Microprocessor Lower Limb Prosthetics: Review of Current State of the Art

Martin, Jay BS, CP, LP; Pollock, Andrew BS; Hettinger, Jessica BS

JPO Journal of Prosthetics & Orthotics:
doi: 10.1097/JPO.0b013e3181e8fe8a

Today, there exist a variety of lower limb prosthetic devices. Each device differs in design and function. A relatively new and particularly interesting category of prosthetics are those that are microprocessor controlled. These devices ultimately seek to mimic the human anatomical control system, surpassing the abilities of mechanical-based prosthetics, by incorporating sensor input, processing, output actuation, and feedback input features. However, it is important to understand that not all microprocessor-controlled devices operate in the same manner; they each enhance amputee ambulation through a combination of different input methods, control system accommodations, and output actuation strategies. Currently, there does not exist a lower limb device that features afferent communication with the brain. Microprocessor-controlled prosthetic devices attain input information intrinsically through computational sensors and/or extrinsically through human interactive sensors. The device processes the sensor inputs collected and appropriately controls for the gait environment. Devices differ in the ability to accommodate for the various environmental factors and in the extent to which accommodation can be achieved. The resultant output of the device incorporates resistive and/or powered actuation strategies into each step. A classification of currently available microprocessor-controlled knees and feet is shown.

In Brief

Microprocessor-controlled prosthetic devices ultimately seek to mimic the human anatomical control system, surpassing the abilities of mechanical based prosthetics by incorporating sensor input, processing, output actuation, and feedback input features. Not all microprocessor controlled devices operate in the same manner; they each enhance amputee ambulation through a combination of different input methods, control system accommodations, and output actuation strategies. This paper provides a classification of currently available microprocessor controlled knees and feet.

Author Information

JAY MARTIN, BS, CP, LP, ANDREW POLLOCK, BS, AND JESSICA HETTINGER, BS, are affiliated with the Orthocare Innovations, Oklahoma City, Oklahoma.

Disclosure: The authors declare no conflict of interest.

Correspondence to: Jessica Hettinger, BS, 840 Research Parkway, Suite 200, Oklahoma City, OK 73104; e-mail: jhettinger@orthocareinnovations.com

Article Outline

Prosthetics innovation has long sought to provide a greater function for amputees. As prosthetics technologies have improved, amputee's abilities have increased. In addition, safety, symmetry, and life-like appearance have all improved, whereas energy expenditure of ambulation has decreased.

As new technologies in computer processing, materials, and actuation strategies have emerged, prosthetic design has begun to reshape what an amputee's experience is like after the loss of a limb. Today, amputees are able to do amazing feats with the use of their prosthetic devices and, in general, can function better than ever. However, there still remains a significant gap between the most advanced prosthetics and the human body, showing that there still exists a great need for innovation.

When comparing the most advanced lower limb prosthetic devices with their anatomical counterparts, we find that there are noticeable differences in kinetics and kinematics.1–5 Vickers et al.1 reported the differences in hip and knee angles experienced by able-bodied and amputee individuals during the different phases of gait, whereas Perry et al.3 and Powers et al.4 reported the associated prosthetic use with extended heel only contact and, thus, limited knee flexion of the residual limb. These differences, and others, create limitations in the ability to perform daily living activities, especially in ambulation environments other than level walking in a patient's normal pace range.1,6–8 According to McIntosh et al.,6 the use of a conventional prosthetic causes instability in downhill walking, as the user attempts to shift his center of mass forward to ambulate the slope amid limited range of motion of the residual limb. Waters et al.9 confirmed that as the level of amputation increases proximally, it is increasingly more difficult to achieve normal ambulation and function.

Mechanical-based prosthetic systems do not have the ability to fully mimic the natural movements of the limbs in all conditions and environments. Normal daily ambulation requires accommodation for force, speed, or terrain alterations past the narrow range offered by mechanical devices. This results in limited and abnormal gait, decreased safety, and higher energy expenditure.5,10,11 Specifically, Vrieling et al.11 identified a loss of balance associated with mechanical prosthetic sloped walking resulting from the user's variations in hip and knee flexion and extension and inability to adjust for the ambulation environment.

Our bodies are not merely mechanical-based systems—rather, we have an advanced controller (brain) to adapt our actuators (limbs) to the surrounding environment. The inclusion of computer-controlled—or intelligent—prosthetics is essential to fully mimic the natural gait cycle in all conditions and in all environments (Figure 1).

In general, this new category of lower limb prosthetic devices is being designed to ultimately replicate the anatomical control system through incorporating sensor input (nerve information from the limb), processing (brain), output (muscle excitation), and feedback input (nerve information from the limb). Fan et al.12 demonstrated through a viability study of a haptic feedback system applied to the lower limbs that this full feedback loop will be essential to provide optimal coordinated movement.

Many of these newer prosthetic devices that have incorporated microprocessor or sensor technology into their designs have demonstrated great benefits to those using them, although full anatomical replication remains to be achieved.13–16 Kaufman et al.13 identified improved knee flexion during limb loading with the use of a microprocessor-controlled knee versus the hyperextension and locking associated with a conventional mechanical knee. Schmalz et al.14 identified enhanced energy management and superior knee resistance throughout the gait cycle in a clinical study of the microprocessor-controlled Otto Bock C-Leg at various ambulation speeds. Johansson et al.16 confirmed smoothness of gait associated with the adaptive C-Leg, whereas Segal et al.15 recognized improved gait symmetry. This field of lower limb computer-controlled prosthetic design remains very much in its infancy, although radical changes are beginning to take place, reshaping lower limb prosthetic design as we know it. The new arena of computer-controlled lower limb prosthetics offers a variety of new approaches that need to be categorized for understanding of their similarities and differences.

Conventional prosthetic knees and feet are categorized according to their general design and function. Feet, for instance, have long been classified into categories including but not limited to solid ankle cushion heel, single axis, multiaxis, energy storing dynamic response, or elastic keel. The knees offer a similar classification system including categories such as constant friction, locking knees, weight-activated stance, polycentric, pneumatic, and hydraulic. Both knees and feet now have a new distinct category—microprocessor controlled.

The category of “microprocessor controlled,” however, can be broken down further into subcategories. It would not be accurate to lump all computer-controlled feet or knees into one category, but rather, there are numerous significant differences within that one group—making the variations in designs distinct in their function abilities and patient population. Each of these devices' approaches has their own set of benefits and drawbacks, many of which are yet to be fully understood, creating a need for further research to quantify their value to the end users.

In an attempt to offer a framework for discussion, the breakdown of input method, control system processing, and output actuation strategies will be used. This is necessary because there are vast differences among these categories between many of the new designs that are currently on the market and in the research arena.

Back to Top | Article Outline


One way to categorize these two general methods of providing input data to the microprocessor that can be used to control the prosthetic device is computational intrinsic control (CIC) or interactive extrinsic control (IEC).14,17,18 CIC is used in the design of the Otto-Bock C-Leg, as described by Schmalz et al.14; the intrinsic sensors detect ambulation cadence and environment and allow adjustment in resistance to accommodate for these variations (Table 1). IEC is present in the design of upper limb prosthetics, which incorporate user myoelectric signaling into functional movement of the device, as documented by Schärer17 and Parker et al.18

CIC takes information for control of the prosthetic movement and function from sensors within the device and not in direct communication with the user. This sensor input is sent to the microprocessor, where the next appropriate movement of the prosthetic device is determined by the microprocessor, not the amputee's brain, and the actuator is adjusted to create a change in movement, and so on (Figure 2).

Examples of sensors frequently used in the field to accomplish this are joint angle sensors (hall effect, potentiometers, and encoders), force sensors (strain gauges, pressure pads, and load cells), and motion sensors (accelerometers, gyroscopes, and goniometers). These may be used to provide input directly (i.e., potentiometer reading giving direct correlation to angle of the joint) or indirectly with additional input information (i.e., potentiometer angle reading being used along with time sensor to extrapolate angular acceleration of the joint). For instance, force information collected from a strain gauge during the gait cycle may be used to determine real-time required stiffness data for the joint's movement, or it may be used to indirectly determine the phase of the gait cycle.

CICs may be used to effectively coordinate movement of a prosthetic device through its own internal feedback loop but do not directly communicate with the body's intended movements based off any form of neural interaction. Because there is, and will likely remain for years to come, a limited amount of neural input available for prosthetic control, extrapolating the available input, even if simply a sensor-based input, into as much useful information as possible is critical. Segal et al.15 consented in their evaluation of microprocessor-controlled knees that this sensor information collected is intended for use in achieving as functional of gait as is currently possible. Therefore, it is essential to find creative methods of extrapolating that limited sensor information into a larger group of worthwhile data to control the device.15

There is also a fine balance between giving the prosthesis too much or too little control on its own. CIC may put the user at risk by limiting stability and consistency of prosthetic movement if the sensor-based microprocessor control algorithms incorrectly determine the appropriate response to the ambulated environment. It is essential for the prosthesis to provide consistent and correct movement independent of the user's activities. Any inconsistencies between the intended movement of the user and the movement of the prosthetic device lead to compromised safety, decreased gait efficiency, and ultimately rejection of the device.

A challenge of some current control methods is the inability of the device to determine and replicate movement parameters for activities other than typical gait cycle motions, such as exercising activities or stairs. Using the available sensor information, the prosthesis can readily provide input to the microprocessor to control typical walking motions effectively; however, typical ambulation requires atypical movements, as the user negotiates the “real world.” The more that a prosthetic device is able to produce properly intended and necessary movements, in all environments and conditions, through CIC strategies, the less dependant the system will be on user input and possibly less dependant the user will be on afferent IEC feedback.

In CIC systems, the sensors collect data and correspond that to real-time alterations in the actuator (minus inherent electronics and mechanical actuator delays). The limitation of such a system, however, is that the human anatomical control system also uses prereal-time sensor data. During ambulation of an able-bodied person, visual and auditory cues alert us to our surrounding environment. Several steps ahead, we can see an obstacle in our path and correspond our leg movement to adjust for that obstacle without looking down once we come upon it. Each of these pieces of information is taken into account for our gait before the event.

By using sensor data only, these important prereal-time cues are missing, and safety may be compromised. Another method may help to overcome this limitation, by ultimately allowing for the user to effect the movement of the device, through electromyogram (EMG), neural integration, or similar strategies.

IEC uses human-interacting methods of determining appropriate movement and function of the prosthetic device (Figure 3). These methods, many of which are currently under development, and not used in conventional lower limb prosthetics, use efferent information from the user by means such as EMG sensors, pattern recognition systems, cortical, or peripheral nerve sensors, and can possibly provide afferent information to the user through haptic feedback or cortical or peripheral nerve feedback. These systems often use computational control parameters in conjunction with human-induced input methods, their ability to take information directly or indirectly from the user offers uniqueness in the control feedback loop.

As is found in upper limb prosthetics, the types and methods of available efferent input from the user to the device are often unique to that user. Not every prosthetics user, because of medical, cost, or other considerations, will be able to have neural input associated with their prosthesis. Similarly, the ability and desire for users to have afferent feedback may differ significantly.

For efferent information from the user to the device, the microprocessor may correlate input parameters to the actuators in conjunction with the intended neural input. Some systems may result in a proportional movement such as that in many current upper limb myoelectric devices,19 whereas other system's functions may result in a preselected movement only, as in upper limb chin switch.20 For example, the EMG-controlled prosthetic limbs described by Ryait et al.19 sense electrical impulses generated through user muscle contractions and incorporate the magnitude of these contractions into the speed and grip strength of the resultant motion of the prosthetic. Conversely, the upper limb system described by Lake and Miguelez20 is actuated by a push from the chin; proportional control is lacking.

In using IEC systems, a full closed loop sensory feedback is increasingly beneficial by allowing the user to provide intended movement information to the device (efferent information) and to have the device communicate that movement information back to the user (afferent information). Able-bodied individuals who lack sensory feedback also have limited ability to control coordinated movements.21 A limb dynamic study of normal individuals compared with those lacking appropriate sensory feedback proprioception by Sainburg et al.21 illustrates that individuals experiencing normal proprioceptive feedforward commands effectively coordinate muscle actions and joint interaction torques, whereas deafferented individuals fall short of attaining the balance. As is often found in current upper limb myoelectric systems, while afferent communication may not be intentionally communicated to the user, the user may use indirect cues from the device, such as differences in the motor sounds or motor vibratory cues, to gain a sense of afferent feedback during use. These second-hand cues are often essential to the user to have an understanding of what movements are taking place within the prosthesis, based on their efferent input.

Whether purposely integrated into a device or taken as second-hand cues from the device, afferent IECs stimulate the user in a manner to achieve communication of kinetic or kinematic data. These signals may generally be characterized as sensory feedback mechanisms.

Afferent IECs are becoming increasingly important in this next generation of computer-controlled prosthetics. The movement from these devices may be otherwise disassociated from the user, resulting in a lack of understanding of proprioception, and both spatial orientation and resistance or actuation characteristics. With conventional systems, users often gain extended physiological proprioception (EPP) with their device.22 A tennis player or a golfer experiences EPP in their ability to have a sense of where the end of their “static” racket or club is at in relation to their body. They, in essence, have a sense of proprioception, or spatial orientation, of a nonanatomical static extension of their body. With “adaptable” prosthetics where the position and stiffness of the joint changes, there may be limited association of joint position of the prosthetic as it adapts to the environment.

With conventional static transtibial prosthetics, users may gain an acute awareness of their prosthesis position in space much the same way that a golfer experiences EPP with the club; however, for transfemoral amputees, the ability to know the spatial orientation of the knee joint of conventional designs is not possible because of the greater variability in possible angles and stiffnesses, thus limiting the user's perceived understanding of EPP. As a transfemoral amputee may experience, if the toe of the prosthetic foot catches on a carpet as they attempt to move the prosthesis through the swing phase of gait, the knee may be in excessive flexion just before heel strike, leading to a fall. Similarly, conventional prosthetic knees often cause abnormalities in gait on varied terrain, which the user has limited EPP of until it is too late.

The use of computer-controlled joints offer differences in perceived EPP, or perceived awareness of appropriate positioning of the limb in space, than with conventional systems. As the control software is able to better match the user's intended movement (through either CIC or IEC), the prosthesis will be able to better adapt to environmental changes and, therefore, enable proper biomechanical movement and hence perceived EPP (Figure 4). By not accommodating for environmental factors as effectively as computer-controlled systems, and hence having poorer biomechanical symmetry in varied environmental conditions, conventional prosthetics limit the perceived EPP of the user—the more intelligent control and more proper biomechanical movement of the prosthesis, the more the user will be able to gain confidence in the spatial orientation of the prosthesis correlated with the ambulated environments. Through offering more appropriate biomechanical movement of the device when compared with conventional designs, computer-controlled joints may “filter” both subtleties (stepping on a crack in the sidewalk) and major gait alteration changes (change in terrain slope for instance) during ambulation, allowing the user to ambulate unaffected, not compromising their perceived sense of EPP.

Newer, more advanced, neural integration strategies are being actively pursued to provide that critical efferent and afferent link between the brain and the microprocessor.23 This work will ultimately result in a better ability to control joint movement in daily activities. One example of such initiative is the Defense Advanced Research Projects Agency Revolutionizing Prosthetics 2009 design of a fully functional hand and arm system currently being developed to restore the function of persons who have experienced high-level upper limb amputation. Control of this prosthetic will involve options of targeted reinnervation, implantable sensors, pattern recognition, and peripheral nerve interface (Figure 5).23

It is possible for prosthetics to incorporate both CIC and IEC mechanisms. This is advantageous because it may allow for the user to have a certain level of control over the intended movement of the device (efferent IEC), communication with the device when available and necessary (afferent IEC feedback), and allows for the device to have intelligent control when no neural input is available (internal microprocessor CIC).

Ultimately, the user has to have full confidence in the control of the prosthetic device and have the prosthesis move corresponding to their intended movement, whether it is controlled by CIC or IEC methods. As the input sensor (CIC) and neural integration strategies (IEC) improve, the ability to enhance function of the prosthesis will increase.

Back to Top | Article Outline


Control strategies of these CIC or IEC input methods are arguably the most difficult technical barrier for the next generation of prosthetics. The movement of the limbs requires precise accommodation for a wide variance of factors, and the ability for the prosthesis to “think, respond, and react” to environmental changes based off the limited number of sensory and neural inputs is challenging.

What makes prosthetics increasingly difficult, when compared with purely robotic systems, is the meshing of man and machine. Although robotic devices have been able to achieve relatively natural bipedal gait, the human factor adds great complexity to the developmental process. Because there is a limited communication platform between the human and the prosthesis, there remains a development challenge to ensure that the “prosthetic brain” (microprocessor/control system) works in conjunction with the human brain (intended movement). Any inconsistencies in movement of the device, compared with the desired movement of the user, result in a disconnect and may lead to asymmetries, impaired balance and safety, and ultimately rejection of the device.

There are many variations in control strategies with the available CIC and IEC input methods. Different available computer-controlled prosthetic devices offer unique abilities and functional parameters (Figure 6). Some of these devices offer a wide range of altering dynamic characteristics, whereas others function within a narrow range of adjustability. In addition, these devices may offer resistive and powered actuation.

In the recent past, lower limb computer-controlled prosthetics have demonstrated greater success in their ability to better adapt to changes in the environment, compared with mechanical-based systems. In a clinical study comparing the two systems, Hafner et al.24 determined the microprocessor-controlled Otto Bock C-Leg to offer superior performance in stair and hill ambulation and lower incidence of stumbling and falling when compared with mechanical-based knees. Many of these computer-controlled systems have the ability to alter their state in angle and resistance to angular change by using sensor technology to provide information to a microprocessor, which in turn changes the resistive actuator to a large range of possible settings, limited only by the response time of the electronics and actuator. Some of these devices have the ability to variably accommodate for alterations of certain characteristics such as force, speed, and terrain changes in the gait cycle.

With the exception of not providing positive energy power to the foot or knee (concentric muscular contractions), this method provides better function during a wide range of ambulatory activities. It provides for a very large variability of angular change within the specified range of motion and a large variability of resistance to angular change. This corresponds to natural kinetic and kinematic requirements of ambulation such as force, speed, and terrain alterations (depending on what the control system accommodates for), at least in eccentric muscular contraction replication. However, this method does not allow for powered spatial orientation of joint placement, which may be essential for certain ambulation activities.

Specific joint angular placement is required to allow for the angular orientation of each joint to be positioned in space at the appropriate angle, independent of specific ambulatory activities, such as descending stairs in which increased plantar flexion angle is desired before foot contact.25 Resistive actuation devices merely restrict joint angular change. These systems may use alternative mechanical methods, such as spring elements or inertial effects, to provide spatial orientation, while using resistance settings of the actuator to control the amount and rate of motion (Figure 7).

Active powered devices may have the ability to provide positive energy into the stance and swing portions of the gait cycle. This acts to mimic concentric muscular contractions found during ambulation. A system such as this could enable a transfemoral amputee to climb stairs step over step, with the prosthesis putting positive energy into the ambulation, or could simply be used for spatial orientation of the prosthetic limb during swing phase of gait.26 For example, in a gait analysis study of the effects of a power knee prosthesis, Cutti et al.26 found that the powered device facilitates knee flexion at the start of the swing phase, promoting foot clearance during the step.

There is a significant functional difference between stance and swing phase powered actuation, but both provide their own benefits to the user.27 Powered actuation of an ankle-foot prosthesis enhances ambulation during stance by adapting joint resistance and providing positive energy, whereas powered actuation during swing enhances ambulation simply by actively controlling the orientation of the foot.27 Some devices may incorporate both stance and swing phase powered actuation.

The use of active powered prosthetics is the most true-to-life system possible, as they mimic each aspect of the anatomical limbs kinematic movement, including both eccentric and concentric muscular contractions. Although not readily implemented into the field, these systems have the potential to use a combination of resistive and active powered actuation to further enhance their power consumption efficiency and gait efficiency of the user. This is comparable with their anatomical counterpart, which uses both resistive and powered movements (concentric and eccentric muscular contractions).28

In considering computer-controlled lower limb products, the actuator technologies of the device used should be considered as well. Considerations for the actuator are weight, noise, size comparison with the anatomical envelope, power efficiency or battery lifespan, and durability. Each of these considerations is a challenge from a development standpoint and often competing requirements. Actuator technologies are also considered in their infancy in many regards. Smaller and more robust hydraulic valve designs are currently under development, for example, in devices such as hydraulic actuation prosthetic hands,29 whereas electric motors and drives are also being improved for use in designs, including powered ankle-foot prostheses.30 The hydraulic actuation artificial hand being developed by researchers at Oak Ridge National Laboratories incorporates very small and power efficient powered actuators that could someday be used in lower limb designs,29 whereas the powered ankle-foot prosthesis in the works at Massachusetts Institute of Technology is being designed to enhance gait symmetry, ambulation speed, and manage energy consumption by providing mechanical power in the stance phase of the gait cycle.30 In addition, state changing materials development is making great strides in becoming practical.31–33 Electroreactive polymers, in particular dielectric elastomers, are being studied by Kornbluh et al.32 for use as artificial muscles that can be implemented to actuate prosthetics someday. As each of these and other actuation technologies improve, they will enable lower limb computer-controlled designs to better mimic the anatomical limb and better meet the requirements and desires of their users.

In the past, mechanical-based systems were able to offer particular benefits to certain populations. Manual lock knees were well suited for lower function-level users who required extra stability, whereas fluid-based designs were generally used for higher functioning levels. With the advent of computer-controlled designs, there becomes a greater distinction between the benefits to the various types of users for this one category of systems (Figure 8). By varying system settings, a microprocessor-controlled knee may have valuable benefit to a broader spectrum of the population. In the past, simple, stable knees were prescribed to lower function users; today, active powered actuation knees may have a significant increase in the benefits to that same population—at least in its potential to offer greater safety and independence. A user who once had difficulty going from a sitting to standing position may now be more apt to perform that task on their own with a power-assist prosthetic. Ambulation potential, such as distance, ease, security, and gait efficiency, are all impacted by the use of intelligent control and actuation methods of the prosthesis.

Back to Top | Article Outline


Various microprocessor-controlled prosthetic designs offer accommodation for different gait characteristics. Although a product may be categorized by its ability to have sensor-based variability in its actuator adjustments—angle, resistance to angular change, powered actuations, etc—each product offers distinct abilities in performance. Some systems are designed to provide joint angle resistance alterations, according to the phase of the gait cycle only, whereas others may accommodate for terrain variances as well.

This is an important factor to consider in determining the most appropriate computer-controlled prosthetic for a given user. If the user rarely alters their gait speed, using a system that alters joint resistance settings for gait cycle phases only may be most appropriate. For users who frequently lift heavy loads, greatly alter gait speed, or walk on varied terrain, the ability of the device to accommodate for those factors should be considered in determining the most appropriate microprocessor-based components.

It is important that the device has the ability to offer real-time functional changes to the prosthetic joint's angle and resistance or actuation, as each step may be significantly different from the previous one. If, for instance, the actuation parameters are in conjunction with sensor-based information from previous steps, the angle and resistance of the joint may not be best suited for the immediate current step. Walking within a typical indoor environment at home or at the office for instance may provide largely even ambulated terrain environment; however, walking along a sidewalk, in the grass, or down a wheelchair ramp may provide significant enough alterations to the ambulated environment step to step, requiring immediate real-time accommodation of those environments. Those that do not may compromise safety and symmetry for the user.

Factors that should be considered are the ability of the device to accommodate for environmental factors (force, speed, and terrain), offer powered actuation in stance, and swing phase (spatial orientation), have dynamic cosmetic effects (such as plantarflexion of the foot during sitting), ability to offer real-time control, and offer a control input method (IEC or CIC) that is well suited for the user.

Inevitably, many advancements will be made and new products will emerge over these next few years in lower limb computer-controlled prosthetics, resulting in vast changes in the way these systems are viewed and used. As new products, exhibiting the many different characteristics of computer-controlled lower limb prosthetics continue to emerge, a greater understanding of how to improve the next generation of prosthetics will surface.

Back to Top | Article Outline


1. Vickers DR, Palk C, McIntosh AS, Beatty KT. Elderly unilateral transtibial amputee gait on an inclined walkway: a biomechanical analysis. Gait Posture 2008;27:518–529.
2. Bateni H, Olney SJ. Kinematic and kinetic variations of below-knee amputee gait. J Prosthet Orthot 2002;14:2–12.
3. Perry J, Boyd LA, Rao SS, Mulroy SJ. Prosthetic weight acceptance mechanics in transtibial amputees wearing the single axis, seattle lite, and flex foot. IEEE Trans Rehabil Eng 1997;5:283–289.
4. Powers CM, Rao S, Perry J. Knee kinetics in trans-tibial amputee gait. Gait Posture 1998;8:1–7.
5. Sanderson DJ, Martin PE. Lower extremity kinematic and kinetic adaptation in unilateral below-knee amputees during walking. Gait Posture 1997;6:126–136.
6. McIntosh AS, Beatty KT, Dwan LN, Vickers DR. Gait dynamics on an inclined walkway. J Biomech 2006;39:2491–2502.
7. Vrieling AH, van Keeken HG, Schoppen T, et al. Gait initiation in lower limb amputees. Gait Posture 2008;27:423–430.
8. Jones SF, Twigg PC, Scally AJ, Buckley JG. The mechanics of landing when stepping down in unilateral lower-limb amputees. Clin Biomech 2006;21:184–193.
9. Waters RL, Perry J, Antonelli D, Hislop H. Energy cost of walking of amputees: the influence of level of amputation. J Bone Joint Surg Am 1976;68:42–46.
10. Nolan L, Wit A, Dudziñski K, et al. Adjustments in gait symmetry with walking speed in trans-femoral and trans-tibial amputees. Gait Posture 2003;17:142–151.
11. Vrieling AH, van Keeken HG, Schoppen T, et al. Uphill and downhill walking in unilateral lower limb amputees. Gait Posture 2008;28:235–242.
12. Fan RE, Culjat MO, King CH, et al. A haptic feedback system for lower-limb prostheses. IEEE Trans Neural Syst Rehabil Eng 2008;16:270–277.
13. Kaufman KR, Levine JA, Brey RH, et al. Gait and balance of transfemoral amputees using passive mechanical and microprocessor-controlled prosthetic knees. Gait Posture 2007;26:489–493.
14. Schmalz T, Blumentritt S, Jarasch R. Energy expenditure and biomechanical characteristics of lower limb amputee gait: the influence of prosthetic alignment and different prosthetic components. Gait Posture 2002;16:255–263.
15. Segal AD, Orendurff MS, Klute GK, et al. Kinematic and kinetic comparisons of transfemoral amputee gait using C-Leg and Mauch SNS prosthetic knees. J Rehabil Res Dev 2006;43:857–870.
16. Johansson JL, Sherrill DM, Riley PO, et al. A clinical comparison of variable-damping and mechanically passive prosthetic knee devices. Am J Phys Med Rehabil 2005;84:563–575.
17. Schärer C. Trainable command recognition for a microprocessor-controlled prosthetic arm. J Microcomput Appl 1993;16:287–292.
18. Parker P, Englehart K, Hudgins B. Myoelectric signal processing for control of powered limb prostheses. J Electromyogr Kines 2006;16:541–548.
19. Ryait HS, Arora AS, Agarwal R. Study of issues in the development of surface EMG controlled human hand. J Mater Sci: Mater Med 2009;20:S107–S114.
20. Lake C, Miguelez JM. Evolution of microprocessor based control systems in upper extremity prosthetics. Tech Disabil 2003;15:63–71.
21. Sainburg RL, Ghilardi MF, Poizner H, Ghez C. Control of limb dynamics in normal subjects and patients without proprioception. J Neurophysiol 1995;73:820–835.
22. Childress DS. Control strategy for upper-limb prostheses. Conf Proc IEEE Eng Med Biol Soc 1998;1998:2273–2275.
23. Weir R, Mitchell M, Clark S, et al. New multifuctional prosthetic arm and hand systems. Conf Proc IEEE Eng Med Biol Soc 2007;2007:4359–4360.
24. Hafner BJ, Willingham LL, Buell NC, et al. Evaluation of function, performance, and preference as transfemoral amputees transition from mechanical to microprocessor control of the prosthetic knee. Arch Phys Med Rehabil 2007;88:207–217.
25. Riener R, Rabuffetti M, Frigo C. Stair ascent and descent at different inclinations. Gait Posture 2002;15:32–44.
26. Cutti AG, Raggi M, Garofalo P, et al. The effects of the ‘Power Knee’ prosthesis on amputees metabolic cost of walking and symmetry of gait—preliminary results. Gait Posture 2008;28:S38.
27. Au S, Berniker M, Herr H. Powered ankle-foot prosthesis to assist level-ground and stair-descent gaits. Neural Netw 2008;21:654–666.
28. DeVita P, Helseth J, Hortobagyi T. Muscles do more positive than negative work in human locomotion. J Exp Biol 2007;210:3361–3373.
29. Cabage B. Prosthetic promise. Oak Ridge National Laboratory Reporter, January–February 2009:1, 6.
30. Au SK, Herr H, Weber J, Martinez-Villalpando EC. Powered ankle-foot prosthesis for the improvement of amputee ambulation. Conf Proc IEEE Eng Med Biol Soc 2007;2007:3020–3026.
31. Herr H, Kornbluh R. New horizons for orthotic and prosthetic technology: artificial muscle for ambulation. Proc SPIE 2004;5385:1–9.
32. Kornbluh R, Bashkin J, Pelrine R, et al. Medical applications of new electroactive polymer artificial muscles. J JSPP 2004;16:631–637.
33. Ashley S. Artificial muscles. Sci Am 2003;289:52–59.

KEY INDEXING TERMS: lower limb prosthetic; microprocessor control; input method; control system; output actuation

Figure. No caption a...
© 2010 American Academy of Orthotists & Prosthetists