CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 1
Robotic Assistive Devices to Improve Quality of Life for Persons with Amputation and Paraplegia
Clare Hartigan, PT, MPTLower Extremity Robotics Project Manager
Clinical Research CoordinatorShepherd Center
Atlanta GA
Michael Goldfarb, PhDH. Fort Flowers Professor of Mechanical Engineering
Professor of Electrical EngineeringProfessor of Physical Medicine and Rehabilitation
Vanderbilt UniversityNashville TN
2013 CPTA Annual ConferenceAugust 15, 2013
Disclosures
• Michael Goldfarb: Inventor on patents owned by Vanderbilt University that relate to the assistive devices described here, which have been licensed to various companies for commercial translation.
• Clare Hartigan: Consultant for Parker‐Hannifin on the clinical aspects of the lower limb exoskeleton.
• Research funded by:• NIH R01HD059832• NIH R21HD068753• NIH R01EB005684
Objectives
3
• Present and discuss the status of emerging robotic leg prostheses intended to enhance mobility for lower limb amputees.
• Present and discuss the status of emerging robotic multigrasp hand prostheses intended to enhance dexterity for upper extremity amputees.
• Present and discuss the status of emerging lower limb robotic exoskeleton technology intended to provide legged mobility and/or locomotor training to persons with SCI, CVA, and MS.
Modern Commercial Leg Prostheses
4
• Typical above‐knee prosthesis consists of a damper at the knee joint and relatively stiff leaf spring for the ankle/foot complex.
• These prostheses are energetically passive devices (i.e., they cannot contribute net power to gait).
• These prostheses provide a relatively small subset of the functionality of the intact limb.
• Amputees walk more slowly, use more energy, stress intact joint, limited mobility, fall frequently
• Recent advances in robotics technology enable a fully powered leg capable of biomechanical levels of torque and power within the size and weight constraints of a lower limb prosthesis.
• Such devices offer the potential to provide a much greater level of functionality to the amputee.
Suction socket
Knee is damper
Ankle/foot complex is leaf spring
Robotic Artificial Leg
5
• Actuation: Two 200W rare‐Earth‐magnet brushless DC motors with ~190:1 transmissions.
• Power: Lithium polymer battery.
• Sensors: Prosthesis configuration and shank load.
• Intelligence: Two on‐board microcontrollers.
• Total mass of leg: 4.2 kg* (9.3 lb).
Generation TwoVanderbilt Prosthesis
*Corresponds to intact limb mass of 105 lb person.
Control of a Robotic Leg
6
• Passive prostheses can only react.
• A powered prosthesis can both act as well as react.
• Prosthesis requires a control interface that provides reliable, robust, natural, and direct control over the movement of the prosthesis.
• Controller overview:• Microcontroller sits at the interface between the user and prosthesis and governs movement of leg.
• Microcontroller constantly observes patterns in sensors information.
• Based on patterns, the leg decides what to do.• User has to be actively engaged with the leg to make it work. If the user stops moving, leg becomes passive (stops moving).
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 2
Level Walking
7
Comparison with Healthy Biomechanics
8
Pas
sive
P
rost
hesi
s
Benefits of Powered Push‐off
9
• Passive prosthesis• Little if any forward propulsion provided by prosthesis.• Hip on the prosthesis side sources all power for swing phase.• Often results in underpowered swing phase with little toe clearance.• Increases likelihood of scuffing and/or stumbling.• Often results in heel hiking, particularly up slopes, uneven terrain.
• Prosthesis with powered push‐off• Powered push‐off from prosthesis propels amputee forward.
• Reduces metabolic energy consumption.• Powered push‐off drives swing leg forward.
• Enhances swing knee flexion and toe clearance.• Decreases likelihood of scuffing or stumbling.• Eliminates tendency for heel hiking.
• Swing phase load on hip is dramatically decreased.
Biomechanical Benefits of Power
10
• Self‐selected speed of level walking:• Passive prosthesis: 4.1 km/hr @ 90 steps/min• Powered prosthesis: 5.1 km/hr @ 90 steps/min• Subjects walk 24% faster with powered prosthesis
• Metabolic energy consumption:• Measurements taken on treadmill @ self‐selected speed for passive prosthesis (3.2 km/hr)
• Oxygen uptake was 23.2% greater with passive prosthesis • If metabolic baseline is subtracted, oxygen uptake was 38.7% greater with passive prosthesis
Functional Flexibility
11
• Previous results for level walking.
• People typically traverse a variety of terrain types (up/down slopes, up/down stairs) and engage in a variety of activities.
• Passive prostheses are particularly limited in their ability to provide appropriate biomechanics across varying terrain types and during a variety of activities.
• Powered prostheses can emulate the behavior of the healthy limb, and therefore are much better able to provide healthy biomechanics across terrain and activity types.
Biomechanics of Upslope Walking
12
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 3
Benefits in Upslope Walking
13
Slope Walking
14
Stairs
15
Running
16
Falls in Lower Limb Amputees
17
• The annual incidence of falls in the lower limb amputee population exceeds that of the elderly population.
• The rate of seeking medical attention as a result of such falling is comparable to that of the institution‐living elderly.
• The incidence of falling (and requiring medical attention due to such falls) is higher in younger than in older amputees.
• In a survey of 435 lower limb amputees, Miller et al. (2001) conclude that “falling and fear of falling are pervasive among amputees.”
• In a survey of 396 lower limb amputees, Gauthier‐Gagnon et al. (1999) report that 50% of respondents reported that they had to “think about every step they made.”
Ground Slope Adaptation
18
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 4
Providing Active Recovery Responses
19
Commercial State‐of‐the‐Art in Hand Prostheses
20
shoulder harness pulls cable to open
hook
body-powered prosthesis
electrodes on skin surface measure
muscle contraction in residual limb and open/close “hand” via electric motor
myoelectric prosthesis
Both are single degree-of-freedom devices (open/close only)
Grasps and Postures in ADLs
21
Vanderbilt Multi‐Grasp Hand
22
VMG Hand
VMG Hand Postures/Grasps
23
VMG Hand Design: Actuation
24
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 5
VMG Hand Video
25
Control of a Multigrasp Hand Prosthesis
26
• Trade‐off exists between functionality and cognitive effort.
• Single DOF myoelectric hand provides intuitive, real‐time, robust, reliable, proportional control.
• User must be able to access multifunctional capability of hand in a natural and efficient manner.
• Multigrasp control interface should provide intuitive, real‐time, robust, reliable, proportional control.
Multigrasp Myoelectric Control Structure
27
MMC Demonstration
28
Control Demonstration
29
Functional Assessment
30
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 6
Preliminary Assessment Results
31
Functionality Profile VMG
Whole Hand Grasps
Extension 89
Spherical 87
Power 85
Precision Grasps
Lateral 88
Tripod 71
Tip 59
Index of Function 87
Indego® Exoskeleton
32
• Provides active movement assistance at both hip and knee joints.
• Allard AFOs provide ankle stability.
• Total weight is 12 kg (27 lbs).
• Modular design (3 pieces) snaps together to facilitate self‐donning/doffing, transport, storage, and handling.
• FES option provides up to 10 channels.
• Compact frontal profile enables sitting in standard wheelchair, car etc.
Allard ToeOFF ®
Collaborative Effort
33
• Indego® is the outcome of a collaborative research effort between Shepherd Center and Vanderbilt University, funded by NIH
• Aug 2010 Vanderbilt – Shepherd Enrolled 1st Subject
• Oct. 2012 Parker Hannifin Licensed Exoskeleton Technology from Vanderbilt Univ.
Primary Indications
34
• Non-ambulatory individuals: Full-assist mode provides legged mobility and the associated health benefits.
• Poorly-ambulatory individuals: Partial-assist mode enables overground locomotor training for neural re-education and functional recovery.
Full-Assist Mode Partial-Assist Mode
Indications and Outcomes
35
Partial‐assist Indego (Lite)
Partial‐assist with supplemental FES
Hypothesized outcomes: Improved gait speed, symmetry, stride length, step height, balance
Planned in future work
Lower limb hemiparesis following stroke
AIS C and D SCI, MS, TBI
Full‐assist Indego
Responsive to FES
Lower motor neuron damage
Hypothesized outcomes:•Secondary health benefits •Improved well‐being •Home/community ambulation
Full‐assist Indego with supplemental FES
Hypothesized outcomes: Enhanced secondary health benefits, relative to without supplemental FES
AIS B or C
AIS B or C
Hypothesized outcome: Possible neurological or functional improvement
AIS A
AIS A
Lower Limb Neurological Deficit
Non‐ambulatory SCI:AIS A, B, or C
Poorly‐ambulatory individuals
Planned in future work
Limited capacity for functional
improvement
Capacity for functional
improvement
Full‐Assist Control Mode
36
• Objective: Provide legged mobility to otherwise non‐ambulatory individuals (thoracic‐level AIS A/ B/weak C).
• Intended primarily as an assistive device.
• Hypothesized benefits:
• Home/limited community ambulation
• Enhanced well‐being and quality of life
• Secondary health benefits of walking, including:
• decrease in levels of pain and spasticity
• improvements in bowel and bladder function
• decrease in fat tissue mass/weight loss
• increase in bone mineral density
• improvement in circulation
• improvement in skin health.
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 7
Posture Based Control
37
“Lean forward, Walk forward”
•Provides user with full control of movement.
•Can be used with standard assistive device (no connection w/external controls).
•Ensures safe movement (system initiates step only when user is in appropriate position).
•Promotes good technique (vibrational feedback encourages balance learning)
10MWT in Full‐Assist Mode
38
T10 AIS A subject
TUG Test in Full‐Assist Mode
39
T10 AIS A subject
Preliminary Case Study
40
• T10 AIS A subject, 39 years old, 9 years post injury• TUG, 10MWT, and 6MWT with exoskeleton and KAFOs • Walker as stability aid for both• Mobility measured by average speed of each instrument• Exertion measured by physiological cost index (PCI)
• Exoskeleton provided 3.5x faster walking during 10MWT• Exoskeleton required 10x less exertion during 10MWT
Preliminary Mobility Study
41
• 5 AIS A subjects, levels of injury T1‐T12 Motor Complete• TUG, 10MWT, and 6MWT with exoskeleton• Either forearm crutches or walker as stability aid• Mobility measured by average speed of each instrument/test
• All subjects demonstrated similar speeds for all assessment instruments, regardless of device experience or stability aid used
• Average walking speed during 10 MWT was ~0.35 m/s
Full‐Assist Stair Ascent/Descent
42
T10 AIS A subject
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 8
Full‐Assist with Supplemental FES
43
• Indego includes up to 10 channels of exoskeleton‐controlled FES:
• Hamstrings • Quadriceps • Gastrocnemius• Tibialis anterior• Anterior trunk• Posterior trunk
• Hypothesized benefits: Enhanced secondary health benefits (relative to full‐assist without FES), including:
• decrease in levels of pain and spasticity• improvements in bowel and bladder function• decrease in fat tissue mass• increase in bone mineral density• improvement in skin health
Preliminary Studies with FES
44
• Hamstring stimulation used for hip extension during stance phase of walking.• Quadriceps stimulation used for knee extension during swing phase of walking.• Stimulation timing and levels automatically adjusted (on step‐by‐step basis) by the exoskeleton controller to provide as much assistive joint torque as possible.
• Joint motion and torque measured by exoskeleton during walking with and without FES.
Walking with Supplemental FES
45
Preliminary Results with FES
46
Hamstrings provide 25% of hip torque during stance
Quadriceps provide 95% of extensive knee torque during swing
Movement is essentially same, with or without FES
Indications and Outcomes
47
Partial‐assist Indego (Lite)
Partial‐assist with supplemental FES
Hypothesized outcomes: Improved gait speed, symmetry, stride length, step height, balance
Planned in future work
Lower limb hemiparesis following stroke
AIS C and D SCI, MS, TBI
Full‐assist Indego
Responsive to FES
Lower motor neuron damage
Hypothesized outcomes:•Secondary health benefits •Improved well‐being •Home/community ambulation
Full‐assist Indego with supplemental FES
Hypothesized outcomes: Enhanced secondary health benefits, relative to without supplemental FES
AIS B or C
AIS B or C
Hypothesized outcome: Possible neurological or functional improvement
AIS A
AIS A
Lower Limb Neurological Deficit
Non‐ambulatory SCI:AIS A, B, or C
Poorly‐ambulatory individuals
Planned in future work
Limited capacity for functional
improvement
Capacity for functional
improvement
Partial‐Assist Control Mode
48
• Objective: Provide overground locomotor training to poorly ambulatory individuals to facilitate faster and more effective recovery of balance and walking.
• Intended primarily as a therapeutic device.
• Hypothesized benefits include improved:• gait speed• balance• step symmetry• stride length• step height
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 9
Exoskeleton for Locomotor Training
49
• The case for an exoskeleton as an assistive device to provide legged mobility is obvious.
• Why use an exoskeleton for locomotor training?• combines coordinated multi‐joint assistance for
promotion of healthy gait patterns with the balance development, weight‐shifting, and whole‐body movement involved in overground walking
• provides body weight support from the ground up in a manner consistent with the biomechanics of overground walking
• enables the patient to dictate the spatiotemporal nature of lower limb movement
• Hypothesis: Exoskeleton‐based overground locomotor training will enhance functional recovery (or the rate of functional recovery) of gait for patients with lower limb hemiparesis
Full‐Assist versus Partial‐Assist
50
• Full‐assist controller relies on the user to trigger a given movement (e.g., a step), but once triggered the exoskeleton dictates the movement trajectories of each joint for the remainder of that movement.
• We hypothesize that (in cases when patients have the capacity for significant functional and/or neurological recovery), recovery is better facilitated when the patient rather than the exoskeleton is principally responsible for movement coordination (i.e., Hebbian learning).
• Therapeutic controller to facilitate recovery of strength and coordination should:
• follow rather than lead movement• assist movement as needed rather than assist fully• adapt assistance gradually as the patient requires less
• We call this a “partial‐assist” controller.
Preliminary Studies
51
• Implemented partial‐assist control on three subjects with hemiparesis from stroke.
• Evaluated single‐session effects over three separate sessions on fast gait speed (FGS), step length asymmetry (SLA), and stride length (SL).
• Each session consisted of:• 5‐min warm‐up without exoskeleton• 10MWT without exoskeleton• don exoskeleton• 20‐25 min walking with exoskeleton (in ~5‐min segments)• doff exoskeleton• 10MWT without exoskeleton
Subject Age (yrs)
Post-stroke (mos)
Affected Side
Stability Aids
Baseline FGS (m/s)
Baseline SLA (%)
Baseline SL (cm)
1 39 3 Right Quad cane, R AFO 0.33 29 88.7 2 42 10 Left Quad cane, L AFO 0.07 115 33.2 3 69 17 Right Quad cane, R AFO 0.19 27 66.3
Preliminary Studies
52
Preliminary Results
53
• data (across all subjects and all sessions) indicate average single‐session improvements of 26% in FGS, 26% in SLA, and 30% in SL
• recent studies incorporating BWSTT for persons with chronic stroke resulted in significant improvements in FGS, but not in SLA
• recent studies in unilateral and split‐belt treadmill training for persons with chronic stroke resulted in significant improvements in SLA, but not in FGS
• results here are preliminary, but indicate that exoskeleton training may provide simultaneous improvements in both FGS and SLA.
Preliminary Results: Persistence
54
• single-session improvements in all gait characteristics persisted 24 hours after training in all of the training sessions
• note that negligible improvement in stride length following Visit 3 is because stride length was representative of healthy walking (for a healthy adult female corresponding to the measured gait speed)
CPTA Annual Conference September 20, 2013
Michael Goldfarb and Clare Hartigan 10
Summary and Future Work
55
Partial‐assist Indego (Lite)
Partial‐assist with supplemental FES
Hypothesized outcomes: Improved gait speed, symmetry, stride length, step height, balance
Planned in future work
Lower limb hemiparesis following stroke
AIS C and D SCI, MS, TBI
Full‐assist Indego
Responsive to FES
Lower motor neuron damage
Hypothesized outcomes:•Secondary health benefits •Improved well‐being •Home/community ambulation
Full‐assist Indego with supplemental FES
Hypothesized outcomes: Enhanced secondary health benefits, relative to without supplemental FES
AIS B or C
AIS B or C
Hypothesized outcome: Possible neurological or functional improvement
AIS A
AIS A
Lower Limb Neurological Deficit
Non‐ambulatory SCI:AIS A, B, or C
Poorly‐ambulatory individuals
Planned in future work
Limited capacity for functional
improvement
Capacity for functional
improvement
Credits
56
Kevin HaMD/PhD candidate
Vanderbilt
Spencer MurrayPhD candidateVanderbilt
Don TruexResearch engineer
Vanderbilt
Ryan Farris, PhDVanderbilt
Hugo Quintero, PhDVanderbilt
Scott HawesPhysical TherapistBeyond Therapy TN
Ismari Clesson, RNDirector of ResearchShepherd Center
Donald P. Leslie, MDMedical DirectorShepherd Center
David F. Apple, MDMedical Director Emeritus
Shepherd Center
Jason MitchellPhD candidateVanderbilt
Brian LawsonPhD candidateVanderbilt
Daniel BennettPhD CandidateVanderbilt
Amanda ShultzPhD CandidateVanderbilt
Skyler Dalley, PhDVanderbilt