic Mechatronic System System Functionally,Functionally, the the design design of of the the mechatroni mechatronic csystem system was was focused focused on on the the similarity similarity withwith real real (people’s) (people’s) upper upper limb limb motions motions ability. ability. Each Each of of the the hand’s hand’s fingers fingers performs performs two two rotationalrotational motions motions for for the the three three phalanges phalanges of of the the whole whole finger, finger, while while the the tip tip (P3) (P3) and and pre- pre- viousvious phalanges phalanges (P2) (P2) (see (see Figure Figure 10) 10) perfor performm one one rotation. rotation. The The hand hand and and the the forearm forearm can can eacheach perform perform two two independent independent rotational rotational motions, motions, one one similar similar to to the the wrist wrist and and the the other other similarsimilar to to the the elbow. elbow. Appl. Sci. 2021, 11, 3077 13 of 29

There were two methods (kinetostatic and virtual power) used for determining the driving torque values (in N·m), both resulting in identical values. This proves the correct calculation results.

3.2. Results of Design and Model of the Mechatronic System Functionally, the design of the mechatronic system was focused on the similarity with real (people’s) upper limb motions ability. Each of the hand’s fingers performs two rotational motions for the three phalanges of the whole finger, while the tip (P3) and Appl. Sci. 2021, 11, x FOR PEER REVIEWprevious phalanges (P2) (see Figure 10) perform one rotation. The hand and the14 forearm of 33 can each perform two independent rotational motions, one similar to the wrist and the other similar to the elbow. TheThe motion motion mechanisms mechanisms are are co complexmplex and and innovative innovative from from the the point point of of view view of of mo- motion tiontransmission transmission components, components, particularly particularly due due to to thethe smallsmall dimensions and and high high geomet- geometric ricprecision precision (tolerances (tolerances of of 0.01 0.01 mm, mm, at at least). least). This This final final aspect aspect (small (small dimensions) dimensions) is is because be- causeof the of limited the limited available available space space for thefor components,the components, considering considering the contourthe contour characteristics charac- teristicsdetermined determined by the by reverse the reverse engineering engineering technique. technique. SomeSome consequent consequent stages stages and and types types of of the the mechatronic mechatronic system system design design are are presented presented in inFigure Figure 13 13..

(a) (b)

(c)

FigureFigure 13. 13. MechatronicMechatronic system system design: design: (a) (firsta) first design design type, type, (b) intermediate (b) intermediate design design type, type, (c) up (c to) up to datedate design—motion design—motion transmission transmission components components

3.3.3.3. Results Results of ofCommand Command and and Control Control System System TheThe command command and and control control system system is a iscomple a complexx one, one,briefly briefly designed designed according according to its to blockits block diagram diagram presented presented in Figure in Figure 14. Basicall 14. Basically,y, the EMG the sensor EMG sensor [25] mounted [25] mounted on the onex- the istingexisting part part of the of theupper upper limb limb will will generate generate the thesignal signal for formotion motion toward toward and/or and/or grabbing grabbing of an object (cup, ball, pencil, etc.). The resistive and infrared sensors [26] will acquire and send to the controller the data regarding prosthesis position, acceleration, pressure on the gripped object, temperature of the object, and motion towards the object. The motion to- ward the object is then available via micro-motors acting on mechanical structure (gears, joints, etc.) and assisted by the gyro sensors. Once the target is reached according to infor- mation on the software set values (psoft, Accsoft, °XTemp, αsoft) and real values from the sensors, the prosthesis motion will stop or reverse. Appl. Sci. 2021, 11, 3077 14 of 29

of an object (cup, ball, pencil, etc.). The resistive and infrared sensors [26] will acquire and send to the controller the data regarding prosthesis position, acceleration, pressure on the gripped object, temperature of the object, and motion towards the object. The motion toward the object is then available via micro-motors acting on mechanical structure (gears, joints, etc.) and assisted by the gyro sensors. Once the target is reached according to Appl. Sci. 2021, 11, x FOR PEER REVIEW information on the software set values (psoft, Accsoft, ◦XTemp, αsoft) and real values15 of 33 from

the sensors, the prosthesis motion will stop or reverse.

FigureFigure 14. Block 14. Block diagram diagram of the of command the command and andcontrol control system system for the for theupper upper limb limb prosthesis. prosthesis.

ThisThis paper paper presents presents research research results results focuse focusedd on the on theconcept, concept, design, design, prototyping prototyping and and initialinitial tests tests of the of prosthesis. the prosthesis. The command The command and control and control system’s system’s testing testingrefers to refers the pre- to the liminarypreliminary testingtesting of the sensors of the sensors and micro-motors and micro-motors on a small on a platform, small platform, such as such Arduino. as Arduino. OneOne conclusion conclusion when when testing testing the theresistive resistive sensors sensors pointed pointed out outthe theneed need for forcareful careful selectionselection of the of base the base material, material, that thatof the of fingers’ the fingers’ surface. surface. This is This because is because it should it should not be not extremelybe extremely rigid—so rigid—so as to allow as to for allow the correct for the fixture correct of fixture flexible of sensors, flexible and sensors, not too and soft— not too so thatsoft—so the material that the does material not to does overtake not to overtakedeformation deformation and thus and influence thus influence the resistive the resistive sen- sor’ssensor’s specific specific deformation. deformation. Images Images taken takenon the on preliminary the preliminary experiments experiments are areshown shown in in Figure 15. In AppendixB the program customization for the display of conventional units Figure 15. In Appendix B the program customi◦ zation for the display of conventional units for forcefor force [N] [N](Figure (Figure A3) A3 and) and tilt tiltangle angle [°] (Figure [ ] (Figure A4) A4 is )presented. is presented. Appl. Sci. 2021, 11, 3077 15 of 29 Appl. Sci. 2021, 11, x FOR PEER REVIEW 16 of 33

(a) (b)

(c)

Figure 15. ExperimentsExperiments on on the the sensors. sensors. (a ()a FSR) FSR pressure pressure sensors—for sensors—for force, force, (b) ( btilt) tiltsensor—for sensor—for finger inclinationfinger inclination angle, angle, (c) EMG (c) EMG sensor sensor test. test.

3.4. Prototyping the the Mechatronic Mechatronic System System Parts Parts One of of the the modern, modern, and and relatively relatively cost cost affo affordable,rdable, techniques techniques for forprototyping prototyping is 3D is 3D printing.printing. Mechanical components components of of the the mech mechatronicatronic system system were were prototyped prototyped using using sev- several rapideral rapid prototyping prototyping techniques techniques due due to various to various offers offers on theon the market. market. The The scope scope was was to obtainto accurateobtain accurate prototypes prototypes for the for parts the parts made made of fit of materials fit materials (polymers, (polymers, resins) resins) and, and, not not least, withleast, moderatewith moderate costs. costs. The first first technique technique used used was was FDM FDM (fused (fused deposition deposition modeling), modeling), the material the material was bi- was biodegradableodegradable thermoplastic thermoplastic polymer, polymer, PLA, PLA, and the and equipment the equipment was Creality was Creality Ender 3-Pro Ender 3- Pro[27]. [ 27Only]. Only the exterior the exterior surface surface of the of index the index finger finger was wasprinted printed for preliminary for preliminary tests tests (see (see Figure 1616).). The prototype proved to bebe accurateaccurate enoughenough forfor manuallymanually testingtesting fingerfinger mobilitymo- andbility for and preliminary for preliminary tests tests of of the the pressure pressure sensors. sensors. Still, Still, the the technique technique had had limitations limitations because of shape deviation from the nominal values (outside the acceptable upper/lower because of shape deviation from the nominal values (outside the acceptable upper/lower limits) and the material’s characteristics (relatively high density and, consequently, not limits) and the material’s characteristics (relatively high density and, consequently, not lightweight enough). lightweight enough). One more step in prototyping the mechanical components was that of considering only the motion of thumb and index fingers—due to their importance in handling everyday objects. If the mechatronic structure with only these two fingers “worked” correctly, then it could be assumed that all the rest of the structure would do so. Appl. Sci. 2021, 11, x FOR PEER REVIEW 17 of 33

Appl. Sci. 2021, 11, 3077 16 of 29 Appl. Sci. 2021, 11, x FOR PEER REVIEW 17 of 33

(a) (b) Figure 16. FDM prototype of the index finger surface. (a) 3D model, (b) FDM prototype.

One more step(a )in prototyping the mechanical components(b )was that of considering only the motion of thumb and index fingers—due to their importance in handling every- FigureFigureday objects.16. 16. FDMFDM If prototype prototypethe mechatronic of of the the index index structure finger finger surface. surface.with only (a ()a 3D) these 3D model, model, two (b (fingers)b FDM) FDM prototype. “worked” prototype. correctly, then it could be assumed that all the rest of the structure would do so. OneThe more 3D model step in to prototyping be prototyped, the conventionallymechanical components named, 2-FPRINT, was that of is considering presented in The 3D model to be prototyped, conventionally named, 2-FPRINT, is presented in onlyFigure the 17 motion. of thumb and index fingers—due to their importance in handling every- Figure 17. day objects. If the mechatronic structure with only these two fingers “worked” correctly, then it could be assumed that all the rest of the structure would do so. The 3D model to be prototyped, conventionally named, 2-FPRINT, is presented in Figure 17.

Figure 17. 3D model of the two fingers prototype, 2-FPRINT. Figure 17. 3D model of the two fingers prototype, 2-FPRINT.

The first prototyping technique considered was the one from EnvisionTEC that is The first prototyping technique considered was the one from EnvisionTEC that is based on “3SP (Scan, Spin and Selectively Photocure) technology to quickly 3D print Figurebased 17. on 3D “3SP model (Scan, of the Spin two fingers and Selectively prototype, 2-FPRINT. Photocure) technology to quickly 3D print highlyhighly accurate accurate parts parts from from STL STL files,files, regardlessregardless ofof geometricgeometric complexity” [28]. [28]. The The print- print- ing equipment is Envisiontec ULTRA 3SP, and the resin is a special one, EnvisionTEC’s ing equipmentThe first prototyping is Envisiontec technique ULTRA considered 3SP, and thewas resin the one is a specialfrom EnvisionTEC one, EnvisionTEC’s that is premium E-Model Light [29], used mainly for orthodontic purposes. Some of the complex basedpremium on E-Model“3SP (Scan, Light Spin [29 and], used Selectively mainly for Photocure) orthodontic technology purposes. to Some quickly of the 3D complex print highlycomponents accurate were parts printed, from STL so asfiles, to testregard thisless 3SP of technique’sgeometric complexity” results. Images [28]. takenThe print- while ingprototyping equipment are is shown Envisiontec in Figure ULTRA 18. 3SP, and the resin is a special one, EnvisionTEC’s premium E-Model Light [29], used mainly for orthodontic purposes. Some of the complex Appl. Sci. 2021, 11, x FOR PEER REVIEW 18 of 33

Appl. Sci. 2021, 11, 3077 17 of 29 components were printed, so as to test this 3SP technique’s results. Images taken while prototyping are shown in Figure 18.

(a)

(b)

(c) Figure 18. 3SP printing of mechanical parts. (a) Model construction and position, (b) models at the Figure 18. 3SP printing of mechanical parts. (a) Model construction and position, (b) models at the final printing process, (c) printed parts. final printing process, (c) printed parts.

After having the printed parts photocured, one important task was that to check their geometric precision (see Figure 19). Dimensions of the holes for mounting the rolling bearings and the micro-motors, respectively, as well as their tolerances, were determined. Additionally, the thread dimensions (pitch of 0.5 mm) were checked. Furthermore, the worm–worm gear transmission system was tested. Appl. Sci. 2021, 11, x FOR PEER REVIEW 19 of 33

After having the printed parts photocured, one important task was that to check their geometric precision (see Figure 19). Dimensions of the holes for mounting the rolling bear- Appl. Sci. 2021, 11, 3077 ings and the micro-motors, respectively, as well as their tolerances, were determined. Ad- 18 of 29 ditionally, the thread dimensions (pitch of 0.5 mm) were checked. Furthermore, the worm–worm gear transmission system was tested.

(a) (b)

(c) (d) Figure 19. 19. CheckingChecking the thegeometric geometric precisio precisionn of the 3SP of the printed 3SP parts. printed (a) Whole parts. dimension (a) Whole in dimension a in a section: 5.87, 5.87, instead instead of of 6 mm, 6 mm, (b) ( bmanual) manual reamer reamer to adjust to adjust the whole, the whole, (c) the ( cgear–worm) the gear–worm gear gear needed needed to be further adjusted for correct motion transmission, (d) incorrect assembly of the micro- tomotor be further and the adjustedcase part for correct motion transmission, (d) incorrect assembly of the micro-motor and the case part. The conclusion was that this printing technique did not prove to be adequate for the accurateThe and conclusion small parts was of the that mechatronic this printing system technique designed. did There not were prove tolerances to be adequate larger for the accuratethan the designed and small ones parts (0.02 of mm), the mechatronic and most hole system dimensions designed. were smaller There werethan nominal, tolerances larger thani.e., smaller the designed than the ones lower (0.02 limits. mm), Moreover, and most all holeexterior dimensions dimensions were resulted smaller in higher than nominal, i.e.,than smaller upper limit than values, the lower so that limits. additional Moreover, processing all exterior would be dimensions required. The resulted thread in pitch higher than upperwas not limit correct, values, and soalmost that additionalproved to be processing threadless. would It could be not required. be assigned The thread a set of pitch was not“rules” correct, for overcoming and almost these proved geometric to be threadless.problems, as It the could deviations not be from assigned nominal a set dimen- of “rules” for overcomingsions were not these similar geometric for all the problems,printed parts. as the deviations from nominal dimensions were not similarAnother for printing all the technique printed used parts. for the rapid prototyping of the mechatronic system parts Anotherwas DLP printing(digital light technique processing). used DLP for technology the rapid prototyping uses a digital ofprojector the mechatronic to project system an image of the layers, processed with specialized software, across the entire build plat- parts was DLP (digital light processing). DLP technology uses a digital projector to project form, curing the whole surface of the layer through a single flash. an imageThe printed of the layers,components processed of the withmechatronic specialized system software, were built across using the the entire printer build “Fig- platform, curingure 4 Standalone” the whole from surface 3DSystems. of the layer The throughmaterial used a single was flash.TOUGH-GRY 10, which is a strongThe production printed plastic components with printing of the speeds mechatronic of up to 10 systemcm/hour werethat is builtextremely using stable the printer “Figureat high humidity4 Standalone” [30]. from 3DSystems. The material used was TOUGH-GRY 10, which is a strongThe mechatronic production system’s plastic components with printing are very speeds accurate, of up with to 10 each cm/hour component that hav- is extremely stableing multiple at high fine humidity geometries, [30 and]. the details needed by the printed parts lead to the usage of a layerThe mechatronicresolution of 10 system’s microns. components are very accurate, with each component having multiple fine geometries, and the details needed by the printed parts lead to the usage of a layer resolution of 10 microns. The materials used by the Figure4 printer are all resin materials, and, because of this, the parts will need support as the build platform rises from the tank, layer by layer. All the support processing, part slicing, and optimization is performed with the help of 3D Sprint, a slicing platform software used for multiple printers, and printing technology from 3D Systems. The support structure is generated and optimized based on the part position in regard to the XY plane. Then, the downface of the components is calculated based on the premise of angle and slope, as can be seen in Figure 20. Basically, the downface is created based on the default angle of 45◦ and the gradient of 1◦ (Figure 20a). Due to the 10 micron Appl. Sci. 2021, 11, x FOR PEER REVIEW 20 of 34

The materials used by the Figure 4 printer are all resin materials, and, because of this, the parts will need support as the build platform rises from the tank, layer by layer. All the support processing, part slicing, and optimization is performed with the help of 3D Sprint, a slicing platform software used for multiple printers, and printing technology Appl. Sci. 2021, 11, 3077 from 3D Systems. The support structure is generated and optimized based on the part19 of 29 position in regard to the XY plane. Then, the downface of the components is calculated based on the premise of angle and slope, as can be seen in Figure 20. Basically, the down‐ face is created based on the default angle of 45° and the gradient of 1° (Figure 20a). Due layer height, we could change the angle to 35◦ and the gradient to 2◦. The result of the to the 10 micron layer height, we could change the angle to 35° and the gradient to 2°. The change can be seen in Figure 20b, and represents the best solution for the parts in regard to result of the change can be seen in Figure 20b, and represents the best solution for the partsthe amount in regard of to supports the amount needed of supports for printing. needed for printing.

(a)

Appl. Sci. 2021, 11, x FOR PEER REVIEW 21 of 34

(b)

Figure 20. 20. TheThe downface downface verification verification based based on angle on angle and andslope. slope. (a) Downface (a) Downface of 45°, of (b 45) downface◦,(b) downface of 35°. of 35◦. Lastly, the support points, pillars, pillar bridges, and base truss are generated. The Lastly, the support points, pillars, pillar bridges, and base truss are generated. The tip tip that is joined with the component surface is generated on the premise of penetration that is joined with the component surface is generated on the premise of penetration length, length, contact point width, etc., settings that are manually optimized based on the com‐ contact point width, etc., settings that are manually optimized based on the component ponent geometries and utility. The support structure for the printed parts is shown in geometries and utility. The support structure for the printed parts is shown in Figure 21. Figure 21.

(a) Appl. Sci. 2021, 11, x FOR PEER REVIEW 21 of 33

Figure 20. The downface verification based on angle and slope. (a) Downface of 45°, (b) downface of 35°.

Lastly, the support points, pillars, pillar bridges, and base truss are generated. The tip that is joined with the component surface is generated on the premise of penetration Appl. Sci. 2021, 11, 3077 20 of 29 length, contact point width, etc., settings that are manually optimized based on the com- ponent geometries and utility. The support structure for the printed parts is shown in Figure 21.

(a)

(b) (c)

Figure 21. 21. OptimizationOptimization and andcreation creation of the ofsupport the support structure. structure. (a) Support (a points) Support generated points by generatedthe by the software, (b) support pillars based on the generated points, (c) support settings. software, (b) support pillars based on the generated points, (c) support settings. Once the optimization and virtual build procedure ended, the printing process Once the optimization and virtual build procedure ended, the printing process started. Appl. Sci. 2021, 11, x FOR PEER REVIEWstarted. The result of the 3D printing can be seen in the images below (Figure 22). 22 of 33 The result of the 3D printing can be seen in the images below (Figure 22).

(a) (b)

(c)

Figure 22. Print job finished for mechanical components. (a) Build table optimized for 3D print, (b) printFigure job 22.finished,Print (c job) build finished platform for before mechanical alcohol part components. wash. (a) Build table optimized for 3D print, (b) print job finished, (c) build platform before alcohol part wash. The printed components shown in the figure above are what, in 3D printing, is called a “green part”. A green part, beside the resin, contains polymers, monomers, and oligo- mers. The green part, before receiving the mechanical proprieties of the material used in 3D printing, needs to be cured with ultraviolet light in a special oven, as can be seen in Figure 23.

(a) (b) (c) Figure 23. Printed green part curing and level of detail. (a) Curing ultraviolet oven, (b) gear and fillet level of detail (thread pitch of 0.5 mm), (c) worm gear level of detail (module of 0.6 mm). Appl. Sci. 2021, 11, x FOR PEER REVIEW 22 of 33

(a) (b)

(c)

Appl. Sci. 2021, 11, 3077 21 of 29 Figure 22. Print job finished for mechanical components. (a) Build table optimized for 3D print, (b) print job finished, (c) build platform before alcohol part wash. The printed components shown in the figure above are what, in 3D printing, is called a The printed components shown in the figure above are what, in 3D printing, is called “green part”. A green part, beside the resin, contains polymers, monomers, and oligomers. a “green part”. A green part, beside the resin, contains polymers, monomers, and oligo- The green part, before receiving the mechanical proprieties of the material used in 3D mers. The green part, before receiving the mechanical proprieties of the material used in printing, needs to be cured with ultraviolet light in a special oven, as can be seen in 3D printing, needs to be cured with ultraviolet light in a special oven, as can be seen in Figure 23. Figure 23.

(a) (b) (c) Appl. Sci. 2021, 11, x FOR PEER REVIEWFigure 23. Printed green part curing and level of detail. (a) Curing ultraviolet oven, (b) gear 23and of 33 Figure 23. Printed green part curing and level of detail. (a) Curing ultraviolet oven, (b) gear and filletfillet level of of detail detail (thread (thread pitch pitch of of 0.5 0.5 mm), mm), (c (c) )worm worm gear gear level level of of detail detail (module (module of of 0.6 0.6 mm). mm).

AfterAfter the the curing curing stage, stage, the the final final parts parts coul couldd be be checked checked for for geometric geometric precision. precision. The The accuracyaccuracy of the printprint dimensiondimension waswas evaluatedevaluated by by comparing comparing the the print print parts’ parts’ measurement measure- mentresults results to the to measurements the measurements done done in the in 3D the Sprint 3D Sprint software software (the (the same same with with the designedthe de- signeddimensions dimensions in CAD in software). CAD software). Not all Not components all components were measured,were measured, as for as the for verification the ver- ificationwe need we only need 2–3 only parts 2–3 from parts each from print each table. print An table. example An example of the measurement of the measurement results is resultspresented is presented in Figure in 24 Figure. 24.

FigureFigure 24. 24. StatisticsStatistics of of the the printed printed parts’ parts’ dimensions. dimensions.

TheThe most most important important and and good good aspect aspect was was th that,at, on on the the whole, whole, tolerances tolerances were were small small enoughenough (according toto lowerlower and, and, respectively, respectively upper, upper prescribed prescribed values) values) that that the plugthe plug gauge gaugecheck check was successful was successful from from the first the test.first test Assembling. Assembling the the parts parts (gears, (gears, bearings) bearings) to to test test(mechanically, (mechanically, for this for moment)this moment) the motion the moti transmissionon transmission proved proved to function to function as intended as in- in tendedthe prosthesis in the prosthesis design (see design Figure (see 25 Figure). 25). The assembly of some DLP printed parts into the 2-FPRINT prototype (see Figure 17), and thus the resulting components of this prototype, is shown in Figure 26.

(a) (b) Appl. Sci. 2021, 11, x FOR PEER REVIEW 23 of 33

After the curing stage, the final parts could be checked for geometric precision. The accuracy of the print dimension was evaluated by comparing the print parts’ measure- ment results to the measurements done in the 3D Sprint software (the same with the de- signed dimensions in CAD software). Not all components were measured, as for the ver- ification we need only 2–3 parts from each print table. An example of the measurement results is presented in Figure 24.

Figure 24. Statistics of the printed parts’ dimensions.

The most important and good aspect was that, on the whole, tolerances were small enough (according to lower and, respectively, upper prescribed values) that the plug gauge check was successful from the first test. Assembling the parts (gears, bearings) to test (mechanically, for this moment) the motion transmission proved to function as in- Appl. Sci. 2021, 11, 3077 tended in the prosthesis design (see Figure 25). 22 of 29 The assembly of some DLP printed parts into the 2-FPRINT prototype (see Figure 17), and thus the resulting components of this prototype, is shown in Figure 26.

Appl. Sci. 2021, 11, x FOR PEER REVIEW 24 of 33

(a) (b)

Appl. Sci. 2021, 11, x FOR PEER REVIEW 24 of 33

(c) (d)

Figure 25. CheckingChecking the the geometric geometric precisio precisionn of ofthe the DLP DLP printed printed parts. parts. (a) Printed (a) Printed parts, parts, (b) plug (b) plug gauge check, ( c)) correct correct gear gear assembly assembly for for motion motion transmission, transmission, (d (d) )bearings bearings mounted. mounted. (c) (d) The assembly of some DLP printed parts into the 2-FPRINT prototype (see Figure 17), Figure 25. Checking the geometric precision of the DLP printed parts. (a) Printed parts, (b) plug gaugeand thus check, the (c resulting) correct gear components assembly for of motion this prototype, transmission, is shown (d) bearings in Figure mounted. 26.

Figure 26. Components of the 2-FPRINT prototype.

4. Discussion The research results presented in this paper are focused on the concept and design of FigureFigure 26. 26. ComponentsComponents of of the the 2-FPRINT 2-FPRINT prototype. prototype. a prosthesis for the upper limb. 4.4. Discussion DiscussionCustomization of this prosthesis is an important issue. Due to the reverse engineering technique,TheThe research research size, dimensions, results results presented presented and otherin in this this geometric pape paperr are are focusedcharacteristics focused on on the the conceptof concept a real and andupper design design limb of of aawould prosthesis prosthesis be achievable for for the the upper upper with limb. limb.this prosthesis. The 3D scan of the upper limb was performed for oneCustomizationCustomization of the authors of of ofthis this this prosthesis prosthesis paper. The is is an anreverse impo importantrtant engineering issue. issue. Due Duetechnique to to the the reverse enabled reverse engineering engineeringdetermin- technique,technique,ing the value size,size, for dimensions, dimensions,any dimension, and and both other other in geometric thegeometric transversal characteristics characteristics and horizontal of aof real aplanes. real upper upper The limb design wouldlimb wouldof this prosthesisbe achievable would with match, this prosthesis. as much as The possible, 3D scan these of thedimensions, upper limb which was are performed the ones forof theone scanned of the authors upper oflimb. this Thispaper. poses The areverse real challenge engineering for the technique concept enabled and design determin- of the ingmechanical the value components, for any dimension, as there both is limited in the transversalavailable space and horizontalwhen considering planes. Thethe reversedesign ofengineered this prosthesis contour would characteristics. match, as much as possible, these dimensions, which are the ones of theThe scanned finger upper mechanism limb. Thiswas posesstudied a realfrom challenge the point for of theview concept of kinematic and design and kineto- of the mechanicalstatic analyses. components, The matrix–vector as there is method, limited availableaided by Matlabspace when software, considering was used the to reverse deter- engineeredmine the joint contour and fingertip characteristics. trajectories. Simulation results for the finger mechanism evi- dencedThe correct finger motions.mechanism was studied from the point of view of kinematic and kineto- staticThe analyses. command The matrix–vectorand control system method, is embedde aided byd andMatlab designed software, for customized was used to applied deter- minemechatronics the joint applications.and fingertip Thetrajectories. novelty ofSimulation this concept results is that for forthe eachfinger of mechanismthe five fingers, evi- denced correct motions. The command and control system is embedded and designed for customized applied mechatronics applications. The novelty of this concept is that for each of the five fingers, Appl. Sci. 2021, 11, 3077 23 of 29

be achievable with this prosthesis. The 3D scan of the upper limb was performed for one of the authors of this paper. The reverse engineering technique enabled determining the value for any dimension, both in the transversal and horizontal planes. The design of this prosthesis would match, as much as possible, these dimensions, which are the ones of the scanned upper limb. This poses a real challenge for the concept and design of the mechanical components, as there is limited available space when considering the reverse engineered contour characteristics. The finger mechanism was studied from the point of view of kinematic and kinetostatic analyses. The matrix–vector method, aided by Matlab software, was used to determine the joint and fingertip trajectories. Simulation results for the finger mechanism evidenced correct motions. The command and control system is embedded and designed for customized applied mechatronics applications. The novelty of this concept is that for each of the five fingers, there are three micro-motors to generate motion (and not servomotors, as in most available myoelectric prostheses). The combination of software and hardware (the mechatronic system included) is specific to this prosthesis. It enables the accurate trajectory of the fingertips and different set rules (gripping pressure, object temperature, acceleration toward the object). The ROM of each joint will be defined once all the tests are completed, and the validation of the results will be performed. Preliminary testing of the sensors and micro-motors on a small platform, Arduino, was performed for the prototype (two fingers, 2-FPRINT). The conclusion from testing the resistive sensors pointed out the need for the careful selection of the base material, specifically the one for the fingers’ surface. The upper limb prosthesis (whole product) will have its PCB, or micro-computer; will be versatile; and will be of small dimensions. The concept of this upper limb prosthesis is aimed at becoming a high-versatility hardware platform. Optionally, for example, a system for sound or vibration warning could be integrated if a visually impaired person was to need the prosthesis. Prototyping the mechanical components has been a challenge because of the high ac- curacy needed for the geometric precision of parts. Several techniques of rapid prototyping have been considered, out of which the DLP (Digital Light Processing) proved to be the right one. As, for the moment, only obtaining the prototype is intended, there is no ethical ap- proval asked from any authorities. Later in time, after the prototype manufacturing is com- pleted and all tests are correct, validation and, perhaps, IRB approval would be required.

5. Conclusions This paper presents aspects of the concept and design of a customized prosthesis for the upper limb. The software and hardware tandem concept of this mechatronic system enables complex motion (in the horizontal and vertical planes) with accurate trajectory and different set rules (gripping pressure, object temperature, acceleration toward the object). The concept of this upper limb prosthesis is different from the many existing prostheses existing on the market. The fingers’ motion is initiated by EMG sensor signal, and the mechanical components are driven by micro-motors, three for each finger (and not by servo-motors, as with most myoelectric prostheses). The rapid prototyping of mechanical components has been considered instead of classical manufacturing techniques (CNC machining by turning, milling, drilling, etc). There were several 3D printing techniques applied of those available and estimated to fulfil the requirements of parts (high geometric precision and accurate finish). It was only DLP (digital light processing) that enabled the manufacturing of mechanical parts with the conditions prescribed in the design. The customization of the prosthesis’ aspect and dimensions, close to those of a per- son’s real limb, is due to the reverse engineering technique. Further development of this research will aim for the careful selection of materials and manufacturing of the all the mechanical components for the upper limb prosthesis, assembling them into a whole Appl. Sci. 2021, 11, 3077 24 of 29

prototype (including micro-motors, controllers, sensors, etc.), and the complete testing and validation of results. The upper limb prosthesis prototype will be in accordance with the body features of the person wearing it, will be lightweight, be easy to maintain, and its cost will be a few thousand Euros.

Author Contributions: Conceptualization, M.I. and C.R.; methodology, M.M.R. and L.M.U.; software, L.M.U. and L.M.; validation, M.I. and C.R.; formal analysis, M.I. and L.M.U.; investigation, C.R. and Appl. Sci. 2021, 11, x FOR PEER REVIEW 26 of 33 M.M.R.; resources, L.M. and C.R.; writing—original draft preparation, M.I. and M.M.R.; writing— review and editing, M.I., M.M.R. and C.R.; supervision, M.I. All authors have read and agreed to the published version of the manuscript. InstitutionalFunding: ReviewThis Board research Statement: received noNot external applicable. funding. InformedInstitutional Consent Statement: Review Board Informed Statement: consentNot was applicable. obtained from all subjects involved in the study Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Not reported yet Data Availability Statement: Not reported yet. Acknowledgement: We especially thank CADWORKS International SRL for the support and help given inAcknowledgments: the elaboration of thisWe work. especially https:// thankwww.cadworks.ro/*** CADWORKS International (accessed SRLat 5 January for the support2021). We and help also thankgiven the in Romanian the elaboration Academy of this for work.their support https://www.cadworks.ro/*** of this research. (accessed at 5 January 2021). We also thank the Romanian Academy for their support of this research. Conflicts of Interest: The authors declare no conflict of interest. Conflicts of Interest: The authors declare no conflict of interest. Appendix A Appendix A

Figure A1. EMG schematic.

Figure A1. EMG schematic. Appl. Sci. 2021, 11, x FOR PEER REVIEW 27 of 33 Appl. Sci. 2021, 11, 3077 25 of 29

Appl. Sci. 2021, 11, x FOR PEER REVIEW 28 of 33

Figure A2. Sensors schematic. Figure A2. Sensors schematic. AppendixAppendix B B

Figure A3. Cont. Appl. Sci. 2021, 11, x FOR PEER REVIEW 29 of 33 Appl. Sci. 2021, 11, 3077 26 of 29

Figure A3. Application software for gripping force measurement. Figure A3. Application software for gripping force measurement. Appl. Sci. 2021, 11, 3077 27 of 29 Appl. Sci. 2021, 11, x FOR PEER REVIEW 30 of 33

Figure A4. Cont. Appl. Sci. 2021, 11, 3077 28 of 29 Appl. Sci. 2021, 11, x FOR PEER REVIEW 31 of 33

Figure A4. Application software for tilt angle measurement. Figure A4. Application software for tilt angle measurement.

References 1. WHO/Health topics/Disability. Available online: https://www.who.int/health-topics/disability#tab=tab_1 (accessed on 23 November 2020). 2. Kim, Y.C.; Park, C.I.; Kim, D.Y.; Kim, T.S.; Shin, J.C. Statistical analysis of amputations and trends in Korea. Prosthet. Orthot. Int. 1996, 20, 88–95. [CrossRef][PubMed] Appl. Sci. 2021, 11, 3077 29 of 29

3. Lee, K.H.; Bin, H.; Kim, K.; Ahn, S.Y.; Kim, B.O.; Bok, S.K. Hand Functions of Myoelectric and 3D-Printed Pressure-Sensored Prosthetics: A Comparative Study. Ann. Rehabil. Med. 2017, 41, 875–880. [CrossRef][PubMed] 4. Zuo, K.J.; Olson, J.L. The evolution of functional hand replacement: From iron prostheses to hand transplantation. Plast. Surg. 2014, 22, 44–51. [CrossRef] 5. Sreenivasan, N.; Gutierrez, D.F.U.; Bifulco, P.; Cesarelli, M.; Gunawardana, U.; Gargiulo, G.D. Towards Ultra Low-Cost Myoacti- vated Prostheses. Biomed. Res. Int. 2018, 2018, 9634184. [CrossRef][PubMed] 6. Oskoei, M.A.; Hu, H. Myoelectric control systems: A survey. Biomed. Signal Process. Control 2007, 2, 275–294. [CrossRef] 7. Devi, M.A.; Udupa, G.; Sreedharan, P. A novel underactuated multi-fingered soft robotic hand for prosthetic. Robot. Auton. Syst. 2018, 100, 267–277. 8. Wang, N.; Lao, K.; Zhang, X. Design and Myoelectric Control of an Anthropomorphic Prosthetic Hand. J. Bionic Eng. 2017, 14, 47–59. Available online: www.sciencedirect.com (accessed on 5 March 2021). [CrossRef] 9. Ventimiglia, P. Design of a Human Hand Prosthesis; Worcester Polytechnic Institute: Worcester, MA, USA, 2012. 10. Badawy, A.; Alfred, R. Myoelectric prosthetic hand with a proprioceptive feedback system. J. King Saud Univ. Eng. Sci. 2019, 32, 388–395. [CrossRef] 11. Richard, F. Chapter 32, Design of artificial arms and hands for prosthetic applications. In Weir, Standard Handbook of Biomedical Engineering & Design; Digital Engineering Library @ McGraw-Hill: New York, NY, USA, 2004. 12. Jamal, M.Z. Chapter 18, Signal Acquisition Using Surface EMG and Circuit Design Considerations for Robotic Prosthesis. In Computational Intelligence in Electromyography Analysis: A Perspective on Current Application and Future Challenge; Naik, G.R., Ed.; InTech: London, UK, 2012; ISBN 978-953-51-0805-4. [CrossRef] 13. Serruya, M.; Donoghue, J. Design Principles of a Neuromotor Prosthetic Device, Chapter 7.9, Series on Bioengineering and Biomedical Engineering. In Neuroprosthetics; University of Utah: Salt Lake City, UT, USA, 2004; pp. 1158–1196. [CrossRef] 14. Schmidt, H.M.; Ulrich, L. Surgical Anatomy of the Hand; Thieme: Stuttgart, Germany, 2003; p. 105, ISBN 978-1-58890-007-4. 15. Marieb, E.N. Human Anatomy & Physiology, 6th ed.; Pearson PLC: Harlow, UK, 2004; p. 237, ISBN 978-0-321-20413-4. 16. Stanišic, M.M. Mechanisms and Machines: Kinematics, Dynamics, and Synthesis; Cengage Learning: Stanford, CA, USA, 2015; ISBN-13: 978-1-133-94391-4, ISBN-10: 1-133-94391-8. 17. Zhang, Y.; Finger, S.; Behrens, S. Rapid Design through Virtual and Physical Prototyping, Chapter Introduction to Mech- anisms. pp. 39–245. Available online: https://www.cs.cmu.edu/~{}rapidproto/mechanisms/chpt4.html (accessed on 17 November 2020). 18. Erdman, A.G.; Sandor, G.N. Mechanism Design: Analysis and Synthesis; Prentice-Hall: Englewood Cliffs, NJ, USA, 1997; pp. 135–137. 19. Yamamoto, T.; Iwatsuki, N.; Ikeda, I. Automated Kinematic Analysis of Closed-Loop Planar Link Mechanisms. Machines 2020, 8, 41. [CrossRef] 20. Makkonen, P. On Multi Body Systems Simulation in Product Design. Ph.D. Thesis, KTH Royal Institute of Technology, Stock- holm, Sweden, 1999. Available online: https://www.diva-portal.org/smash/get/diva2:8521/FULLTEXT01.pdf (accessed on 5 December 2020). 21. MechDesigner & MotionDesigner R14.2. Available online: http://mechdesigner.support/index.htm?md-kinematics_dyads.htm (accessed on 10 November 2020). 22. Petrescu, F.T. Teoria Mecanismelor/Mechanism Theory; Create Space: Scotts Valley, CA, USA, 2012; ISBN 978-1-4792-9362-9. 23. Moise, V.; Simionescu, I.; Ene, M.; Rotaru, A. Analiza mecanismelor plane cu bare articulate—Aplica¸tiiîn MATLAB; Editura Printech: Bucure¸sti,Romania, 2015; ISBN 978-606-23-0403-4.

24. Moise, S, .I.; Moise, V. Matrix Vector Method for Kimematic Analysis of Spatial Mechanism with Open Contours, Research people and actual tasks on multidisciplinary science. In Proceedings of the Fifth International Conference, Lozenec, Bulgaria, 24–28 June 2015; Volume 1, pp. 248–253. 25. Senzor de Activitate Musculară MyoWare. Available online: https://www.optimusdigital.ro/ro/senzori-senzori-de-atingere/28 20-senzor-de-activitate-musculara-myoware.html (accessed on 21 December 2020). 26. FSR®400. Available online: https://www.interlinkelectronics.com/fsr-400 (accessed on 21 December 2020). 27. Radu, C.; Ro¸su,M.M.; Iliescu, M. Design and Model of a Prosthesis for Hand. In IOP Conference Series: Materials Science and Engineering, Proceedings of the ModTech International Conference Modern Technologies in Industrial Engineering, Eforie Nord, Romania, 23–27 June 2020; IOP Publishing: Bristol, UK, 2020; Volume 916. [CrossRef] 28. ULTRA 3SP and ULTRA 3SP Hi-Res. Available online: https://envisiontec.com/wp-content/uploads/2016/09/2017-ULTRA- 3SP-Series.pdf (accessed on 28 December 2020). 29. EnvisionTEC Launches Two Orthodontic Materials at LMT Lab Day Chicago. Available online: https://envisiontec.com/ orthodontic-materials-launched-at-lmt-lab-day-chicago/ (accessed on 28 December 2020). 30. Figure 4™ TOUGH-GRY 10. Available online: https://www.3dsystems.com/materials/figure-4-tough-gry-10 (accessed on 28 December 2020).