On The Kinematic Design of Anthropomorphic Lower Limb Exoskeletons and Their Matching Movement
On The Kinematic Design of Anthropomorphic Lower Limb Exoskeletons and Their Matching Movement
On The Kinematic Design of Anthropomorphic Lower Limb Exoskeletons and Their Matching Movement
Abstract
The lower limb exoskeleton is a wearable device for assisting medical rehabilitation. A classical lower limb exoskeleton
structures cannot precisely match the kinematics of the wearer’s limbs and joints in movement, so a novel anthro-
pomorphic lower limb exoskeleton based on series–parallel mechanism is proposed in this article. Then, the human lower
limb movements are measured by an optical gait capture system. Comparing the simulation results of the series–parallel
mechanism with the measured human data, the kinematics matching model at the hip joint is established. The results show
that the kinematic matching errors in the X, Y, and Z directions are less than 2 mm. So, the proposed kinematics matching
model is effective and the anthropomorphic series–parallel mechanism has a significant improvement in tracing the human
positions at the hip joint.
Keywords
Lower limb exoskeleton, series–parallel mechanism, kinematics matching model, human hip joint, optical gait capture
system
Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 International Journal of Advanced Robotic Systems
joint with three DOFs, that is, flexion/extension (FE), velocity in various moving situations, such as walking or
abduction/adduction (HAA), and internal/external rotation climbing stairs. Compared to exoskeleton-based systems
(HIE). The knee joint is considered as a revolute joint that can support the knee during the stance phase, the
between the femur and tibia, which has one DOF, that is, end-effectors based devices may require manual assistance
flexion/extension (KFE). The ankle joint which is similar to during stance phase. Rutgers Ankle20,21 is a typical ankle
the hip joint has three DOFs, that is, dorsiflexion/plantar- rehabilitation robot based on Stewart platform, which is a
flexion (ADP), abduction/adduction (AAA), and internal/ 6-DOF parallel mechanism. The movement of this parallel
external rotation (AIE).2 Therefore, the LLE should be mechanism is realized by the synergistic movement of its
designed with seven DOFs for each leg to follow the loco- six electric cylinders. A parallel robot for ankle rehabili-
motion generated by the wearer in an anthropomorphic and tation22 was proposed in Italian Institute of Technology to
ergonomic way.6 carry out the required exercises, but this device only
Serial mechanisms are adopted in many multiple-joint allows movements in plantar/dorsiflexion and inversion/
exoskeleton systems, which have larger workspace eversion. In addition to the research above, there are many
compared to the parallel mechanisms. With an anthropo- exoskeletons based on parallel mechanisms, such as ankle
morphic design, BLEEX7,8 has two three-segment-linkage rehabilitation robot ARBOT 23 with 3UPS/U1 parallel
legs, which are analogous to the human thigh, shank, and mechanism developed by Saglia et al. A wearable ankle
foot. Each leg has seven DOFs whose HFE, HAA, HFE, parallel robot (PMAs)24,25 is developed by Jamwal et al.
and ADP are powered by the actuator, while HIP, AIE and and a 3-DOF ankle parallel robot is developed by Xie and
AAA that equipped with mechanical impedances are pas- Jamwal26 from the University of Auckland.
sive. The commercial lower limb exoskeleton {Hybrid The series–parallel LLE combines the advantages of
Assistive Limb} (LLE HAL-5)9–11 is designed with six series and parallel mechanisms, especially the sufficient
DOFs in total. It provides the motion assistance on the active DOFs to maximize the anthropomorphism. Pan
sagittal plane. The HFE and KFE joints are driven by direct et al.27 proposed a novel LLE model with a 6-DOF (four
current servomotors mounted with harmonic reducer, while active DOFs and two passive DOFs) series–parallel
the movement of ADP is passive. There are also some mechanism. And the kinematic accuracy, reliability,6 and
discussions in underactuated Robotics,12 redundant parallel dynamic characteristics28,29 of this device are studied to
manipulator,13 and series–parallel hybrid mechanics.14 The explore and find out the characteristics of this novel LLE
ReWalk Rehabilitation and ReWalk Personal 6.0 devel- model.
oped by Argo Medical Technologies are commercialized Inspired by the research results achieved in the litera-
LLE to assist patients with spinal cord injury to walk again ture,3,25–27 a new novel bionic LLE model is proposed
independently.15,16 Wearers equipped with these devices based on the series–parallel mechanism. This LLE model
need to hold on crutches to maintain body balance during has twenty DOFs, with one 6-spherical–prismatic–spheri-
walking. ReWalk has six DOFs. The HFE and KFE are cal (SPS) parallel mechanism at the pelvis, two 3-SPS par-
actuated by motors to move the wearer’s legs, while the allel mechanisms at the thighs and foots, one 1-SPS serial
ADP moves passively by using a spring. Most of these mechanism at the shank.By analyzing this series–parallel
existing multijoint exoskeletons with serial mechanisms hybrid structures, we obtained the analysis result that this
are less than seven DOFs, which means it is hard for these model can realize a more anthropomorphic gait pattern,
devices to accurately match the lower limb kinematics of while the motion of some existing LLEs is limited to the
wearer. sagittal plane through the comparison between optical
Existing limb exoskeletons based on parallel mechan- motion capture experiment results and kinematical analysis
isms are mainly single-joint exoskeletons used for hip or of our proposed model. The assessment of the synchronous
ankle joint rehabilitation. Parallel mechanisms have better effect of our model at the hip joint indicates this series–
precision, stiffness, and load-carrying capacity than serial parallel hybrid structure can realize a good matching result
mechanisms,17 and it can match the kinematics of the wear- at the hip joints during the movement of the wearer, and
er’s single joint accurately. Therefore, the parallel mechan- further research will verify the matching effect at the knee
ism is suitable for the configuration design of the pelvis and joint and ankle joint.
the hip joint due to its abundant configuration modes, var- The rest of this article is organized as follows: In the
ious DOFs, and flexible wearing forms. Yu et al.18 pro- second section, we propose the anthropomorphic series–
posed a parallel hip joint exoskeleton arranged in a parallel hybrid mechanism of LLE. In the third section, the
parallel kinematic chain fashion, which consists of a 6- kinematic model of the hip joint in the series–parallel LLE
DOF platform, with three universal–prismatic–spherical is established. In the fourth section, we design a gait cap-
serial chains per side, surrounding the user hip articulation. ture experiment to evaluate the matching effect of the
Yoon et al.19 presented a 6-DOF gait rehabilitation robot. anthropomorphic series–parallel mechanism and assess the
The foot end-effector of this gait rehabilitation robot is kinematics matching model at the hip joint. In the fifth
designed as a parallel mechanism driven by two linear section, discussion and conclusion, and some related
actuators and it allows patients to alter their walking research that still need to be further studied are given.
Ren et al. 3
Figure 4. (a) Squatting gait and (b) layout diagram of markers and coordinate system.
Figure 6. Three-dimensional coordinate comparison of marker 3 (on human body) and point B1 (on exoskeleton).
Figure 7. Error of marker 3 (on human body) and point B1 (on exoskeleton).
In Figure 6, the abscissa is the gait period, and the ordi- angle in YOZ plane, which is the rotation angle of the hip
nate is the coordinate value of X, Y, and Z. X coordinate is the joint around X-axis, is the smallest and the smoothest. In
lateral displacement of the hip joint (marker 3) when lower addition, the peak errors are concentrated at 40% of the gait
limbs move, Y coordinate is the tilt displacement when mov- cycle. When the gait cycle is 40%, the test is switched from
ing forward, and Z coordinate is the vertical displacement. squatting to standing. At this time, no matter the coordi-
The comparison results indicate that the absolute value nates or the projection angle, there is relatively large track-
of motion error in the X, Y, and Z directions are less than 3.5 ing error.
mm, and the absolute value of angle error projected into
coordinate plane is less than 2 . Among them, the Y direc-
tion, which is the forward direction during walking, has the Discussion and conclusion
smallest error and the smallest error fluctuation in the In the process of squatting down and standing up, the hip
whole gait cycle. For projection angle, error of projection joint has lateral stride movement, forward flexion and
Ren et al. 7
extension, and rotation around the vertical direction. Typ- hip of the lower limbs. It can realize the walking function in
ical LLE robots, such as those used in medical rehabilita- all directions, so as to achieve the goal of accurately match-
tion, can only move in a single plane and cannot actively ing the movement of the wearer’s lower limbs and joints in
realize lateral stride and rotation around the vertical direc- kinematics.
tion. HAL is the typical commercial exoskeleton, used for When designing the series–parallel LLE, the accessi-
lower limb assistance, which can only move in sagittal ble workspace of the parallel mechanism at each joint
plane, so that it cannot meet the needs of lower limb assis- should also be considered. The isomerism of human–
tance in outdoor auxiliary space. For a lower limb paralysis, machine system is a common problem in existing LLE
passive freedom is difficult to play a role because the user systems. There are two main consequences caused by the
cannot control the movement of the thigh. In this article, the human–machine isomerism system: (1) the absence of
anthropomorphic LLE is based on a 20-DOF series–paral- exoskeleton structure movement space compared with the
lel mechanism, which is active in any direction to drive the wearer’s movement space which will result in the
8 International Journal of Advanced Robotic Systems
IEEE 13th international conference on rehabilitation 23. Saglia JA, Tsagarakis NG, Dai JS, et al. Control strategies
robotics (ICORR), Seattle, WA, USA, 24 June–26 June for patient-assisted training using the ankle rehabilitation
2013, pp. 1–5. robot (ARBOT). IEEE/ASME Trans Mech 2012; 18:
16. Zeilig G, Weingarden H, Zwecker M, et al. Safety and toler- 1799–1808.
ance of the ReWalk™ exoskeleton suit for ambulation by 24. Jamwal PK, Xie SQ, Hussain S, et al. An adaptive wearable
people with complete spinal cord injury: a pilot study. parallel robot for the treatment of ankle injuries. IEEE/ASME
J Spinal Cord Med 2012; 35: 96–101. Trans Mech 2012; 19: 64–75.
17. Fan Y, Guo Z, and Yin Y. SEMG-based neuro-fuzzy control- 25. Tsoi Y, Xie S, and Mallinson G. Joint force control of par-
ler for a parallel ankle exoskeleton with proprioception. Int J allel robot for ankle rehabilitation. In: Proceedings of 2009
Robot Autom 2011; 26: 450. IEEE international conference on control and automation,
18. Yu Y, Liang W, and Ge Y. Jacobian analysis for parallel Christchurch, New Zealand, December–11 December 2009,
mechanism using on human walking power assisting. In: pp. 1856–1861.
Proceedings of 2011 IEEE international conference on 26. Xie SQ and Jamwal PK. An iterative fuzzy controller for
mechatronics and automation, Beijing, China, 7 August–10 pneumatic muscle driven rehabilitation robot. Exp Syst Appl
August 2011, pp. 282–288. 2011; 38: 8128–8137.
19. Yoon J, Novandy B, Yoon CH, et al. A 6-DOF gait rehabili- 27. Pan D, Gao F, Miao Y, et al. A new study on the relative
tation robot with upper and lower limb connections that kinematic accuracy reliability of a novel exoskeleton with
allows walking velocity updates on various terrains. IEEE/ series–parallel topology. Mech Base Des Struct Mach 2015;
ASME Trans Mech 2010; 15: 201–215. 43: 383–406.
20. Girone M, Burdea G, Bouzit M, et al. A Stewart platform- 28. Pan D, Gao F, Huang X, et al. Kinematic accuracy reliability
based system for ankle telerehabilitation. Auton Robot 2001; research of a novel exoskeleton with series–parallel topology.
10: 203–212. Proc Inst Mech Eng C J Mech Eng Sci 2014; 228: 1767–1776.
21. Meng W, Liu Q, Zhou Z, et al. Recent development of 29. Pan D, Gao F, and Miao Y. Dynamic research and analyses of
mechanisms and control strategies for robot-assisted lower a novel exoskeleton walking with humanoid gaits. Proc Insti-
limb rehabilitation. Mechatronics 2015; 31: 132–145. tut Mech Eng C J Mech Eng Sci 2014; 228: 1501–1511.
22. Saglia JA, Tsagarakis NG, Dai JS, et al. A high-performance 30. Sandau M, Koblauch H, Moeslund TB, et al. Markerless motion
redundantly actuated parallel mechanism for ankle rehabili- capture can provide reliable 3D gait kinematics in the sagittal
tation. Int J Robot Res 2009; 28: 1216–1227. and frontal plane. Med Eng Phys 2014; 36: 1168–1175.