1. Test Bed 6- Human Assist Devices (Fluid-powered ankle-foot-orthoses)
Faculty
Students
Liz Hsiao-Wecksler (UIUC)
Will Durfee (UM)
Geza Kogler (GT)
Zongliang Jiang (NCAT)
Doug Cook (MSOE)
Vito Gervasi (MSOE)
Tom Chase (UM)
David Kittelson (UM)
Eric Barth (Vanderbilt)
Yifan David Li(UIUC)
Morgan Boes (UIUC)
Mazhar Islam (UIUC)
Jicheng Xia (UM)
Nebiyu Fikru (UM)
Lei Tian (UM)
Mark Hofacker (Vanderbilt)
Dan Cramer (Vanderbilt)
Davorin Stajsic (NCAT)

2. Overview Motivation Test Bed 6 Timeline Progress on Pneumatic AFO
Progress on Hydraulic AFO
TB6 Affiliated Projects
Future Work

3. Major question being answered
Do energy-to-weight and power-to-weight advantages of fluid power (FP) continue to hold for tiny, mobile FP systems ( W)?
Drive development of enabling FP technologies
Create new portable, wearable, FP assist devices
TB6 Product Platform: Ankle-Foot Orthosis
Numerous pathologies / injuries create below the knee muscle weakness and impair gait
Currently no portable powered ankle-foot orthosis avalible for treatment
Stroke (4.7M*)
Polio (1M*)
Multiple sclerosis (400K*)
Spinal cord injuries (200K*)
Cerebral palsy (100K*)
Trauma
* Number of people in US that would benefit from an active lower limb orthosis [Dollar and Herr, IEEE Trans Robotics, 24(1): , 2008]

4. Potential applications
Rehabilitation of lower leg muscles (possible from the patient’s personal residence instead of restricted to clinic)
Assistance in walking (including flexibility to walk outside)
General use for public commercial applications where large amounts of walking are required

5. TB6 Development Timeline
Multiple Design Versions
Started with motion control and progressed to powered actuation
ca 2008
Custom integrated (Untethered)
CCEFP affil. proj.
psi
6-20 Nm
Light-weight
Safer
Powered pneumatic AFO for both motion control and assistance (Tethered)
Pneumatic AFO
Off-the-shelf components
ca 2012
ca 2007
ca 2010
IMU based mode recognition
Functional energy efficiency analysis
Passive
pneumatic
power- harvesting AFO for motion control (Untethered)
Portable Powered pneumatic AFO for both motion control and assistance (Untethered)
IMU
Hydraulic AFO 5

6. Setup for Pneumatic AFO
Structure:
Shell, lines, integration
Power: Engine or CO2
Actuation: Valves, Regulators, Actuator
Electronics: Sensors and driving electronics
The world’s lightest, most compact, untethered, pneumatically powered AFO

7. Inertial Sensor Based Gait Mode Recognition and Actuation Control
Goal: recognize different gait modes (stairs and ramps) and control the actuation accordingly
Level: Dorsiflexion, create clearance for limb advancement
Stair Descent: Plantarflexion, prepare for next landing
Different Functional Need for Stair Descent
Actuation for LEVEL GROUND
Actuation for Stair Descent
Mode recognized by IMU
Plantarflexor (0-20GC and GC)
IMU

8. Inertial Sensor Based Gait Mode Recognition and Actuation Control
Stair and ramp modes can be recognized by tracking the vertical position differences and foot angle for each step
Experimental Protocol
Three gait mode conditions:
Outdoor stairs, one step traverse
Outdoor stairs, two step traverse
Indoor stairs, two step traverse
Level ground was assessed during approach for ascent and descent the stairs
Three PPAFO actuation algorithms:
Passive: no actuation provided
Mode Controller: level controller except stair descent (plantarflexor torque during swing)
Level Controller: actuation provided using original level ground walking mode controller for all gait modes
Only (c) tested for gait mode conditions 2-4.
Subject
Indoor
Two Stairs
Mode Ctrl
Outdoor
One Stair 1 98
92.5
98.1 2
92.1
71.5
78.2 3
90.5
88.9
96.4 4
96.7
97.2
95.5 5
92.8
92.9
91.5
Avg
94.0
88.6
Number of Subjects: 5 (male) Weight: average 82.0kg (70-97kg)
Age: average 23.4 (20-27) Height: average 178.6cm ( cm)

9. Functional Efficiency Analysis
Goal: Compare fuel consumption and work output between different control algorithms
CO2 Bottle
Wireless Data Logging
Micro
Controller
Computer
Data Logging Receiver
Force, Angle, Pressure
Control Mode Condition
Direct Event
State Estimation – no recycling
State Estimation – with recycling
Procedure Notes:
Four healthy subjects (22-30 y.o.)
3 minute trials of walking, ~150Hz sampling
CO2 bottles were weighed before and after each trial for fuel consumption
Each bottle was warmed to room temperature in a water bath before use
4 bottles were rotated for use to allow warming

10. Control Mode Condition Average Fuel consumption (per 3 min trial)
Results on fuel consumption and net work output for different controllers
Control Mode Condition
Average Fuel consumption (per 3 min trial)
Average Net
Work output
(per step)
Direct Event (DE)
57.5 g
4.7 J
State Estimation – no recycling (SE)
61.0 g
3.5 J
State Estimation – with recycling (SER)
50.5 g
3.3J
Results
SE took on average 6% more fuel than DE.
SER saves 17.5% of fuel compared to the fuel consumption of SE.
The DE scheme did more work than SE or SER.
SE and SER did approximately the same amount of work per gait cycle

11. Hydraulic Ankle Foot Orthosis: First Platform
Torque: 90 N-m
(600 lbs force for a 3cm moment arm)
Small packaging space
Weight: < 1 kg
Portability: untethered
Longevity: 10,000 steps
Peak power: 250 W
Artist rendering

12. Many System Level Questions Need to Be Answered Before Specifying Each Component
Electric Motor
Battery
Pump
Conduits
Actuators
Gearhead? or
Not?
Piston? or
Vane?
Long? or
Short?
Linear? or
Rotary?

13. The Efficiency and Weight of Each Component Can Be Modeled Analytically
Electric DC Motor Weight
Gear-head Efficiency
Hydraulic Cylinder Weight
Axial-Piston Pump Efficiency

14. The Established Efficiency and Weight Models Can Identify the AFO Configuration
Power Pack

15. Design Variables Can Be Expressed by Known Parameters*
Pump displacement (cc/rev):
Num. of pistons
Pump piston area (m^2):
Pitch radius
Swash-plate angle
Pump piston bore (m):
Pump efficiency:
* Key pump design variable: pump displacement, pump piston bore and pump efficiency. To simplify the design problem, other pump parameters were adopted from the three-piston Oildyne pump.

16. Design Variables Can Be Plotted on P-n Plot

17. The Hydraulic AFO Components Can Be Specified Based on the Design Map
Part Number
Wt (g)
Dia (mm)
Len (mm)
Actuator Package
PF Cylinder
Custom part
165
17.2
100
DF Cylinder 23 7
Joint Pulley
TBD 60 NA
Total
188 72
130
Power Source
Valves
Pump
190 32 53
Gear Head
Maxon
118 27
DC Motor
Maxon
141 45
449
107
Energy Source
Battery
TP2700-6SPL25
465
50 x 34 x 102

18. Conclusions
System level analyses are necessary to identify the design guidelines for the hydraulic AFO.
The analytical efficiency and weight models for the system level analyses are achievable.
For the hydraulic AFO, the actuators would better be separated from the power source, similar to an excavator.
The analytical efficiency and weight models are also needed to specify each component in the hydraulic AFO system.

19. Affiliated Projects: 2F MEMS Valve (Nebiyu Fikru, UMN)
Goal : Create an efficient MEMS based proportional valve for controlling air flow in pneumatic systems
Targets
High flow rate (40 slpm at 6 → 5 bar)
Compact (< 4 cm3)
Low power usage (< 1 mW)
Progress
Port plate with array of orifices successfully fabricated and tested for flow and pressure
Displacement sensor integrated to meso-scale prototype valve
Fabrication of MEMS unimorph actuator nearing completion
Next Steps
Complete and test MEMS unimorph actuator
Demonstrate proportional control on meso-scale valve
Fabricate MEMS bimorph PZT actuator

20. Affiliated Projects
Clinician Centered AFO Interface (Davorin Stajsic, NCAT)
Spring 2013 plans
NCAT will continue to work on Quanser and XNA integration, and further game and GUI development. Currently there are some XNA and Quanser API incompatibilies that need to be resolved.
Plan on doing experimental research using CybexNorm system (shown) to emulate a game therapy session in order to test the effects of different factors that may have an impact on game performance (social interaction among patients, leg dexterity, different seating positions, etc.)
This experiment will help with further development of the game with features that ensure patient improvement, and a game therapy that is effective on a more diverse patient population.

21. Affiliated Projects – 2D MSOE
Passive HCCI thermal-management
Successful testing w/ surrogate source
Currently not funded by CCEFP
Developing high-efficiency pneumatic actuation system
>60% ηthermodynamic using <15g fuel/hr
CCEFP rejected 2011 proposal
Proposal submitted to NSF National Robotics Institute
Co-robotics applications – legged robots, assisting humans

22. Affiliated Projects
Elastomeric Accumulator (Dan Cramer, Vanderbilt)
Goal : Create a gravimetrically and volumetrically efficient strain accumulator for pneumatic systems
Targets
Pressures 7-10 bar
Max volume of 34 ml
Progress
Accumulator operating below 7 bar fitted
Next Steps
Fabricate accumulator for use with pressures up to 150 psi
Evaluate gravimetric and volumetric efficiencies

23. Affiliated Projects: 2B.4 Controlled Stirling Thermocompressor
(Eric Barth, Vanderbilt)
Goal : Create a compact, near silent, pneumatic power source with low amounts of vibration
Targets
20 Watts
80 psig
Mounted on ankle-foot orthosis
Progress
Completed dynamic model of thermocompressor
Constructed single stage prototype
Novel take on Stirling cycle device
Piston controlled directly by brushless DC motor and reciprocating lead screw
Developed method of achieving high rates of heat transfer in compact device
In process of patenting
High Temperature/High Efficiency
Enabled by use of fused quartz and machinable ceramic
Next Steps
Instrument and test prototype
Refine model
Build multistage compressor capable of high pressure
Stainless Steel Heater Head
Fused Quartz Cylinder
Reciprocating Lead Screw
Macor Machinable Ceramic
Pictured above are pictures, schematics, finite element analysis, schematic of operation, and predicted performance of the Stirling Thermocompressor for an ankle-foot orthosis.  It is meant to be a 20 Watt pneumatic power supply air at 80 psi. Meant to be mounted on the back of the user’s calf, the device must be compact, light weight, near silent, and have low amounts of vibration while still having sufficient power to provide walking assistance for over one hour. These metrics are achieved through the novel use of a reciprocating lead screw, brushless dc motor, and heat exchangers that efficiently convert large amounts of thermal energy into pneumatic potential while using only small amounts of electrical power.
Pressure Transducer
DC Motor