This article describes the design of Austin exoskeleton – a minimally actuated medical exoskeleton with mechanical swing-phase gait generation and sit-stand assistance. The Austin exoskeleton is an accessible lightweight system that enables individuals with paraplegia to walk. The gait generation hardware of the Austin exoskeleton suit consists of three major components: hip actuation, a hip-knee coupler, and a controllable locking knee. Users operate the exoskeleton with a simple wireless user interface consisting of two push buttons that are installed on the handle of the stability aid. Electrical components are located on the back of the exoskeleton. A single actuator per leg and a mechanical hip-knee coupler power the knee during swing phase and provide assistance for sitting and standing. The suit’s design embeds gait generation into hardware, decreasing controller complexity. By using a bio-inspired coupling mechanism, the Austin system is able to power both the hip and knee joints using a single hip actuator.

Article

The AUSTIN exoskeleton (Figure 1) is an accessible lightweight system that enables individuals with paraplegia to walk. A single actuator per leg and a mechanical hip-knee coupler power the knee during swing phase and provides assistance for sitting and standing. The suit's design embeds gait generation into hardware, decreasing controller complexity.

graphic

In 2012, an estimated 270,000 individuals in the U.S were afflicted by spinal cord injury (SCI), according to the National Spinal Cord Injury Statistical Center (NSCISC). Fortythree percent of the estimated 12,000 new spinal cord injuries sustained each year result in paraplegia.

Because they are confined to a wheelchair, SCI patients risk developing secondary injuries and medical complications such as acute pressure ulcer development, osteoporosis, decreased range of joint motion, urinary tract infection, and impaired respiratory and cardiovascular functions 22,33.

Devices to reduce secondary injuries

Passive orthoses are commonly prescribed by doctors as means of delaying the onset of secondary injuries 44. Devices such as the knee-ankle foot orthosis (KAFO) or reciprocating gait orthoses (RGO), as shown in Figure 2, are designed to enable users to achieve a quasi bipedal gait by swinging one leg in front of the other using upper body strength. Although passive orthotics are available for personal ownership, studies have shown that the majority of RGO users discontinued use within 5 years due to rapid fatigue and sluggish ambulation.

graphic

Powered Exoskeletons

In an effort to improve walking energetics, several research groups have developed powered exoskeleton systems that use actuators to assist with locomotion. Figure 3 shows four of the most well-known powered exoskeletons today. ReWalk, eLEGS and Vanderbilt University Exoskeleton use four electric motors to power the knee and hip joint motions along the sagittal plane. The Rex exoskeleton uses ten joint actuators to power the hip, knee, and ankle 66. The more involved actuation scheme of the Rex generates a very slow walking gait that puts the user through a series of quasi-statically balanced postures.

graphic

Coupling inspired by biomechanics

By using a bio-inspired coupling mechanism, the Austin system is able to power both the hip and knee joints using a single hip actuator. Hip-knee joint coupling refers to motion where the knee flexes simultaneously with hip flexion and extends simultaneously with hip extension. Biologically, this happens partially due to biarticular muscles. Biarticular muscles such as the hamstrings act across multiple joints to produce coupled motion. Various examples of this coupling behavior during walking, sitting, and standing are described in the literature 77,99. To more clearly illustrate this coupling relationship, Figure 4 shows a simulated human leg motion with a fixed hamstring length while flexing the hip.

graphic

Joint Coupling Walking Mechanics

The human gait cycle is commonly divided into stance and swing phase. Swing phase can be further divided into two phases: swing-flexion and swing-extension as illustrated in Figure 51010.

graphic

PHASE I:Stance During stance, the person's foot is in contact with the ground and the leg is bearing weight. The hip goes through extension while the knee stays locked.

PHASE II:Swing-flexion At maximum hip extension, swing-flexion phase begins. In this phase, both the knee and hip flexes to achieve toe clearance.

PHASE III:Swing-extension At maximum knee flexion, swing-extension phase begins and the knee re-extends to prepare for heel strike.

Mechanical Gait Generation

Knee actuators can be replaced by a joint coupling system so that only a single actuator per leg is required to produce the three-phase gait cycle. In this system, knee motion becomes a function of hip motion and the knee has two states. In the coupled state, activated during swing-flexion, the knee will flex and extend with the hip, powered by a coupling mechanism that transfers power to it from the hip. During the uncoupled state, the knee is locked against flexion but still free to extend, allowing the knee-spring to fully extend the leg during swing-extension and provide locking during stance.

A walking gait is generated by toggling between the coupled and uncoupled states, in the following three steps (see Fig. 5).

1 During stance, the hip and knee are uncoupled and the knee stays locked while the hip goes through extension.

2 Going into swing-flexion, the hip and knee are coupled to flex simultaneously as observed in a natural human gait.

3 At the point of maximum toe clearance, the coupling mechanism is deactivated, allowing the knee to extend and prepare the leg for stance.

System Architecture

The gait generation hardware of the Austin exoskeleton suit consists of three major components: hip actuation, a hip-knee coupler, and a controllable locking knee. Users operate the exoskeleton with a simple wireless user interface consisting of two push buttons that are installed on the handle of the stability aid. Electrical components are located on the back of exoskeleton as shown in Figure 6.

graphic

Hip Actuation

The hip actuator is a Parker BE231D brushless DC motor mounted at the lower back of the exoskeleton. The motor is connected to an 80 to 1 transmission linkage system that moves motor torque to the hip. This system can provide 53 Nm of continuous and about 160 Nm of maximum intermittent hip torque.

Coupling Mechanism

The coupling mechanism (Figure 7) consists of a pulley at the hip (hip pulley) that is 4 inches in diameter and a smaller pulley, 2 inches in diameter, at the knee (knee pulley). The hip pulley freely rotates coaxially to the hip joint. The knee pulley is fastened to the tibia link of the exoskeleton, generating knee rotation from the knee pulley. A pair of wire ropes connects the two pulleys: rotating the hip pulley rotates the knee at twice the angular velocity.Coupled motion of the exoskeleton leg is generated when a mechanical brake fixes the hip pulley relative to the torso of the exoskeleton. In this state, the hip pulley becomes analogous to the pelvis where the hamstring and rectus femoris are attached. The pair of wire ropes connecting the pulleys now spans across both the hip and knee joint—in effect acting as two very stiff antagonistic biarticular muscles. Just like their biological counterparts, these wire ropes become able to generate knee flexion during hip flexion and knee extension during hip extension. The coupling ratio is determined by the relative diameter of the pulleys. The two states of the coupling system are used to produce a walking cycle (Figure 8).

graphic

graphic

Knee Design

The exoskeleton knee plays a critical role in gait generation since the locking and unlocking behavior of the knee needs to work in concert with the hip coupling mechanism. The knee has three basic functions. First, it automatically locks against flexion at heel strike. Secondly, it smoothly unlocks when coupled to the hip mechanism during swing-flexion phase, and thirdly, it is free to extend during swing-extension phase.

By default, the knee in this system is free to extend and locked against flexion. This behavior is achieved by using an off-the-shelf one-way locking gas spring controlled by a cam coupled to the knee pulley. The lockable gas spring constant can be tuned to provide knee extension torque for individuals with knee joint spasticity and contractures.

Determining Coupling Gait Trajectory

As previously described, the walking gait of the exoskeleton is generated through the activation and deactivation of the hip-knee coupling system, in other words, through engaging and disengaging the hip pulley. The synchronization of hip pulley engagement as a function of hip angle defines the gait trajectory. Clinical gait analysis (CGA) data published by Winter was used as reference to determine the optimal walking trajectory 1111.

By superimposing hip joint data on top of knee joint data, the coupling gait trajectory can be determined. This reveals regions of stance, swing-flexion, and swing-extension. The vertical dashed lines in Figure 9 mark the point of maximum hip extension and maximum knee flexion. The segment between the dotted lines is the swing-flexion phase of the gait cycle when the hip and the knee are both flexing. Based on this data, the swing-flexion phase of the gait begins when the hip is extended to approximately -17̊ and ends after approximately 30̊ of hip flexion, when the hip angle reaches 13̊.

graphic

At maximum knee flexion, the coupling system can be deactivated, allowing the knee to freely re-extend with the assistance of gravitational force and a knee extension spring. As shown in the swing-extension segment of Fig. 9, knee angle returns to zero while the hip joint can be controlled to follow an independent and arbitrary trajectory. During stance phase, the hip is driven through a desired trajectory while the knee remains uncoupled (knee angle remains constant).

Discussion of Experimental Data

Preliminary experiments using hip-knee coupling for ambulation assistance were conducted with the Austin exoskeleton on a 21 year-old male test pilot, who weighed 200 lbs at six foot two, and who had sustained a complete T12 injury 3 years prior. Video point tracking data (Figure 10) of the subject walking with the exoskeleton was collected for comparison with published CGA data.

graphic

Figure 11 plots point tracking data of two steps, with markers indicating the moment of “heel-off” for each step. The plot provides a comparison between CGA data and the full gait cycle of the Austin system. As expected, the data shows that the paraplegic pilot spends more time in double-stance getting ready for the next step. However, if the moments of heel-off are superimposed onto one another, the data shows that the mechanical gait generator produces a swing phase gait trajectory very similar to the CGA reference.

graphic

Conclusion and Future Work

The Austin exoskeleton illustrated the feasibility of using a mechanical gait generator to mimic a natural human gait. Eliminating knee actuation by embedding knee control into intelligent hardware design, allows the Austin system to become more compact and lightweight than some existing powered exoskeletons.

1.
NSCISC, 2012, “Spinal Cord Injury Facts and Figures at a Glance,” 35(4) https://www.nscisc.uab.edu.
2.
Brown-Triolo
,
D.L.
,
M. J.
Roach
,
K.
Nelson
, and
R.J.
Triolo
,
2002
,
“Consumer perspectives on mobility: implications for neuroprosthesis design.,”
J. Rehabil. R&D
,
39
(
6
), p.
659
–p.
669
3.
Phillips
,
L.
,
M.
Ozer
,
P.
Axelson
, and
J.
Fonseca
,
1987
, Spinal Cord Injury: A Guide for Patient and Family, Raven Press.
4.
Center for Orthotics Design, 2013, “Isocentric RGO,” http://www.centerfororthoticsdesign.com/isocentric_rgo/index.html.
5.
Sykes
,
L.
,
J.
Edwards
,
E.S.
Powell
, and
E.R.
Ross
,
1995
,
“The reciprocating gait orthosis: long-term usage patterns.,”
Arch. Phys. Med. Rehabil.
,
76
(
8
), pp.
779
–pp.
783
6.
Little
,
R.
, and
R.
Irving
,
2011
, “Self contained powered exoskeleton walker for a disabled user,” U.S. Pat. App. 12/801,809.
7.
Doorenbosch
,
C.A.
,
J.
Harlaar
,
M.E.
Roebroeck
, and
G.J.
Lankhorst
,
1994
,
“Two strategies of transferring from sit-to-stand; the activation of monoarticular and biarticular muscles,”
J. Biomech.
,
27
(
11
), pp.
1299
–pp.
1307
8.
Kumamoto
,
M.
,
T.
Oshima
, and
T.
Yamamoto
,
1994
,
“Control properties induced by the existence of antagonistic pairs of bi-articular muscles—Mechanical engineering model analyses,”
Hum. Mov. Sci.
,
13
, pp.
611
634
.
9.
Schenau
,
G.
van
I.
,
1989
,
“From rotation to translation: constraints on multi-joint movements and the unique action of bi-articular muscles,”
Hum. Mov. Sci.
,
8
, pp.
301
337
.
10.
Sutherland
,
D. H.
,
K.R.
Kaufman
, and
J.R.
Moitoza
,
1994
,
“Kinematics of normal human walking,” Human Walking
,
Williams & Wilkins
, pp.
23
44
.
11.
Winter
,
D. A.
,
1990
,
Biomechanics and Motor Control of Human Movement
, 2nd ed.,
John Wiley & Sons
.