A powered prosthetic ankle joint for walking and running
© The Author(s) 2016
Published: 19 December 2016
Current prosthetic ankle joints are designed either for walking or for running. In order to mimic the capabilities of an able-bodied, a powered prosthetic ankle for walking and running was designed. A powered system has the potential to reduce the limitations in range of motion and positive work output of passive walking and running feet.
To perform the experiments a controller capable of transitions between standing, walking, and running with speed adaptations was developed. In the first case study the system was mounted on an ankle bypass in parallel with the foot of a non-amputee subject. By this method the functionality of hardware and controller was proven.
The Walk-Run ankle was capable of mimicking desired torque and angle trajectories in walking and running up to 2.6 m/s. At 4 m/s running, ankle angle could be matched while ankle torque could not. Limited ankle output power resulting from a suboptimal spring stiffness value was identified as a main reason.
Further studies have to show to what extent the findings can be transferred to amputees.
The current standard for prosthetic ankle joints are passive SACH (solid ankle cushioned heel) or carbon fiber ESAR (energy storage and return) feet. In contrast to the stiff SACH feet, ESAR feet are able to store energy during the stance phase and release it later during push-off [1, 2]. Through this they are able to mimic the function of the Achilles tendon . In contrast to human muscles, carbon feet are not able to create net positive work for ankle plantar- or dorsiflexion. Thus, actuation systems are required to achieve able-bodied ankle behavior. Different approaches with pneumatics  and electric motors [5–11] have been developed in recent years. In order to support amputees in common daily life activities like walking on flat terrain , stairs  or slopes , these activities were investigated and biomechanical characteristics were implemented in the active ankle joints.
In addition to the daily life movement requirements, amputees want to participate in social activities like sports. Cycling, swimming, and running are some possible sporting activities, with running also being fundamental to multiple activities such as ball games. Some special prosthetic solutions for different sports have been designed [14–16] such as waterproof legs and arms for swimming, electronic controlled knee joints with programmable modes specifically for cycling or skiing, (C-Leg, Otto Bock) as well as running and sprinting prostheses. As the limited range of motion (ROM) of existing prosthetic walking feet used for daily life make it unfavorable to run, amputees typically change their prosthesis when running. Ankle prosthesis designs for running and especially for sprinting have no heel element to make it possible to roll over the foot, resulting in a gait similar to forefoot running. The missing heel element increases effort for standing and other tasks of daily life and therefore makes the feet designed for running less appropriate for daily usage. Nevertheless, the limitations in ROM  passive walking feet can be also used for running. Czerniecki et al.  compared different generations of passive feet at 2.8 m/s transtibial amputee running. They found that a passive SACH (solid ankle cushioned heel) foot was only able to return 31% of the stored energy. ESAR (energy storage and return) feet were able to return 52% (Seattle Foot) to 84% (Flex Foot) of energy. For the Flex Foot about 0.19 J/(kg m) returned (stride length 2 m). In comparison, about 0.35 J/(kg m) of positive work at the ankle joint are required to perform running for able-bodied at the same speed . Positive work output for different running feet (Flex Run, Cheetah, Catapult) was about 0.1 J/(kg m) when evaluating various running speeds (2.5–3.5 m/s, ). Prosthetic foot peak power output of the same study was much lower (2.2–2.9 W/kg) compared to able-bodied running data (8.7 W/kg, 2.6 m/s, ).
A powered ankle prosthesis can provide positive work and overcome the limitations of ROM. Therefor sufficient motor power especially for higher walking and running speeds is required . Calculations show that different arrangements of elastic elements can decrease these requirements, especially in running gait [21, 22]. When considering the assistive effect of a series spring to mimic the ankle joint torque-angle curves with a motor, about 0.6 to 1.3 W/kg mechanical peak power output should be provided in walking (1.1–1.6 m/s) and 2.6–2.8 W/kg  for medium marathon running speeds of 2.6–3 m/s . A running speed of 4 m/s would require four times (3.9 W/kg) the motor peak power of the preferred walking speed.
Along with reduced power demands, elastic elements can also reduce energy requirements for a powered prosthesis. Modeling of a powered ankle prosthesis (to match able-bodied torque-angle profile) demonstrated that walking would require about 0.14 J/(kg m) to 0.18 J/(kg m) (1.1–1.6 m/s) and running about 0.22 J/(kg m) (2.6, 3 and 4m/s) mechanical work input when using a series spring for assistance . The design of a powered running prosthesis should be adapted to meet the increase of about 22 to 57%.
In addition to the power and energetic requirements, the control needs an adaptation to distinguish between walking and running and the desired speed. This could be critical when amputees want to run below or walk above preferred transition speed (2.1 m/s, PTS, ). Transitions from walking to running and running to walking must be realized quickly to support acceleration or deceleration during the transition process. Transition between movement and standing are critical for safety.
To investigate these topics, powered prosthetic ankles were designed to be capable of walking and medium speed running. First concepts were published by Bellman et al. in . The proposed design included two motors to realize tasks like jumping and running. Based on this idea a first running ankle was built in 2009 and tested with a military amputee [26, 27]. Evaluations demonstrated that the weight of the system and inertial properties of the actuators had to be improved in follow up versions.
The next generation of a powered running ankle was built in 2012. Instead of two brushed 150 W motors a brushless 200 W DC motor was used for the Walk-Run ankle. A controller to change between standing, walking, and running with speed adaptation was developed. For a first proof of concept, the system was evaluated with one able-bodied subject wearing the prosthesis in parallel to the fixed healthy foot. This article includes the results of the biomechanical evaluation of this prototype and the comparison to able-bodied subject data and corresponding prosthetic model estimations. Next to powered ankles also combined system with powered knee and ankle have been evaluated while running . Control concepts were developed to perform walk-run transitions .
The evaluation of the control response to speed changes and gait transitions is not a part of this study. The topic should be addressed in future work.
Design of the Walk-Run ankle
The current design has a weight of 1.9 kg, not including the battery and the electronics. Both the PC-104 used for control and the 4400 mAh 25.9 V battery (powering ankle and PC-104) are stationary on a test bench. For a subject (61 kg, 1.7 m height) with an amputation height of 0.31 m, the amputated limb mass is estimated to be about 2.3 kg according to calculations using the equation from . Thus, the weight of the artificial foot, including the adapters and socket, is only slightly higher compared to the healthy condition. A belt drive is used as a transmission between the motor and roller-screw.
Control of the Walk-Run ankle
All three signals (1000 Hz) were converted from voltage to acceleration or velocity in the “Raw Signal Processing” block. In addition, all the data was filtered for “Gait Determination” with a 2nd order Butterworth filter using a cutoff frequency of four to get a general representation of the gait behavior.
In the “Gait Determination” the incoming three sensor signals are compared to sensor reference trajectories starting at the beginning of gait cycle. The most similar reference is identified calculating a cumulative sum of error between the curves. This results in a measurement of the current gait mode and the movement speed.
As a last step, the gait mode, the speed, and the gait percentage are used in the “Motor Trajectory Generation” block to determine the motor trajectory using a reference lookup table.
The desired motor trajectory is followed by the implementation of PD control. PD values enabled the system to follow the desired motor trajectories with a mean difference of 0.2 mm over the gait cycle in 1.6 m/s walking. As previously described, a belt drive is used in conjunction with a roller-screw to effect dorsiflexion or plantarflexion at the prosthetic ankle joint.
Control and model reference data
Three different experiments were used to define reference data for the prosthetic control.
In the first experiment , able-bodied (21 subjects) walked and ran (0.5–2.6 m/s) on a treadmill with integrated 3D force sensors (Kistler, 1000 Hz). Kinematics were recorded by high-speed infrared cameras (Qualisys, 240 Hz). A second, similar experiment (7 subjects), was repeated for higher running speeds (3–4 m/s). Ankle angles and torques were calculated [24, 30] and used to determine motor trajectories for the powered ankle based on the method presented in [31, 32]. The used method determines the deflection of the Walk-Run ankle spring when applying recorded torque data from the able-bodied experiments. Motor trajectories must accomplish the length change to match human ankle angle data.
Reference times, used in the controller to determine the gait percent, were taken from both treadmill experiments.
In the third experiment, one subject (height: 173 m, mass: 63.5 kg, age: 23, male) walked and ran with the same speeds as in the previous experiments over level ground. The pace was set by a bicycle. A two axis accelerometer and a rate gyro sensor were mounted to a small wireless board. The board was affixed at the shank near the ankle joint to measure reference values for all gait conditions. The fixation height of the system was similar to the height of the sensors of the Walk-Run ankle. About 40 strides were measured for each condition at the instrumented leg. Mean values for the gait cycle were created and filtered in like manner to the incoming data in the prosthetic controller (2nd order Butterworth filter, cutoff frequency 4 Hz). The frequency was selected in order to get the general shape of the signal rather than measuring the real kinematics.
All the experiments were in compliance with the Helsinki Declaration.
Evaluation of Walk-Run ankle
Calculation of biomechanical parameters
Biomechanical parameters of the Walk-Run ankle were calculated using the motor encoder and the ankle angle encoder. The length of the spring was determined by the position of the nut (motor) in combination with the ankle angle. Using Hooke’s law, the spring force could be calculated. Spring and nut velocity were determined by numerical differentiation. By using the lever arm of the spring towards the ankle joint, ankle torque was calculated. Ankle torque could be related to force applied by the nut using the lever arm of the ankle joint towards the roller screw. By multiplying spring velocity and spring force the mechanical power was calculated. Nut output power was calculated by multiplying nut force and nut velocity. The sum of spring power and nut power is equal to the ankle joint power.
Stride length of the bypass site was calculated using the speed of the treadmill and the time required for the gait cycle.
Walking 1.6 m/s
Mechanical work provided by the motor (nut output, positive + absolute of negative), the spring (positive) and the overall robotic ankle in J/(kg m)
The maximum power output of 3 W/kg is almost equal, but the timing of the maximum power of the Walk-Run ankle is later compared to the model curve.
Running 2.6 m/s
Running 4.0 m/s
For running at a treadmill speed of 4.0 m/s, the controller identified a speed of 3.7 m/s. The reference group used a stride length of 2.81 m at 4 m/s running. The subject with the bypass had a mean stride length of about 2.67 m. When comparing to the reference data of 4.0 m/s, we identified the same ankle angle behavior as in 2.6 m/s running. Dorsiflexion happens earlier and plantarflexion later compared to the model. After takeoff, a larger difference between model and system ankle angle occurs. Similar fluctuations after takeoff with less amplitude can be found for 2.6 m/s running and 1.6 m/s walking. Similar to running at 2.6 m/s, there is a gap between the reference nut position and the real nut position in the stance phase. The spring is not elongated in the same way as the model (Fig. 6). As a result of nut position and spring length, ankle torque is less compared to the model values (Fig. 7). In addition, peak power and work [nut output: 0.12, spring: 0.17 J/(kg m)] are less compared to the model data [nut output: 0.22, spring: 0.31 J/(kg m), Table 1]. Joint peak power should be about 13.4 W/kg. The robotic ankle can provide about 8 W/kg in total combining the effect of the spring and the motor.
Replication of human gait
The Walk-Run ankle could mimic reference subject walking and running behavior when testing using a bypass system on a healthy subject. When comparing the reference and the experimental data only subtle differences could be identified in walking. For running at 2.6 m/s minor differences for the ankle torque and the ankle angle exist. Work output at the ankle at this speed is comparable to able-bodied data. Thus we demonstrated that the Walk-Run ankle can overcome limitations in ROM  and in generating positive work compared to passive prosthetic walking (used for running, ) and running  feet. Human like ROM was also achieved at 2.25 m/s running for the Vanderbilt prototype  and at 2.5 m/s running for the Ruggedized Odyssey Ankle  during amputee testing.
At 4.0 m/s the biomechanics of the Walk-Run ankle differ from able-bodied gait. The positive work output is reduced compared to the model reference but still double compared to passive running feet (model 0.35, RA 0.2, passive 0.1 J/(kg m), ). To some extent differences between model and experiment could be caused by the difference of the detected speed (3.7 instead of 4.0 m/s) and the resulting difference in desired motor trajectory. In running, a gap between the desired nut reference (Nut Ref) and the achieved nut position of the robotic ankle exists (Nut RA) between 10 and 43% of the gait cycle.
As motor current is not at its maximum (almost reached between 16 and 28%) during some part of this period, it could be possible to reduce the difference by an adaptation of the control parameters. During Midstance (4 m/s), the motor was not able to deliver enough torque to move the nut, but motor torque was enough to hold position. It could be tested to what extent an earlier pretensioning of the motor, when less torque is applied, could improve elastic recoil.
The differences in nut trajectory during stance cannot explain the difference in ankle angle mainly caused by the spring deflection at the beginning of the stance phase. The spring is elongated faster than the model predicts. For a comparable joint torque, such a behavior could appear when spring stiffness is too low. Additional reasons for the increase in ankle torque could be a subject specific gait pattern, differences of the prosthesis to the human in structure and control, but also asymmetries caused by the bypass system. Shank velocity data and video show that the subject is touching the ground while the shank is in forward rotation. This is in contrast to the natural pattern where at touch down the shank is already in backward motion. The so called leg retraction  is used to avoid impacts. The longer leg (about 3.5 cm) of the bypass system seems to cause an increased impact that results in higher ankle torques compared to the reference data. For a more detailed analysis on the cause, motion capturing and force measurements of both legs would be required.
Model assumptions on the energy flow
The modeling approach to determine the optimal energetic interaction of the spring and the motor uses mean values of torque and angle data of several subjects as an input. Based on the torque related spring deflection it calculates the corresponding length change for the nut to mimic desired ankle angle. It is assumed that lower extremity kinematics and kinetics will be similar when using the Walk-Run ankle including nut trajectories based on this approach. The data shows that when using the bypass design such an assumption seems to be acceptable for walking. Higher model agreement of the energy flow between the nut output and the spring was achieved without the large differences in leg length and leg mass when testing the Walk-Run ankle at an unilateral amputee .
In contrast, the running power curves show (Fig. 5) large differences between model predictions and achieved robotic ankle performance. Due to the results for amputee walking, the authors assume that the majority is caused by the bypass testing approach (leg length, mass) and the Walk-Run ankle limitations in peak torque. In addition some part of the differences may result from individual gait characteristics but also structural differences of the device compared to biology. Several muscles are represented by just one actuator and a biarticular coupling to the thigh is missing.
The powered ankle was designed to add additional functionality, to reduce user effort during locomotion, and to regain interlimb symmetry (unilateral amputee). Also when the energy flow between the spring and the motor is comparable to the model assumption and the torque and angle curves are almost similar to the reference, a transtibial amputee must not necessarily have lower locomotion effort combined with higher symmetry. Thus, further measurements are required to evaluate the user benefits of having powered assistance during running but also walking at the ankle joint. In level walking benefits for preferred walking speed , oxygen consumption , and reductions in stance to swing duration asymmetries  could be demonstrated.
Spring fluctuations after take off
After takeoff, fluctuations of the spring occurred. This was more distinctive for the higher speeds. It is unclear how much these fluctuations in the early flight phase affect the gait stability. It could be worthwhile to investigate if it is possible to reduce fluctuations by using the motor for damping. As a result of the fluctuations, ankle angle after takeoff differs from reference data especially at 4.0 m/s running.
Touch down detection
Limitations of the Walk-Run ankle
The Walk-Run ankle nut was able to deliver about 3 W/kg mechanical peak power at 4 m/s running for a subject weight including bypass and prosthesis of about 67 kg. This 200 W maximum was almost reached for 2.6 m/s running. For running at 4.0 m/s, the model predicted a required nut peak power output of 8 W/kg for the chosen stiffness. The motor was not able to deliver these 536 W. Using an optimal stiffness could reduce requirements for running at 4 m/s to 3.9 W/kg . This could greatly improve performance for the fastest running speed.
Peak torque limits during the experiment were at about 140 Nm. This limit was reached in both running conditions while in 4 m/s running it clearly limited the powered ankle performance.
Also, while the desired and achieved motor trajectories for 4 m/s running were not matching, the final ankle angle was almost equal (neglecting angle caused by fluctuations after take off) to the reference data. Thus we assume there could be a better solution that is a compromise between following of the desired motor trajectory and resulting similarity to able-bodied gait biomechanics and symmetry measures. Precise following will require a large amount of energy but gait quality may only improve slightly. Investigations on this topic could be made by measuring kinematics and kinetics to compare symmetry, joint work, limb loading, locomotion speed, balance, and by using a spiroergometry system to measure user effort.
For the first preliminary study on the Walk-Run ankle, we found that it is possible to mimic human reference ankle joint behavior for walking and running up to a speed of 2.6 m/s. It was demonstrated that the Walk-Run ankle can overcome limitations in ROM and in generating positive work compared to passive prosthetic walking and running feet.
Compared to the reference data, at 4 m/s running, ankle torque from the prosthesis was not adequate. As the desired ankle angle was almost achieved even without the correct torque, the authors think that it is worthwhile to investigate the compromise between following calculated motor trajectories and achieved similarity to able-bodied gait biomechanics. The authors believe that there is a high potential to decrease peak power and energy requirements when motor trajectories are manipulated in a way to have reduced acceleration and less changes in direction.
Some hardware and control issues, like differences in motor desired and achieved trajectory, fluctuations of the spring, and touch down detection, could be identified. Some of these may occur due to the bypass setup. Thus, the next step would be to test the ankle on an amputee to avoid larger differences in leg length or mass distribution in between legs.
The data collected and presented in the paper was derived from the controller M. Grimmer developed for this system. This information is being used to influence the design of a ruggedized powered ankle that also has running capability. The new and improved design is being updated by SpringActive to include weight reducing titanium springs that have a longer life, higher efficiency bearings that increased system efficiency, and embedded electronics. First preliminary data can be found in .
MG was responsible for programming the high and medium level gait control. In addition he was in the lead for drafting of the manuscript and experiments. MH was in the lead for project planing, prosthesis design specifications and organization. RH was responsible for the design of the Walk-Run ankle, sensor integration, wiring and assembling. AB was responsible for the setup of the PC-104 platform, including sensor and motor interfaces, used to control the robotic ankle. In addition he was programming the main framework for the prosthetic controller including the low level control. JW provided supervision on control developments. In addition he assisted for interpretation on results to improve future prosthetic developments. MH, RH, AB, and JW assisted for experiments. AS assisted for the biomechanical analysis. Depending on there special field of knowledge, all authors including KH and TS participated in the critical discussion of the results and helped for revision of the manuscript. All authors read and approved the final manuscript.
The authors acknowledge S.W. Lipfert for providing human data.
Matthew Holgate, Alexander Boehler, Jeff Ward, Kevin Hollander and Thomas Sugar are owners of SpringActive. Robert Holgate was previously an employee of SpringActive. SpringActive, Inc. supplied the Walk-Run ankle for this research and is currently developing commercial powered ankle prostheses. Martin Grimmer and Andre Seyfarth are with the Technische Universität Darmstadt and have no competing interests.
Availability of data and materials
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.
Consent for publication
The subject gave consent to puplish gait data recorded in the study.
This article has been published as part of BioMedical Engieering OnLine Vol 15 Suppl 3, 2016: Towards Active Lower Limb Prosthetic Systems: Design Issues and Solutions. The full contents of the supplement are available online at http://biomedical-engineering-online.biomedcentral.com/articles/supplements/volume-15-supplement-3.
Ethics approval and consent to participate
The research was performed in accordance with the Declaration of Helsinki. The subject gave informed consent to participate in the study. The study was approved by the external Institutional Review Board, Argus IRB, Inc. Tucson, Arizona (IRB registration number with the Department of Health and Human Services: IRB00001496).
Research reported in this publication was supported by the the DFG grant SE1042/8 and the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under award number R43HD072402. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Matthew Holgate, Alexander Boehler, Jeff Ward, Kevin Hollander, Robert Holgate, and Thomas Sugar are with SpringActive. Contributions were funded by SpringActive and NIH. Martin Grimmer and Andre Seyfarth were employees at Technische Universität Darmstadt. Martin Grimmer was also funded by the DFG grant SE1042/8. The publication costs of this article are covered by internal funds of the Lauflabor Locomotion Laboratory, Technische Universität Darmstadt.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Gitter A, Czerniecki JM, DeGroot DM. Biomechanical analysis of the influence of prosthetic feet on below-knee amputee walking. Am J Phys Med Rehabil. 1991;70(3):142.View ArticleGoogle Scholar
- Czerniecki JM, Gitter A, Munro C. Joint moment and muscle power output characteristics of below knee amputees during running: the influence of energy storing prosthetic feet. J Biomech. 1991;24(1):63–5.View ArticleGoogle Scholar
- Farris DJ, Sawicki GS. Human medial gastrocnemius force-velocity behavior shifts with locomotion speed and gait. Proc Natl Acad Sci USA. 2012;109(3):977–82.View ArticleGoogle Scholar
- Versluys R, Lenaerts G, Van Damme M, Jonkers I, Desomer A, Vanderborght B, Peeraer L, Van der Perre G, Lefeber D. Successful preliminary walking experiments on a transtibial amputee fitted with a powered prosthesis. Prosthet Orthot Int. 2009;33(4):368–77.View ArticleGoogle Scholar
- Hitt JK, Sugar TG, Holgate M, Bellman R. An active foot-ankle prosthesis with biomechanical energy regeneration. J Med Devices. 2010;4:011003.View ArticleGoogle Scholar
- Zhu J, Wang Q, Wang L. Pantoe 1: biomechanical design of powered ankle-foot prosthesis with compliant joints and segmented foot. In: IEEE/ASME international conference on advanced intelligent mechatronics (AIM). IEEE; 2010. p. 31–6.Google Scholar
- Sup F, Varol HA, Mitchell J, Withrow TJ, Goldfarb M. Self-contained powered knee and ankle prosthesis: initial evaluation on a transfemoral amputee. In: IEEE international conference on rehabilitation robotics (ICORR). IEEE; 2009. p. 638–44.Google Scholar
- Suzuki R, Sawada T, Kobayashi N, Hofer E. Control method for powered ankle prosthesis via internal model control design. In: international conference on mechatronics and automation (ICMA). IEEE; 2011. p. 237–42.Google Scholar
- Herr HM, Grabowski AM. Bionic ankle-foot prosthesis normalizes walking gait for persons with leg amputation. Proc R Soc B Biol Sci. 2012;279(1728):457–64.View ArticleGoogle Scholar
- Ferris AE, Aldridge JM, Rábago CA, Wilken JM. Evaluation of a powered ankle-foot prosthetic system during walking. Arch Phys Med Rehabil. 2012;93(11):1911–8.View ArticleGoogle Scholar
- Cherelle P, Matthys A, Grosu V, Brackx B, Van Damme M, Vanderborght B, Lefeber D. The amp-foot 2.0: a powered transtibial prosthesis that mimics intact ankle behavior. In: 9th National Congress on theoretical and applied mechanics; 2012.Google Scholar
- Aldridge JM, Sturdy JT, Wilken JM. Stair ascent kinematics and kinetics with a powered lower leg system following transtibial amputation. Gait Posture. 2012;36(2):291–5.View ArticleGoogle Scholar
- Sup F, Varol HA, Goldfarb M. Upslope walking with a powered knee and ankle prosthesis: initial results with an amputee subject. IEEE Trans Neural Syst Rehabil Eng. 2011;19(1):71–8.View ArticleGoogle Scholar
- Webster JB, Levy CE, Bryant PR, Prusakowski PE. Sports and recreation for persons with limb deficiency. Arch Phys Med Rehabil. 2001;82(3):38–44.View ArticleGoogle Scholar
- Radocy B. Upper-extremity prosthetics: considerations and designs for sports and recreation. Clin Prosthet Orthot. 1987;11(3):131–53.Google Scholar
- Bragaru M, Dekker R, Geertzen JHB. Sport prostheses and prosthetic adaptations for the upper and lower limb amputees: an overview of peer reviewed literature. Prosthet Orthot Int. 2012;36(3):290–6.View ArticleGoogle Scholar
- Sanderson DJ, Martin PE. Joint kinetics in unilateral below-knee amputee patients during running. Arch Phys Med Rehabil. 1996;77(12):1279–85.View ArticleGoogle Scholar
- Grimmer M, Seyfarth A. Mimicking human-like leg function in prosthetic limbs. In: Neuro-robotics. Berlin: Springer; 2014. p. 105–55.Google Scholar
- Baum BS. Kinetics in individuals with unilateral transtibial amputations using running-specific prostheses. PhD thesis, University of Maryland, College Park; 2012Google Scholar
- Grimmer M, Eslamy M, Seyfarth A. Energetic and peak power advantages of series elastic actuators in an actuated prosthetic leg for walking and running. In: Actuators, vol. 3. Basel: Multidisciplinary Digital Publishing Institute; 2014. p. 1–19.Google Scholar
- Grimmer M, Eslamy M, Gliech S, Seyfarth A. A comparison of parallel-and series elastic elements in an actuator for mimicking human ankle joint in walking and running. In: 2012 IEEE international conference on robotics and automation (ICRA). IEEE; 2012. p. 2463–70.Google Scholar
- Eslamy M, Grimmer M, Seyfarth A. Effects of unidirectional parallel springs on required peak power and energy in powered prosthetic ankles: comparison between different active actuation concepts. In: IEEE international conference on robotics and biomimetics (ROBIO); 2012.Google Scholar
- Leyk D, Erley O, Ridder D, Leurs M, Ruther T, Wunderlich M, Sievert A, Baum K, Essfeld D. Age-related changes in marathon and half-marathon performances. Int J Sports Med. 2007;28(6):513–7.View ArticleGoogle Scholar
- Lipfert SW. Kinematic and dynamic similarities between walking and running. Hamburg: Verlag Dr. Kovac; 2010.Google Scholar
- Bellman, R.D., Holgate, M.A., Sugar, T.G.: Sparky 3: Design of an active robotic ankle prosthesis with two actuated degrees of freedom using regenerative kinetics. In: Biomedical Robotics and Biomechatronics, 2008. 2nd IEEE RAS & EMBS international conference on BioRob 2008. IEEE; 2008. p. 511–16.Google Scholar
- Hitt J, Merlo J, Johnston J, Holgate M, Boehler A, Hollander K, Sugar T. Bionic running for unilateral transtibial military amputees. MILITARY ACADEMY WEST POINT NY; 2010.Google Scholar
- Schinder A, Genao C, Semmelroth S. Methodology for control and analysis of an active foot-ankle prosthesis. Methodology. 2011.Google Scholar
- Shultz A, Lawson BE, Goldfarb M. Running with a powered knee and ankle prosthesis. 2014.Google Scholar
- Huff AM, Lawson BE, Goldfarb M. A running controller for a powered transfemoral prosthesis. In: Engineering in medicine and biology society (EMBC), 2012 annual international conference of the IEEE. IEEE; 2012. p. 4168–71.Google Scholar
- Günther M, Sholukha V, Keßler D, Wank V, Blickhan R. Dealing with skin motion and wobbling masses in inverse dynamics. J Mech Med Biol. 2003;3(3–4):309–35.View ArticleGoogle Scholar
- Hollander KW, Sugar T. Design of the robotic tendon. In: Design of medical devices conference (DMD); 2005.Google Scholar
- Grimmer M, Seyfarth A. Stiffness adjustment of a series elastic actuator in an ankle-foot prosthesis for walking and running: the trade-off between energy and peak power optimization. In: IEEE international conference on robotics and automation (ICRA). IEEE; 2014. p. 1439–44.Google Scholar
- Ward J, Schroeder K, Vehon D, Holgate R, Boehler A, Grimmer M. A rugged microprocessor controlled ankle-foot prosthesis for running. In: 39th annual meeting of the american society of biomechanics (ASB). Columbus; 2015.Google Scholar
- Seyfarth A, Geyer H, Herr H. Swing-leg retraction: a simple control model for stable running. J Exp Biol. 2003;206(15):2547–55.View ArticleGoogle Scholar
- Grimmer M. Powered lower limb prostheses. PhD thesis, Technische Universität Darmstadt; 2015. http://tuprints.ulb.tu-darmstadt.de/4382.
- Mancinelli C, Patritti BL, Tropea P, Greenwald RM, Casler R, Herr H, Bonato P. Comparing a passive-elastic and a powered prosthesis in transtibial amputees. In: Engineering in medicine and biology society, EMBC, 2011 annual international conference of the IEEE. IEEE; 2011. p. 8255–58.Google Scholar