Utafiti huu unachunguza prototype ya small-scale hip exoskeleton inayoendeshwa na bidirectional artificial muscles zilizoundwa kutoka shape memory alloy (Shape Memory Alloy, SMA) springs kwa mtazamo wa thermo-mechanical characterization, temperature estimation, system identification na adaptive PID control. Opposing spring groups mbili, kila moja ikiwa na SMA springs sita, zilizalisha flexion na extension directions, na prototype ilisogezwa kati ya −20° na 200° katika characterization experiments. Kwa sababu ya difficulty ya kupima moja kwa moja temperature ya thin SMA springs, researchers walitengeneza heat-transfer-based temperature-estimation method inayounganisha thermocouple measurements na electrical voltage na current data. System identification ilitoa transfer functions kwa controller tuning; katika sinusoidal-trajectory experiment ya adaptive control, hip-angle tracking error iliripotiwa kuwa 0,94° RMSE. Hata hivyo, study inategemea small-scale experimental prototype; haionyeshi wearable exoskeleton performance kwa binadamu, long-term durability au clinical rehabilitation success.

Artificial muscle ya prototype inategemea SMA spring groups mbili zinazopingana, zilizoongozwa na antagonistic-motion principle ya human muscles. Kila group ina springs sita. Wire diameter ya spring iliyotumiwa ni 0,5 mm, outer spring diameter 3,5 mm na number of coils ni 25. SMA inapopashwa joto hubadilika kutoka martensite phase kwenda austenite phase na kujikaza kuelekea trained shorter shape; system hubadilisha linear displacement hii kuwa rotational motion kupitia transmission wire na pulley. Energizing opposing spring group hubadilisha direction ya motion.

Control system inategemea feedback mbili kuu: SMA temperature na hip-joint angle. Thermocouples nne zilitumika; thermocouples mbili zilifuatilia flexion group na mbili extension group. Angular velocity ilipimwa na inertial measurement unit (Inertial Measurement Unit, IMU), na angle ikakadiriwa kwa Kalman filter. Sampling time ya control loop ni 50 ms, frequency ni 20 Hz. PID mbili hudhibiti temperature, na adaptive PID mbili nyingine hudhibiti hip angle kulingana na movement direction.

Main engineering limitation ya study ni thermal behavior ya SMA. Joule heating husababisha contraction, huku cooling ikiachwa kwa natural ambient conditions. Researchers waliona kwamba, ikilinganishwa na previous experiments, more heating ilihitajika over time kufikia same movement distance na wakapunguza movement range hadi 20° katika subsequent adaptive-control experiments ili kupunguza risk ya thermal degradation ya SMA. Kwa hiyo, study inaonyesha kwamba high range of motion pekee haitoshi; temperature, cooling time, control accuracy na material life lazima zitathminiwe pamoja.

Kwa mtazamo wa Uturuki: Study si exoskeleton experiment iliyofanywa na human users nchini Uturuki na results haziwezi kuhamishwa directly kwenda clinical rehabilitation. Hata hivyo, SMA-based compact actuator, temperature-estimation na adaptive-control approach vinatoa engineering approach inayoweza kuchunguzwa na research groups katika universities zinazotengeneza soft robotics, rehabilitation robots na wearable-system prototypes. Ili kuhamia kwenye real-user application nchini Uturuki, full-scale prototype, user safety, thermal skin contact, active cooling, control under different loads, cycle life na long-term material fatigue lazima vivalidated separately.

Main problem ya research ni nini?

Shape memory alloys ni actuators zinazovutia katika wearable robots kwa sababu zinaweza kutoa high force katika compact size, kufanya kazi silently, na kuunda soft na flexible mechanisms ukilinganisha na traditional motors. Kwa upande mwingine, motion yake inategemea directly temperature. SMA inayocontract wakati wa heating hupoa relatively slowly; material ina hysteresis na overheating inaweza kuharibu shape-memory behavior.

Kwa hiyo, research haijafocus tu kwenye swali “je, SMA spring inaweza kusogeza hip joint?” Main engineering problem ni kufuatilia kwa reliable manner SMA spring temperature, kumodeli motion dynamics na kudhibiti target hip angle licha ya changing system behavior.

Bidirectional SMA artificial muscle inafanyaje kazi?

System ina opposing SMA spring groups mbili. Figure 2 ya study inaonyesha detailed mechanical arrangement ya small-scale prototype. Group A ina SMA springs sita, Group B ina SMA springs sita zinazofanya kazi dhidi yake. Linear movement ya springs huhamishwa kupitia steel transmission wire kwenda rotating linkage. Heating group moja husogeza linkage katika direction moja, na heating group nyingine katika opposite direction.

Figure 1 ya study inaonyesha thermo-mechanical behavior ya SMA kati ya martensite na austenite phases. Katika cold state martensite phase inatawala. Heating hubadilisha material kuelekea austenite phase na spring hurudi kuelekea trained shape yake. Researchers walitumia physical phase transformation hii kama actuation mechanism inayofanana na contraction katika biological muscle.

Spring geometry imetolewa katika source kama ifuatavyo:

Range of motion ya prototype ni kiasi gani?

Katika thermo-mechanical characterization experiments, small-scale prototype ilisogezwa kati ya −20° na 200°. Hata hivyo, katika Figure 2a ya study, normal flexion-extension motion ya human hip iliyotolewa kama reference inaonyeshwa approximately kati ya −15° extension na 120° flexion. Kwa hiyo, experimental mechanical range ya prototype haipaswi kutafsiriwa kama physiological motion limit ya human hip.

Katika subsequent adaptive-control experiments, usable movement distance ilipunguzwa intentionally hadi 20° ili kupunguza thermal-degradation risk. Change hii inaonyesha kwamba maximum mechanical range katika initial characterization experiment na safe operating range katika sustainable-control experiment si concept ileile.

Kwa nini ni vigumu kupima temperature ya thin SMA spring?

Kwa kuwa phase transformation ya SMA actuator inategemea temperature, temperature ni critical variable kwa system safety na control. Hata hivyo, si rahisi kufunga thermocouple tip reliably kwenye thin wire yenye diameter ya 0,5 mm. Sensor lazima iwe na good thermal contact na pia iwe electrically insulated.

Researchers walitumia polyimide tape kwa electrical insulation na nylon fishing line kwa mechanical fixation ili kuunganisha thermocouple tip kwenye SMA. Pamoja na hayo, kwa kuwa small contact point inaweza kushindwa wakati wa measurement, thermocouples mbili ziliwekwa kwenye kila SMA group. Hivyo total thermocouples nne zilitumika.

Pamoja na hardware redundancy hii, researchers walitengeneza computational temperature-estimation method inayoweza kutumika ikiwa sensor itashindwa. Model huhesabu SMA temperature kwa kutumia applied voltage, current, ambient temperature na experimentally calibrated thermal parameters.

Heat-transfer model ya SMA ilijengwaje?

Researchers walifafanua temperature-change rate kupitia energy balance kati ya electrical input power na heat inayohamishwa kwenda environment:

\[ \dot{T}= \frac{I V-h_cA_{sp}(T-T_{amb})} {m_{sp}C_p} \]

Hapa \(V\) ni applied voltage, \(I\) current, \(m_{sp}\) mass ya SMA spring, \(C_p\) specific heat capacity, \(T_{amb}\) ambient temperature na \(A_{sp}\) spring surface area inayobadilishana heat na environment.

Convection coefficient haikuchukuliwa kuwa constant na ilimodeliwa kwa relation:

\[ h_c=h_0+h_2T^n \]

Heat-transfer surface area ya spring imetolewa katika source kama:

\[ A_{sp}=\pi dL \]

. Hapa \(d\) ni wire diameter na \(L\) total length ya SMA spring wire.

Temperature-estimation model ilitambuliwaje kutoka experimental data?

Kwa kutoa ambient temperature:

\[ T_e=T-T_{amb} \]

ilifafanuliwa na temperature dynamics ikapunguzwa kuwa:

\[ \dot{T}_e=a_1VI+a_2T_e^n \]

. Katika first-order assumption, \(n=1\) ilitumika na definitions \(Z=\dot{T}_e\), \(X=VI\) na \(Y=T_e\) zikatolewa:

\[ [Z]=[X\;\;Y] \begin{bmatrix} a_1\\ a_2 \end{bmatrix} \]

na coefficients zikahesabiwa kutoka experimental samples kwa pseudo-inverse:

\[ \begin{bmatrix} a_1\\ a_2 \end{bmatrix} = [X\;\;Y]^{-1}[Z] \]

Same procedure ilirudiwa kwa \(n=2\) na \(n=3\). Kila model ilitathminiwa chini ya initial conditions mbili: initial temperature ikiwa measured na known, na initial temperature ikiwa unknown na assumed equal to ambient temperature.

Ni RMSE values zipi zilipatikana katika temperature estimation?

Kwa T1 wakati initial temperature ilipimwa, first-order model inatoa lowest error kwenye table kwa 0,40 °C RMSE. Initial temperature isipokuwa measured, 1,53 °C imeripotiwa kwa all three models kwa T1.

Kuna source-internal interpretation problem katika T2 results. Text inasema second-order model ilitoa lowest RMSE kwa T2; lakini wakati initial temperature imepimwa, 2,31 °C ya third-order model iko chini ya 3,78 °C ya second-order model. Wakati initial temperature haijulikani, second-order model yenye 3,51 °C inatoa lower error kuliko other two T2 estimates. Kwa hiyo, model ipi ni “best” inategemea initial condition.

Katika same section, researchers wanasema estimation error ilibaki ndani ya ±2 °C. Hata hivyo, kwa kuwa table hapo juu ina RMSE values nyingi zaidi ya 2 °C, statement hii haiwezi kugeneralized kwa all model na initial conditions.

Hip angle inapimwaje na kudhibitiwaje?

Katika prototype, IMU moja hupima angular motion. Researchers hupitisha angular-velocity data kutoka IMU kupitia Kalman filter ili kuestimate hip angle.

Total PID controllers nne zilitumika. PID controllers mbili za kwanza huregulate temperatures za SMA groups mbili. Controllers mbili nyingine huactivate flexion au extension SMA group kulingana na direction ya error kati ya desired na measured hip angle.

Temperature error imefafanuliwa kama:

\[ e_T=T_m-T_d \]

. Temperature PID control signal ni:

\[ C_{T1,T2} = K_{pT}e_T+ K_{iT}\int e_Tdt+ K_{dT}\frac{de_T}{dt} \]

. Imeelezwa specifically kwamba controllers hizi zinaactive only wakati wa heating. SMA cooling haitolewi na active system, bali hutokea kwa natural cooling kupitia ambient air.

Katika angle control, tracking error ni:

\[ e_x=\theta_{hip}-\theta_{Des} \]

na adaptive PID control signal:

\[ C_{x1,x2}(t) = K_Pe(t)+K_I\int_0^t e(t)dt+K_D\frac{d}{dt}e(t) \]

.

Adaptive PID inabadilisha nini?

Tofauti na fixed PID, adaptive structure huupdate \(K_P\), \(K_I\) na \(K_D\) gains wakati wa operation:

\[ K_P(n+1)=K_P(n)+\gamma_1e_x(t) \]

\[ K_I(n+1)=K_I(n)+\gamma_2K_I\int_0^t e_x(t)dt \]

\[ K_D(n+1)=K_D(n)+\gamma_3K_D\frac{d}{dt}e_x(t) \]

\(\gamma_1\), \(\gamma_2\) na \(\gamma_3\) ni positive learning rates. Researchers wanaeleza wazi kwamba initial PID gains na learning rates ziliamuliwa kwa experimental iterative tuning na kwamba unsuitable parameters zinaweza kupeleka system kwenye undesirable behavior.

Kwa nini system identification ilifanywa?

SMA system haifanyi kazi kama simple linear actuator kwa sababu ya temperature, phase transformation, mechanical load na natural cooling. Badala ya kuacha controller tuning kwa trial-and-error pekee, researchers walitumia MATLAB R2021a System Identification Toolbox kutoa transfer functions kutoka experimental input-output data.

Flexion na extension directions ziliidentified separately. Sababu ni kwamba system dynamics za opposing SMA groups katika directions mbili si identical.

Kwa flexion side, highest fitness kwenye table ni %96,51 kwa underdamped pole pair with delay. Hata hivyo, researchers walichagua two-pole-with-delay model yenye %95,15 fitness kwa subsequent PID tuning. Source haielezi wazi kwa nini choice hii ilipendelewa dhidi ya model yenye higher fitness.

Kwa extension side, underdamped pole pair with delay yenye %88,1 ina highest fitness kwenye table na pia ndiyo iliyochaguliwa na researchers katika controller design.

Thermo-mechanical characterization ilionyesha nini?

Katika characterization experiments, flexion motion iliundwa kwanza kupitia Group B. 10 V DC ilitumika kwenye SMA springs na PWM-controlled current ikaongezwa taratibu. Kadiri temperature ilivyoongezeka kwa Joule heating, linkage angle pia ilibadilika. Kwa Group B, heating ilisimamishwa kulingana na relevant limits kabla temperature kufikia 120 °C, hip angle 200° au experiment duration 400 seconds.

Baadaye natural cooling ilisubiriwa na reverse-direction motion ikawekwa kupitia Group A. Katika Group A experiment, temperature limit ilikuwa tena 120 °C, angle limit −20° na time limit 800 seconds. Experiments zilirudiwa mara tano ili kutathmini consistency.

Repeats tano katika Figure 13 zinaonyesha kwamba joint angle inaweza kupanda hadi upper position karibu 200° na kisha kurudi karibu −20° wakati opposing spring group inapoactivate. Temperature graph ya figure ileile inaonyesha kwamba kila spring group hupata heat rapidly wakati wa Joule heating na hupoa gradually katika ambient conditions baada ya power kukatwa.

Kwa nini movement range ilipunguzwa baadaye hadi 20°?

Researchers wanaripoti kwamba walipolinganisha current experiments na PID-control experiments zilizofanywa takriban 10 months awali, higher temperature ilihitajika kuzalisha similar angular motion. Authors wanahusisha hili na overheating ya SMA na risk ya degradation ya shape-memory properties.

Kwa hiyo, katika subsequent control experiments za adaptive PID, range of motion ililimitishwa hadi 20°. Purpose haikuwa kuzalisha bigger motion, bali kuchunguza control behavior huku SMA ikilindwa dhidi ya high temperature.

Result ya adaptive-control experiments ilikuwa nini?

Katika step-trajectory experiment iliyofanywa chini ya 20 V input, RMSE ya adaptive control ilitolewa kuwa 2,98°. Katika results section ya study, delay time ya 2,09 s, rise time ya 2,51 s na overshoot ya %10,5 zimeripotiwa kwa same step experiments.

Katika sinusoidal reference trajectory, desired na actual angle curves zilifuatana kwa kiasi kikubwa. Figure 17 inaonyesha kwamba adaptive control ilirecover trajectory tracking licha ya sudden disturbances kuongezwa kwenye angle measurement. Katika experiment hii:

RMSE = 0,94°

iliripotiwa. Result hii inawakilisha angle-tracking performance katika experimental setup ya study; si clinical accuracy value iliyopatikana katika gait ya real user au rehabilitation application.

Matokeo yanayoungwa mkono na utafiti

• Opposing spring groups mbili, kila moja ikiwa na SMA springs sita, ziliweza kutoa bidirectional joint motion katika small-scale prototype.
• Katika thermo-mechanical characterization experiments, prototype ilionyesha mechanical motion range kati ya −20° na 200°.
• Temperature-estimation approach inayounganisha voltage na current measurements na heat-transfer model ililinganishwa na experimental thermocouple data.
• Kwa T1 temperature, 0,40 °C RMSE iliripotiwa kwa first-order model na measured initial temperature.
• System identification ilionyesha different dynamic behaviors kwa flexion na extension directions.
• Adaptive PID control ilifikia angle-tracking RMSE ya 0,94° katika sinusoidal experiment.
• Thermal load ilionekana experimentally kuwa mojawapo ya main limiting factors kwa SMA range of motion na long-term usability.

Matokeo ambayo utafiti hauungi mkono au haujatesti

• Study si wearable hip-exoskeleton experiment kwenye real human.
• 0,94° RMSE si clinical rehabilitation success au patient outcome.
• Prototype motion kati ya −20° na 200° si recommended operating range ya human hip.
• Long-term cycle life ya SMA system haijaonyeshwa experimentally.
• Active cooling system haikutumika katika study hii.
• Thermal safety haijavalidated katika full-scale exoskeleton inayogusa skin.
• Haijaonyeshwa kwamba temperature-estimation method itatoa same accuracy kwa kila SMA geometry au different ambient conditions.
• Material fatigue na performance loss katika long-term use hazijatestwa comprehensively.

Mbinu na Matokeo ya Utafiti

Experimental-system summary

Thermo-mechanical relation ya SMA

Phenomenological model inayohusisha stress, temperature, shear strain na martensite fraction kwa SMA spring imetolewa katika source kama:

\[ \dot{\tau} = G\dot{\gamma} + \Omega\dot{\xi} + \Theta\dot{T} \]

Hapa \(G\) ni shear modulus, \(\Omega\) transformation tensor, \(\Theta\) thermoelastic tensor, \(\gamma\) shear strain, \(T\) temperature na \(\xi\) martensite-phase fraction. Katika source, \(\xi=1\) imefafanuliwa kuwa fully martensitic, yaani cold state; \(\xi=0\) fully austenitic, yaani hot state.

Conversion kutoka linear movement kwenda rotational movement

Linear displacement inayozalishwa na SMA spring group hubadilishwa kuwa rotation ya hip linkage kupitia pulley:

\[ \theta_x= \frac{\Delta L}{r_{pulley}} \]

\(\Delta L\) inawakilisha linear movement inayozalishwa na SMA springs, \(r_{pulley}\) pulley radius, na \(\theta_x\) rotation angle ya linkage.

Flexion system identification

Researchers walichagua two-pole-with-delay model yenye %95,15 fitness kwa PID tuning:

\[ G_{flex}(s) = \frac{648.2} {348.8s^2+37.35s+1} e^{-6.46s} \]

PID gains zilizowekwa kwa model hii:

• \(K_p = 0.00518\)
• \(K_i = 0.00001\)
• \(K_d = 0.05697\)

Katika simulation, rise time ya 8,7 s, settling time ya 56,8 s na overshoot ya %6,9 ziliripotiwa.

Extension system identification

Transfer function iliyochaguliwa kwa extension:

\[ G_{ext}(s) = \frac{0.003174} {281.2s^2+15.33s+1} e^{-22.4s} \]

PID parameters:

• \(K_p = 120.3254\)
• \(K_i = 5.7343\)
• \(K_d = 631.2162\)

Katika simulation, rise time ya 39,3 s, settling time ya 290 s na overshoot ya %8,15 ziliripotiwa. Large difference hii kati ya flexion na extension dynamics inaonyesha kwamba directions mbili hazikuwakilishwa na one identical dynamic model.

Initial parameters za adaptive-control experiment

Baada ya movement range kupunguzwa hadi 20° kwa sababu ya thermal risk, initial PID parameters katika adaptive-control experiments zilitolewa katika source kama:

• \(k_p = 0.621732 \times 255\)
• \(k_i = 0.007068 \times 255\)
• \(k_d = 0.0341826 \times 255\)

Adaptation learning rates:

• \(\gamma_1 = 6.0 \times 10^{-6}\)
• \(\gamma_2 = 10^{-7}\)
• \(\gamma_3 = 5 \times 10^{-9}\)

Experimental control results

Main engineering limitations za study

• Natural cooling time ya SMA inalimit system response.
• Excessive temperature ina risk ya degradation ya shape-memory behavior na reduction ya actuator life.
• Compared with earlier experiments, more heat ilionekana kuhitajika kwa same movement.
• Kwa hiyo movement distance ilipunguzwa hadi 20° katika subsequent experiments.
• Full-scale human-wearable exoskeleton haikutestwa ndani ya scope ya study hii.
• Long-term cycling na durability tests hazikufanywa.
• Active cooling haikutumika; researchers waliitaja kama future-work topic.

Maelezo ya Chanzo na Mbinu

Jina kamili asilia la utafiti: Temperature Estimation of Thin Shape Memory Alloy Springs in a Small-Scale Hip Exoskeleton with System Identification and Adaptive Control

Waandishi: Hussein F. M. Ali, Youngshik Kim, Ejaz Ahmad, Shuaiby Mohamed.

Author order: Order iliyotolewa katika source imehifadhiwa exactly.

Co-first / equal contribution: Hakuna co-first au equal-contribution statement katika source.

Corresponding authors: Hussein F. M. Ali na Youngshik Kim.

Taasisi:

• Department of Mechanical Engineering, Benha University, Benha, Egypt.
• Department of Smart Computing, Kyungdong University Global, Goseong, Republic of Korea.
• Department of Mechanical Engineering, Hanbat National University, Daejeon, Republic of Korea.
• Department of Mechatronics Engineering, Assiut University, Assiut, Egypt.

Aina ya chanzo: Peer-reviewed experimental research article; SMA actuator, system identification, temperature estimation na adaptive-control study.

Jarida: Actuators.

Mchapishaji: MDPI.

Bibliographic citation: Actuators 2026, 15(1), 26.

Article number: 26.

DOI: 10.3390/act15010026.

Official publication link:https://doi.org/10.3390/act15010026

Submission date: 23 August 2025.

Revision date: 23 November 2025.

Acceptance date: 16 December 2025.

Publication date: 3 January 2026.

Leseni: Creative Commons Attribution (CC BY).

Funding: Study ilipokea funding kutoka grants 2017R1A2B4008056, 2021R1H1A2093798 na 2022R1A2C1011462 zinazoungwa mkono na Korean government MSIT kupitia National Research Foundation of Korea (NRF). Pia support ya 2021RIS-004 chini ya Regional Innovation Strategy inayofadhiliwa na Ministry of Education, pamoja na support 2019H1D3A1A01071124 chini ya NRF Brain Pool program kwa first author, ziliripotiwa.

Data availability: Imeelezwa kwamba research data zinaweza kutolewa upon request.

Institutional ethics board: Source imeonyesha “Not applicable”.

Informed consent: Source imeonyesha “Not applicable”.

Conflict of interest: Waandishi walitangaza kwamba hakuna conflict of interest.

Author contributions: Conceptualization H.F.M.A. na Y.K.; methodology H.F.M.A.; software H.F.M.A. na S.M.; validation H.F.M.A. na S.M.; formal analysis na investigation H.F.M.A.; resources Y.K.; data curation H.F.M.A. na S.M.; original draft H.F.M.A., E.A. na S.M.; review na editing H.F.M.A.; visualization H.F.M.A. na S.M.; supervision, project administration na funding acquisition Y.K. zimeripotiwa.

AI-use statement: Waandishi wameeleza wazi kwamba wakati wa preparation ya article walitumia ChatGPT, model GPT-5, kwa literature-review support, text editing, language improvement na technical assistance ya LaTeX formatting. Waandishi wanasema walireview na kuedit all outputs na wanachukua responsibility ya publication content.

Scientific-content boundary: Experimental setup, formulas, temperature values, RMSE results, transfer functions, control parameters na scientific interpretations katika Verianla article hii zinategemea examined primary study. Hakuna new experimental finding kutoka external sources iliyoongezwa isipokuwa bibliographic verification.

Important inconsistencies ndani ya source

• Katika sentence moja ya temperature-estimation section, both T1 na T2 zimefafanuliwa kama “SMA springs group A”. Katika control architecture katika sections nyingine za study, T1 na T2 zinahusu opposing SMA groups mbili tofauti. Hii imehifadhiwa kama source-internal labeling inconsistency.
• Katika flexion system identification, highest fitness value ya Table 1 ni %96,51 kwa “underdamped pair with delay” model. Hata hivyo, authors wanachagua “two poles with delay” model yenye %95,15 fitness kwa controller design. Source haisemi wazi reason ya preference hii.
• Katika temperature-estimation evaluation, authors wanasema second-order model ilitoa lowest RMSE kwa T2. Lakini wakati initial temperature imepimwa, third-order model inatoa 2,31 °C huku second-order model ikitoa 3,78 °C RMSE. Wakati initial temperature haijapimwa, result ya 3,51 °C ya second-order model ndiyo lowest value.
• Ingawa source inasema temperature-estimation error ilibaki ndani ya ±2 °C, Figures 10–12 zina RMSE results zaidi ya 2 °C kama 2,31, 2,64, 3,51, 3,78, 4,15, 4,44, 7,65 na 13,75 °C. Kwa hiyo, ±2 °C statement haijawasilishwa kama general result inayotumika kwa all test conditions.
• Katika explanatory text ya thermo-mechanical characterization, Figure 13b imeelezwa kuwa temperature na Figure 13c angle; katika visible figure na figure caption, 13b ni joint angle na 13c ni temperature.

Scientific-interpretation boundary: Research inachunguza small-scale experimental prototype. Prototype kuonyesha mechanical motion kati ya −20° na 200° haimaanishi kwamba range hii ni safe au physiological kwa human hip. 0,94° RMSE katika adaptive-control experiment pia haiwakilishi performance katika real-patient rehabilitation. SMA temperature ni important safety issue kwa human contact na researchers wanaeleza wazi kwamba active cooling na long-term durability studies zinahitajika katika future work.