Tesi etd-04102017-204001
Link copiato negli appunti
Tipo di tesi
Perfezionamento
Autore
D'ELIA, NICOLO
Indirizzo email
nico.delia@hotmail.it
URN
etd-04102017-204001
Titolo
Ergonomics of Wearable Robots
Settore scientifico disciplinare
ING-IND/34
Corso di studi
INGEGNERIA - Biorobotics
Commissione
relatore Prof. VITIELLO, NICOLA
Parole chiave
- Active Pelvis Orthosis
- Ergonomics
- Passive degrees of freedom
- Wearable robotics
Data inizio appello
05/06/2017;
Disponibilità
completa
Riassunto analitico
In the field of human-centered robotics, exoskeletons are becoming relevant for addressing needs in the healthcare and industrial domains, both as tools for rehabilitation treatment and clinical assessment and for augmented reality applications (haptics or augmentation). Despite the increasing interest and number of developed prototypes and commercial systems, the design of exoskeletons still has many open issues, such as those related to the development of the physical human-robot interface. Owing to their close interaction with the user, safety and ergonomics are critical features that heavily influence the functionality and the dependability of a wearable robot. In general, these devices are designed to generate and transfer mechanical power to human joints: therefore, optimal kinematic coupling is required between the corresponding human and robot rotation axes.
Misalignment between the human and robot joint axes can cause undesired forces that overload human articulations, thus resulting in an uncomfortable or even painful interaction with the robot. Undesired forces originating from joint axis misalignments can also lead the orthotic shells of the exoskeleton to slide along the human limb segments, leading to unreliable assistive torque transmission and possible skin inflammation or even sores.
As a consequence, modern exoskeletons are provided with adjustable regulations and/or passive degrees of freedom (pDoFs), in accordance with recent ergonomic design principles for the development of wearable robots. However, the introduction of passive DoFs into the design of a WR is not free of drawbacks; the tradeoff between the degrees of laxity and the system complexity may affect the overall human-robot kinematics coupling. On the one hand, by increasing the degree of laxity of the powered joints, there is a risk of increasing the overall inertia and friction of the moving parts. On the other hand, a lack of adequate laxity partially affects the human-robot joint axis self-alignment and thus hinders the spontaneous movement of the user. As a consequence, in the development and design of an exoskeleton, the assessment of its kinematic compatibility with user biomechanics is of paramount importance. In particular, it is important the evaluation of ergonomics. Anyhow, a systematic evaluation methodology for the assessment and benchmark of ergonomics is still missing in literature.
In this thesis, we propose a methodology for the assessment of ergonomics of wearable robots. In particular, by taking as a test bed an active pelvis orthosis (APO) for hip flexion/extension assistance, we carried out a set of experiments where we assessed ergonomics against three main factors at the same time: i) the deviation from natural walking (NW) kinematics deriving from wearing the device in transparent mode (TM)—the APO shadows the wearer in zero–torque control—and in assistive mode (AM) with different levels of assistance—i.e. the APO deliver torque to the wearer; ii) the stability of the human-robot interaction; and iii) the discrepancy between the human and robotic hip joint kinematics. These three factors led to the definition of three quantitative indicators:
• The root mean square (RMS) of the difference of the human hip, knee, and ankle flexion/extension (f/e) angle between the NW and TM/AM conditions; we named these variables ‘human hip angle deviation’ (H-HAD), ‘human knee angle deviation’ (H-KAD), and ‘human ankle angle deviation’ (H-AAD), respectively.
• The standard deviation (SD) of the relative displacements between the markers placed on the orthotic cuffs of the APO and those placed on their corresponding body segments; we named this set of variables ‘physical human-robot interface displacement’ (pHR-ID); (iv) the human joints’ ROM in the sagittal plane (hip, knee, and ankle) and in the frontal plane (only at the hip); and (v) spatio-temporal parameters (i.e., step length, stance time, and cadence).
• The RMS of the difference between the APO hip f/e angle and the anatomical hip f/e angle; we named this variable ‘human-robot hip angle deviation’ (HR-HAD).
The proposed metrics was validated in two experimentations. In the first study, we reported the methods and results of an experiment with healthy users engaged in tasks of treadmill–based gait while the APO provided hip flexion-extension at different speeds and assistance levels. In particular we wanted to understand whether the pDoFs of the APO hip joint allowed an appropriate human-robot joint axes self-alignment.
The results show: (i) negligible interference of the APO in human kinematics under all the experimented conditions: H–HAD, H–KAD, H–AAD below 3.9°, 4.0°, 3.8° respectively; (ii) high stability (pHR-ID < 1 cm) between the APO cuffs and the corresponding body segments; and (iii) human-robot kinematics discrepancy at the hip flexion-extension joint (HR-HAD) remains below 7.2°.
Hence, APO mechanics and actuation have negligible interference in human locomotion. Human kinematics was not affected by the APO under all tested conditions. In addition, under all tested conditions, there was no relevant relative displacement between the orthotic cuffs and the corresponding anatomical segments. Thereby, the physical human-robot coupling is reliable. These facts prove that the adopted mechanical design of passive degrees of freedom allows an effective human–robot kinematic coupling.
In the second study, we reported the methods and results of the experiment with healthy users engaged in tasks of ground–level walking and stair ascending in ecological conditions (no treadmill). A new version of the APO (which feature halved weight and no intra/extra rotation pDoF) provided hip f/e at different speeds and assistance levels. The final goals of this study were: (i) test the applicability of the proposed metrics to different tasks and conditions of walking; (ii) compare and benchmark the ergonomics of the previous version of the APO with the new one.
The results show (i) negligible interference of the APO onto human kinematics in ground–level walking (H–HAD, H–KAD, H–AAD below 3.0°, 5.0°, 3.0° respectively) but non–negligible impact in stair ascending (H–HAD, H–KAD, H–AAD below 9.0°, 9.0°, 5.0° respectively); (ii) high stability (pHR-ID < 1 cm) between the APO cuffs and the corresponding body segments under both locomotion tasks; and (iii) HR-HAD below 8.0° and 9.0° ground–level walking and stair ascending respectively.
Two different conclusions regarding APO ergonomics can be reached for ground–level walking and stair ascending. In the case of ground–level walking, since the physiological kinematics is not severely altered (low H-HAD, H-KAD, H-AAD), pHR-ID are negligible (< 1 cm) and all subjects did not report discomfort while walking, we may conclude the APO to be ergonomic, regardless the possible JAxM. Interestingly, in ground level walking, the reduced self-alignment capability (with respect to previous version), do not bring a worsening in the ergonomics of the device in terms of the analyzed indicators.
Instead, in the case of stair ascending, we observed low (< 1 cm) pHR-ID, similar values of HR-HAD and no reported discomfort by the wearer, but natural kinematics seems to be relevantly altered. This result evidences a minor kinematic compatibility between the wearer and the robot suggesting a revision of its kinematic design for stair ascending tasks.
In conclusion, the proposed metrics proved to be a reliable tool to investigate the ergonomics of lower limb wearable robots and assess their performance. Furthermore, it can be a valid aid to benchmark and compare the several platforms which have been developed in the state of the art and will be developed in the future.
Misalignment between the human and robot joint axes can cause undesired forces that overload human articulations, thus resulting in an uncomfortable or even painful interaction with the robot. Undesired forces originating from joint axis misalignments can also lead the orthotic shells of the exoskeleton to slide along the human limb segments, leading to unreliable assistive torque transmission and possible skin inflammation or even sores.
As a consequence, modern exoskeletons are provided with adjustable regulations and/or passive degrees of freedom (pDoFs), in accordance with recent ergonomic design principles for the development of wearable robots. However, the introduction of passive DoFs into the design of a WR is not free of drawbacks; the tradeoff between the degrees of laxity and the system complexity may affect the overall human-robot kinematics coupling. On the one hand, by increasing the degree of laxity of the powered joints, there is a risk of increasing the overall inertia and friction of the moving parts. On the other hand, a lack of adequate laxity partially affects the human-robot joint axis self-alignment and thus hinders the spontaneous movement of the user. As a consequence, in the development and design of an exoskeleton, the assessment of its kinematic compatibility with user biomechanics is of paramount importance. In particular, it is important the evaluation of ergonomics. Anyhow, a systematic evaluation methodology for the assessment and benchmark of ergonomics is still missing in literature.
In this thesis, we propose a methodology for the assessment of ergonomics of wearable robots. In particular, by taking as a test bed an active pelvis orthosis (APO) for hip flexion/extension assistance, we carried out a set of experiments where we assessed ergonomics against three main factors at the same time: i) the deviation from natural walking (NW) kinematics deriving from wearing the device in transparent mode (TM)—the APO shadows the wearer in zero–torque control—and in assistive mode (AM) with different levels of assistance—i.e. the APO deliver torque to the wearer; ii) the stability of the human-robot interaction; and iii) the discrepancy between the human and robotic hip joint kinematics. These three factors led to the definition of three quantitative indicators:
• The root mean square (RMS) of the difference of the human hip, knee, and ankle flexion/extension (f/e) angle between the NW and TM/AM conditions; we named these variables ‘human hip angle deviation’ (H-HAD), ‘human knee angle deviation’ (H-KAD), and ‘human ankle angle deviation’ (H-AAD), respectively.
• The standard deviation (SD) of the relative displacements between the markers placed on the orthotic cuffs of the APO and those placed on their corresponding body segments; we named this set of variables ‘physical human-robot interface displacement’ (pHR-ID); (iv) the human joints’ ROM in the sagittal plane (hip, knee, and ankle) and in the frontal plane (only at the hip); and (v) spatio-temporal parameters (i.e., step length, stance time, and cadence).
• The RMS of the difference between the APO hip f/e angle and the anatomical hip f/e angle; we named this variable ‘human-robot hip angle deviation’ (HR-HAD).
The proposed metrics was validated in two experimentations. In the first study, we reported the methods and results of an experiment with healthy users engaged in tasks of treadmill–based gait while the APO provided hip flexion-extension at different speeds and assistance levels. In particular we wanted to understand whether the pDoFs of the APO hip joint allowed an appropriate human-robot joint axes self-alignment.
The results show: (i) negligible interference of the APO in human kinematics under all the experimented conditions: H–HAD, H–KAD, H–AAD below 3.9°, 4.0°, 3.8° respectively; (ii) high stability (pHR-ID < 1 cm) between the APO cuffs and the corresponding body segments; and (iii) human-robot kinematics discrepancy at the hip flexion-extension joint (HR-HAD) remains below 7.2°.
Hence, APO mechanics and actuation have negligible interference in human locomotion. Human kinematics was not affected by the APO under all tested conditions. In addition, under all tested conditions, there was no relevant relative displacement between the orthotic cuffs and the corresponding anatomical segments. Thereby, the physical human-robot coupling is reliable. These facts prove that the adopted mechanical design of passive degrees of freedom allows an effective human–robot kinematic coupling.
In the second study, we reported the methods and results of the experiment with healthy users engaged in tasks of ground–level walking and stair ascending in ecological conditions (no treadmill). A new version of the APO (which feature halved weight and no intra/extra rotation pDoF) provided hip f/e at different speeds and assistance levels. The final goals of this study were: (i) test the applicability of the proposed metrics to different tasks and conditions of walking; (ii) compare and benchmark the ergonomics of the previous version of the APO with the new one.
The results show (i) negligible interference of the APO onto human kinematics in ground–level walking (H–HAD, H–KAD, H–AAD below 3.0°, 5.0°, 3.0° respectively) but non–negligible impact in stair ascending (H–HAD, H–KAD, H–AAD below 9.0°, 9.0°, 5.0° respectively); (ii) high stability (pHR-ID < 1 cm) between the APO cuffs and the corresponding body segments under both locomotion tasks; and (iii) HR-HAD below 8.0° and 9.0° ground–level walking and stair ascending respectively.
Two different conclusions regarding APO ergonomics can be reached for ground–level walking and stair ascending. In the case of ground–level walking, since the physiological kinematics is not severely altered (low H-HAD, H-KAD, H-AAD), pHR-ID are negligible (< 1 cm) and all subjects did not report discomfort while walking, we may conclude the APO to be ergonomic, regardless the possible JAxM. Interestingly, in ground level walking, the reduced self-alignment capability (with respect to previous version), do not bring a worsening in the ergonomics of the device in terms of the analyzed indicators.
Instead, in the case of stair ascending, we observed low (< 1 cm) pHR-ID, similar values of HR-HAD and no reported discomfort by the wearer, but natural kinematics seems to be relevantly altered. This result evidences a minor kinematic compatibility between the wearer and the robot suggesting a revision of its kinematic design for stair ascending tasks.
In conclusion, the proposed metrics proved to be a reliable tool to investigate the ergonomics of lower limb wearable robots and assess their performance. Furthermore, it can be a valid aid to benchmark and compare the several platforms which have been developed in the state of the art and will be developed in the future.
File
| Nome file | Dimensione |
|---|---|
| TESI_Nic...iewed.pdf | 3.84 Mb |
Contatta l'autore |
|