Secondary Logo

Journal Logo

Original Research Article

A Powered Inverting and Everting Prosthetic Foot for Balance Assistance in Lower Limb Amputees

Panzenbeck, Jason T. MS; Klute, Glenn K. PhD

Author Information
JPO Journal of Prosthetics and Orthotics: October 2012 - Volume 24 - Issue 4 - p 175-180
doi: 10.1097/JPO.0b013e31826e86c3
  • Free

The loss of all or part of a lower limb can reduce the ability to balance while standing or walking. Among amputees able to walk in household and community environments, 65% report low balance confidence1 and more than half have fallen within the last 12 months.2 Of those who have a fear of falling, three quarters indicated that they avoided certain activities in light of their fear.1 Low balance confidence or fear of falling often heralds the beginning of a downward spiral of decreasing mobility and participation in social activities.3 Decreased mobility can lead to a deterioration of muscle strength, endurance, and coordination,3–7 which may, in turn, lead to a further deterioration of balance.8

Balance is one of several considerations in optimizing mobility with a prosthesis. One balance strategy involves ankle inversion and eversion to adjust the individual’s center of pressure (COP).9–12 Because amputees lack ankle musculature, they often exhibit greater COP movement (sway) while standing than nonamputees do.13 While walking, balance can be influenced by foot placement at the end of swing10,14,15 and by ankle musculature contributions during stance.10,11,16 Once foot placement is established, inversion and eversion moments can act to correct for small errors in foot placement. Unfortunately for amputees, their variability in foot placement is greater than that for nonamputees17 and the ankle inversion eversion strategy is unavailable to them.

The purpose of this article was to describe our progress in developing a powered inverting and everting prosthetic foot to replace the lost musculature of the amputated limb and discover if such a device can improve amputee balance. A prototype ankle mechanism capable of inversion and eversion was designed and built. A robotic bench test was performed to measure system response time. A human subject test with a transfemoral amputee was performed to measure the effect of the device on standing and walking balance.



The prototype prosthesis (Figure 1) has five components: a four-bar mechanism, a linear actuator, a load cell, the keel of a conventional prosthetic foot, and a control system. The 1.4-kg four-bar mechanism was designed for high torque-force transmission ratio and a full 22-degree inversion to 6-degree eversion range of motion.18 The 0.7-kg linear actuator system can output nearly 400 N of force. It consists of a motor (model EC-Powermax; Maxon, Sachseln, Switzerland) operated with a Maxon (model EPOS 70/10) motor controller. The motor rotary motion was transformed to linear motion with a spindle transmission (Ultra Motion, Cutchogue, NY, USA). When connected to the four-bar mechanism, the output varied from 23 to 34 N-m about the ankle joint center depending on the angular position. A 0.4-kg three-axis load cell (Futek Advanced Sensor Technology, Irvine, CA, USA), mounted proximal to the four-bar mechanism, allowed measurement of the pylon-axis force, the medial-lateral force (orthogonal to the pylon), and the coronal moment. The distal component was a 0.4-kg modified Seattle Carbon LightFoot prosthetic foot (TruLife, Poulsbo, WA, USA). This foot (model SCF18526LN3) is intended for medium- to high-activity individuals (i.e., able to participate in light sports) with a weight range from 66 to 106 kg. The proximal elements were removed from the foot to provide access to the keel upon which the four-bar mechanism was fastened. The control system operated at 100 Hz using LabView (National Instruments, Austin, TX, USA) in a Windows operating environment. The control system communicated with the motor controller over a Controller Area Network bus and with the load cell through a 100-Hz analog antialias filter and a data acquisition card (National Instruments).

Figure 1
Figure 1:
Powered inverting and everting prosthetic foot. The system includes a four-bar mechanism, a linear actuator, load cell, a control system, and the keel of a conventional prosthetic foot.

The objective of the control system is to assist in maintaining balance by inverting or everting the foot (Figure 2). This approach is similar to that used in biped robots to correct for small errors in foot placement.19–21 The controller determines the zero moment point (ZMP), the location under the plantar surface of the foot where the resultant ground reaction force acts, using data sampled from the pylon-mounted load cell. For assisting with postural stability, the controller then calculates the divergence of the ZMP from a ZMP setpoint acquired from an a priori standing trial. The divergence provides feedback through a proportional-integral controller whose gains were determined using the Zeigler-Nichols tuning method.22 The output drives ankle movement to minimize the feedback error. When the ZMP moves medially from the setpoint, the control system gives an inversion command. When the ZMP moves laterally from the setpoint, the control system gives an eversion command. For assisting with gait stability, the divergence of the ZMP from a ZMP setpoint trajectory, acquired from an a priori gait trial, is used as the feedback signal.

Figure 2
Figure 2:
Control loop schematic. The proportional-integral (PI) feedback loop uses the difference between the zero moment point (ZMP) calculated at 100 Hz from the pylon-mounted load cell and the setpoint determined during quiet stance to alter ankle position.


To measure the response time of the system, the prosthesis was loaded using a 6–degree of freedom robot capable of mechanically simulating human gait (Figure 3). The prosthesis was neutrally aligned in a horizontal position as if the person was supine and rigidly attached to the robot frame. The robot (Parallel Robotics System Corporation, NH, USA) had a force plate (Kistler, Amherst, NY, USA) mounted vertically to simulate the ground. This arrangement maximizes use of the robot workspace such that the entire stance phase of gait, from heel contact to toe-off, can be simulated. For this test, the proximal end of the prosthesis was held in a fixed position while the robot moved the force plate to provide the desired test conditions. The prosthesis was loaded to 600 N and the ZMP setpoint was defined during the equivalent of quiet stance. The robot was then instructed to perform a 2-degree rotation in the coronal plane of the prosthesis. The time for the prosthesis to invert or evert and reposition the ZMP back to within ±2 mm of its setpoint was measured.

Figure 3
Figure 3:
Robotic bench tests. Measurement of system response time was performed with a force plate mounted on a robot. The stationary prosthesis was loaded by superior translation of the force plate and then rotated in the coronal plate of the prosthesis.


To measure the effect of the powered inverting and everting prosthesis on standing and walking balance, unilateral amputees were recruited to participate in this institutional review board (IRB)–approved study. Individuals were eligible if they were between 18 and 70 years of age, able to wear their prosthesis for at least 6 hrs each day, and could walk without assistance. Individuals were excluded if they had a disorder, pain, or injury that interfered with their gait. The protocol included a test of standing balance and a test in which their gait was disturbed. For these tests, the participant’s pylon and prosthetic foot were detached and an appropriately sized pylon and the prototype prosthesis were attached and aligned by a certified prosthetist.


This test involved standing with the feet together on a force plate (Kistler) for 30 seconds. The test was first performed with eyes open and then with eyes closed. The control system of the powered prosthesis was turned on and off in random order while the position of the COP was recorded from the force plate using a data acquisition system sampling at 1200 Hz (Vicon, Centennial, CO, USA). The data used for analysis were a 20-second sample, which did not include the initial or final 5 seconds of the full 30-second trial to insure steady-state conditions.

The standard deviation of the COP position was hypothesized to be smaller in the medial-lateral direction with the control system turned on for both eyes open and eyes closed conditions. If the prosthesis did not function as desired, the standard deviation of the COP position would not be smaller.


This test involved walking on a motorized treadmill while the participant was occasionally subjected to medially and laterally directed disturbances.23 The subject wore a harness, which was belayed from an overhead position in the event of a fall (none occurred). For random gait cycles, a burst of air was used to disturb their prosthetic foot placement during the swing phase of gait. A flexible hose and small elbow joint were attached to either the lateral or medial side of the subject’s prosthesis. The hose was connected to a ballast tank of compressed air controlled by a solenoid valve which was placed adjacent to the treadmill. The timing between successive foot-ground contact, triggered by a foot switch taped to the bottom of the participant’s shoe, was calculated with Labview software and was used to determine when to open the solenoid valve such that the burst of air would occur ∼135 milliseconds before initial foot-ground contact. The burst duration was calculated using an estimation of leg segment weight24 to produce a medially or laterally directed foot placement shift between 25 and 50 mm.

The subject walked continuously on the instrumented treadmill at self-selected speed while 17 trials were collected, including eight medially directed disturbance trials, eight laterally directed disturbance trials, and one trial with no disturbance. Data were collected with a 12-camera motion analysis system (Vicon) at 250 Hz, and force plate data were collected at 1200 Hz. During a trial, the disturbed step was defined as the first foot-ground contact of the ipsilateral limb after a swing phase disturbance. The undisturbed step was defined as the preceding foot-ground contact of the contralateral limb. The recovery step was defined as the next foot-ground contact of the contralateral limb after the disturbed step. The undisturbed step width was recorded as the medial-lateral distance between the heel marker positions at initial foot-ground contacts before a disturbance. The disturbed step width was recorded between the contralateral heel marker during the undisturbed step and ipsilateral heel marker positions during the disturbed step. The recovery step width was recorded between the positions of the contralateral heel marker after the disturbed step and the ipsilateral heel marker after the disturbed step.

The recovery step width was hypothesized to be equal to the undisturbed step width for both medially and laterally directed disturbances with the control system turned on.



The robotic bench test was performed to provide a performance benchmark for the current prototype design. A 2-degree coronal rotation, requiring ∼150 milliseconds to complete, was applied to the prosthesis while under a 600-N load. When the control system was turned off, the 2-degree coronal rotation resulted in a ∼10 mm ZMP movement (Figure 4). The ZMP position continued to settle over the next whole second.

Figure 4
Figure 4:
Robotic bench test results with the controller off. A control system off trial shows that a 2-degree rotational input results in a ∼10-mm zero moment point (ZMP) disturbance.

When the control system was turned on, the prototype prosthesis was able to return the ZMP to within ±2 mm of its setpoint in ∼180 milliseconds in response to a 2-degree coronal rotation input from the robot (Figure 5). Interestingly, only a 0.73-degree coronal prosthetic ankle rotation was required to achieve steady state. As in the off condition, the ZMP position continued to settle after the movement. The observed settling behavior may be a result of the foam from which the prosthetic foot is constructed.

Figure 5
Figure 5:
Robotic bench test results with controller on. In response to a 2-degree coronal rotation input, the prototype prosthesis required ∼180 milliseconds to return the zero moment point (ZMP) to within ±2 mm of its setpoint (solid line). Only a 0.73-degree coronal ankle rotation was required (dashed line), which may be related to compression settling of the foam from which the prosthetic foot is constructed.


One individual gave informed consent to participate in this IRB-approved protocol. The participant was a transfemoral amputee of traumatic etiology, was 36 years of age, and had a body weight of 84 kg and height of 1.76 m.

The participant’s existing prosthetic prescription included an ischial-containment skin suction socket, a single-axis pneumatic weight activated stance control knee (model OP4; Medi, Whitsett, NC, USA), an aluminium pylon, and an energy storage and response prosthetic foot and cover (model 1C31=L27-1-P; Otto Bock, Minneapolis, MN, USA).


The standing balance test was performed to explore effects of the current prototype design during stationary conditions. The test documented the standard deviation of the position of COP during a 20-second steady-state sample (Table 1). When the subject was allowed visual feedback (eyes open condition), the control system appeared to have no effect on reducing sway. The standard deviation of the COP location was 3 ± 1 mm when control system was on or off. When the subject was not allowed visual feedback (eyes closed condition), the control system appeared to reduce the variation in COP location by 2 mm. The standard deviation of the position of the COP was 4 ± 1 mm with the control system on and 6 ± 2 mm with it off.

Table 1
Table 1:
Center of pressure (COP) movements during standing balance


The gait disturbance test was performed to explore the effects of the current prototype design during walking. The test measured step width obtained from an undisturbed step, a disturbed step, and a recovery step for both medially and laterally directed disturbances (Table 2). The medially directed disturbance test did not produce the desired result. When the control system was turned off, the medially directed disturbance resulted in a recovery step width of smaller magnitude than the disturbance step width, indicative of the need for additional steps to recover from the disturbance as expected. However, when the control system was turned on, the medially directed disturbance again resulted in a recovery step width of smaller magnitude than the disturbance step width. The system did not result in a recovery step width of the same size as the undisturbed step.

Table 2
Table 2:
Step widths observed during gait disturbance testing

The laterally directed disturbance test produced the desired result. When the control system was turned off, the laterally directed disturbance resulted in a recovery step width of greater magnitude than the disturbance step width, suggesting the need for additional steps to recover from the disturbance. When the control system was turned on, the system responded as desired to a laterally directed disturbance as the recovery step width was approximately the same size as the undisturbed step.


The purpose of this research was to discover if a powered inverting and everting prosthetic foot can improve amputee balance. The bench test provided a performance metric that can be compared with future iterations of the device. The torque output appeared to be sufficient, but the response time may be insufficient. Use of an embedded controller would likely improve response times. An important limitation is the excessive mass (2.9 kg) of the current design. A reduction in weight would likely improve patient acceptance of the device. A further limitation of the current design is that only transtibial subjects with a short residual limb or transfemoral subjects could be fit with the device owing to the relatively long length of the four-bar mechanism. A shorter mechanism would enable use by a larger patient population.

The transfemoral subject who participated in this test appeared to exhibit less sway than that measured in a population of transtibial amputees.13 Transtibial amputees of traumatic etiology (n = 18), participating in a similar protocol with the standard deviation of the position of the COP as a metric, exhibited 5 ± 2 and 7 ± 3 mm for eyes open and closed, respectively. This population was older (64 ± 10 years) and might be expected to exhibit more sway than the much younger subject (36 years) in this test. However, differences in amputation level might play a more significant role than age. The results from this case study with a prototype device suggest that balance may be improved with powered inversion and eversion, but only in reduced vision environments such as night time.

While walking, the difficult nature of lower limb amputee balance control is evident in amputees’ more variable17 and more lateral25,26 foot placement. In the case study reported here, the mean undisturbed step width ranged from 168 to 204 mm. These means are somewhat larger than the step widths of transfemoral amputees walking at different speeds12 and transtibials wearing ankles of different stiffnesses.27 Further enrollments would be necessary to determine if this is simply a specific subject aberration.

The case study results presented here suggest that the current prototype prosthesis may improve amputee walking balance in response to laterally directed disturbances. No improvement in balance was noted for medially directed disturbances. Importantly, the prosthetic foot keel properties may have a significant role; if the foot is too compliant in the coronal plane, the actuator will require a greater ankle coronal rotation to move the ZMP. The foot used in the current prototype is thought to be moderately stiff in the coronal plane and exhibits relatively low energy storage and return capacity. Further work is necessary to explore and optimize the coronal plane keel properties.

The developmental status of the current prototype prosthesis precluded testing on a larger sample population at this stage. Further development will enable use outside the laboratory, enabling an acclimation period. Tests with human subjects will then allow for measures of balance confidence, an important indicator of amputee stability.


It is widely accepted that lower limb amputees have significant balance problems. A powered prosthetic foot capable of producing inversion and eversion moments is one possible approach to reducing incidence of falls and improving balance confidence in this population. The results presented here suggest that powered inversion and eversion may improve standing balance under limited vision conditions and walking balance may be recovered more quickly in response to lateral disturbances.


1. Miller WC, Speechley M, Deathe AB. Balance confidence among people with lower-limb amputations. Phys Ther 2002; 82 (9): 856–865.
2. Miller WC, Speechley M, Deathe B. The prevalence and risk factors of falling and fear of falling among lower extremity amputees. Arch Phys Med Rehabil 2001; 82 (8): 1031–1037.
3. Miller WC, Deathe AB, Speechley M, Koval J. The influence of falling, fear of falling, and balance confidence on prosthetic mobility and social activity among individuals with a lower extremity amputation. Arch Phys Med Rehabil 2001; 82 (9): 1238–1244.
4. Miller WC, Deathe AB. A prospective study examining balance confidence among individuals with lower limb amputation. Disabil Rehabil 2004; 26 (14–15): 875–881.
5. Dillingham TR, Pezzin LE, MacKenzie EJ, Burgess AR. Use and satisfaction with prosthetic devices among persons with trauma-related amputations: a long-term outcome study. Am J Phys Med Rehabil 2001; 80 (8): 563–571.
6. Hagberg K, Branemark R. Consequences of non-vascular trans-femoral amputation: a survey of quality of life, prosthetic use and problems. Prosthet Orthot Int 2001; 25 (3): 186–194.
7. Legro MW, Reiber G, del Aguila M, et al.. Issues of importance reported by persons with lower limb amputations and prostheses. J Rehabil Res Dev 1999; 36 (3): 155–163.
8. Maki BE, Holliday PJ, Topper AK. Fear of falling and postural performance in the elderly. J Gerontol 1991; 46 (4): M123–M131.
9. King D, Zatsiorsky VM. Periods of extreme ankle displacement during one-legged standing. Gait Posture 2002; 15: 172–179.
10. MacKinnon CD, Winter DA. Control of whole body balance in the frontal plane during human walking. J Biomech 1993; 26 (6): 633–644.
11. Matsusaka N. Control of the medial-lateral balance in walking. Acta Orthop Scand 1986; 57 (6): 555–559.
12. Hof AL, van Bockel RM, Schoppen T, Postema K. Control of lateral balance in walking: experimental findings in normal subjects and above-knee amputees. Gait Posture 2007; 25: 250–258.
13. Hermodsson Y, Ekdahl C, Persson BM, Roxendal G. Standing balance in trans-tibial amputees following vascular disease or trauma: a comparative study with healthy subjects. Prosthet Orthot Int 1994; 18 (3): 150–158.
14. Bauby CE, Kuo AD. Active control of lateral balance in human walking. J Biomech 2000; 33 (11): 1433–1440.
15. Townsend MA. Biped gait stabilization via foot placement. J Biomech 1985; 18 (1): 21–38.
16. Hof AL, van Bockel RM, Schoppen T, Postema K. Control of lateral balance in walking. Experimental findings in normal subjects and above-knee amputees. Gait Posture 2007; 25 (2): 250–258.
17. Klute GK, Segal AD, Glaister BC, et al.. Variation in amputee foot placement during straight and turning gait. In: Proceedings of the 12th World Congress of the International Society for Prosthetics and Orthotics; Vancouver, Canada; 2007.
18. Lewis GSaP, Stephen J. Functional location of subject-specific joint axes of the human ankle complex. In: Proceedings of the International Mechanical Engineering Congress and Exposition; Anaheim, CA: ASME; 2004.
19. Vukobratovic M, Frank AA, Juricic D. On the stability of biped locomotion. IEEE Trans Biomed Eng 1970; BME-17 (1): 25–36.
20. Vukobratovic M, Juricic D. Contribution to the synthesis of biped gait. IEEE Trans Biomed Eng 1969; 16 (1): 1–6.
21. Hirai K, Hirose M, Haikawa Y, Takenaka T. The development of Honda humanoid robot. In: Proceedings of the IEEE International Conference on Robotics and Automation. Leuven, Belgium: Proceedings of IEEE Conference;1998.
22. Ziegler JG, Nichols NB. Optimal settings for automatic controllers. Trans Am Soc Mech Eng 1942; 64 (11): 759.
23. Segal AD, Treadwell CR, Klute GK. Biomechanical response to an error in foot placement. In: Proceedings of the 16th Annual Gait and Clinical Movement Analysis Society Meeting; Bethesda, MD; 2011.
24. Dempster WT. Space requirements of the seated operator: geometrical, kinematic, and mechanical aspects of the body with special reference to the limbs. Wright-Patterson Air Force Base, Ohio, Wright Air Development Center, 1955. WADC Technical Report; 55–159.
25. Murray MP, Drought AB, Kory RC. Walking patterns of normal men. J Bone Joint Surg Am 1964; 46: 335–360.
26. Murray MP, Kory RC, Sepic SB. Walking patterns of normal women. Arch Phys Med Rehabil 1970; 51 (11): 637–650.
27. Ventura JD, Klute GK, Neptune RR. The effect of prosthetic ankle energy storage and return properties on muscle activity in below-knee amputee walking. Gait Posture 2011; 33 (2): 220–226.

lower limb amputee; transtibial; artificial limb; prosthesis

© 2012 American Academy of Orthotists & Prosthetists