June 2008 • Vol. 4, No. 3
	Advancing Orthotic and Prosthetic Care Through Knowledge

Focus on Upper Extremity:
   Future of Upper-Limb Design:
   Clinical Perspectives of the DARPA RP09 Program

Jay Martin, BS, CP, LP

Abstract

Conventional high-level upper-limb prostheticsXcan offer*-three to four isolated degrees of freedom with limiteacontrol options and feedback mechanisms. Compared to the anatomical counterpart with 28 degrees of freedom, full active neural control, and sensory feedback, there remains significant room for innovation.

With recent federal prosthetics funding increases, the prosthetics field is entering a Renaissance in technological advancements. The Defense Advanced Research Projects Agency Revolutionizing Prosthetics 2009 (DARPA RP09) program, a $30.4-million initiative to advance upper-limb prosthetics design, is reshaping the limits of what prosthetics can do.

Made up of a team of leading scientists and engineers from numerous organizations, the DARPA RP09 program seeks to restore full functionality to upper-limb amputees, closely resembling the anatomical counterpart. Using peripheral and cortical nerve interface technologies, pattern recognition, and targeted nerve reinnervation, control of the program's 28 degrees of freedom upper-limb prosthesis will offer new abilities to amputees for living and working with their prosthesis.

With the completion of Phase I objectives, radical advancements have already been achieved. This paper will discuss the program's objective and challenges as they progress into their Phase II efforts.

Keywords: Upper-limb prosthetics, engineering, control strategies, interface design, design challenges

Future of Upper-Limb Design:
Clinical Perspectives of the DARPA RP09 Program

Conventional high-level upper-limb prostheses can offer three to four isolated degrees of freedom, with limited control options and feedback mechanisms. Compared to the anatomical counterpart, with 28 degrees of freedom, full active neural control, and sensory feedback, there remains significant room for innovation.

New advancements in engineering technology have resulted enabling technologies for prosthetic development. In years past, advanced prosthetic developments have been limited by the need for further advancements in technologies such as actuator design, computing power, electronics complexity, and neural connectivity to enable certain milestones to be met in function and control of advanced upper-limb prosthetics.

Overall Requirements of the DARPA RP09 Program

The DARPA RP09 program, a $30.4-million initiative to advance upper-limb prosthetics, is reshaping the limits of prosthetics design.

Made up of a team of leading scientists and engineers from numerous organizations, the DARPA RP09 program seeks to restore full functionality to upper-limb amputees, closely resembling the anatomical counterpart. Using peripheral and cortical nerve interface technologies, targeted nerve reinnervation, and pattern recognition strategies, control of the program's 28 degrees of freedom upper-limb prosthesis will offer new abilities to amputees for living and working with their prosthesis.

The DARPA RP09 program has called for far-reaching design goals that push the limits of what even the most up-to-date engineering technologies will allow. A full shoulder disarticulation-level prosthetic device is desired, with full anatomical range of motion, degrees of freedom, neural afferent and efferent communication, fitting within the anatomical envelope, weight comparable to the anatomical limb, and being cosmetically appropriate. As one can imagine, many of these listed requirements are seemingly competing. Further, the engineering task of successfully accomplishing such a far-reaching set of goals is challenging.

Not only are the engineering goals of such a program challenging, but making a full robotic arm system practical for actual clinical prosthetics application adds an entirely new level of complexity to the design. Many research labs in the past have successfully created robotic arms for various purposes, with dexterity and strength rivaling the anatomical counterpart. Incorporating those designs to mesh with the human body in a clinically practical way is very different.

Developers on the RP09 program have focused their efforts closely on the listed requirements to help ensure that the final deliverable will be suitable for practical clinical application. These requirements very much drive the direction of innovation from every angle—electronics, mechanical design, control strategies, materials, and many others.

TABLE 1

Advanced Actuators, Electronics, and Software

The human body relies on its "electronics and software" equivalent to control the movement of the mechanical system (bones, muscles, ligaments). In many conventional upper-limb prosthetics, the movement of the device is either single-joint control or passive.

Creating multi-joint systems whose movement is initiated by intelligent control—through autonomous, sensor-based, or neural integration strategies—stands to radically affect functional outcomes. On the RP09 program, a number of actuator-based joint system platforms are being developed to be used through a combination of autonomous, sensor-based, and neural control. Each of these various platforms, including technologies ranging from powered micro-hydraulics, powered pneumatics, motors/gears/tendons, etc., have a unique set of advantages and disadvantages for practical clinical application. Each of these designs face similar challenges of fitting within the anatomical envelope, maintaining sufficient strength while keeping power consumption to a minimum, maintaining durability, and providing adequate resolution of control. There may be no "perfect" actuator-based joint system platform, or prosthetic design for that matter, but rather a tradeoff in competing requirements is carefully being considered in creating a design that may be optimal for practical clinical outcomes. The development of each of the actuator methods under consideration has its own subset of challenges and competing requirements as well. Creating a system with such a large number of active degrees of motion, fitting within the anatomical envelope, and having strength comparable to the anatomical counterpart is very challenging. Additionally, a number of other technical requirements can be found in table 1. Each of these requirements must be adequately addressed in order to ensure practical clinical application for this program's far-reaching goals.

In general, competing requirements for development of the actuator and joint system platforms include strength versus weight, space to fit within the anatomical envelope, precision of control, number of actuators required to provide necessary degrees of freedom versus degrees of motion, actuator attachment means (i.e., tendons versus gears), and length of space required (affecting its use for various levels of amputation), among many others.

Conventional hands, for instance, typically offer one to five active degrees of motion (DOM), and one to ten degrees of freedom (DOF) within the anatomical size envelope.^1,2 The RP09 program is pursuing designs with 11 to 15 active DOM and 19 to 21 DOF within similar size constraints. Adding these additional DOM and DOF is especially challenging considering that weight, space, and durability cannot be compromised in comparison to conventional designs. Adding these extra DOM and DOF does, however, stand to provide dexterity comparable to the anatomical limb. The challenge, however, is controlling such a system.

In conventional upper-limb myoelectric prosthetics, the movement of the device or system still relies on a very limited amount of available information from the user for intended movement—typically single joint movement such as "hand open," "hand close," or "elbow flexion," while other joints' movements may solely be passive. This control is often initiated through at most two- to three-site myoelectric sensors, with possible inclusion of other switch mechanisms—all using a single-joint movement at a time. While a considerable amount of research has gone into better enabling prosthetics to be controlled more effectively, many engineering limitations have prohibited those developments from being applied clinically to their full potential.

Over the years, many researchers have pushed the limits of prosthetic control strategies, but seeing their efforts come to commercialization have often been stifled by the practical application capabilities of their base technology. Computing power, in early pattern-recognition research, for instance, required large computers that were impractical for mobility requirements of prosthetics users. Today, that same computing power requirement can be incorporated into a small microprocessor-based board. Electronics design overall has largely progressed, allowing complicated electronics design that once took a considerable amount of space and weight to now be fit onto small circuit boards.

The electronics designs incorporated into the prototype limbs on the RP09 program are very advanced. Creatively integrating a vast amount of complicated circuitry and wiring into the multi-joint limb is challenging. In a system capable of 18 to 21 active degrees of motion (with 25-28 degrees of freedom) with multi-joint simultaneous control, sensory feedback, and sensor-based control, wiring alone requires creative and hightech solutions—all affecting weight, durability, and space considerations.

Figure 1

Neural Integration Considerations

As these mechanical, computer, and electronics designs on the prototype limbs have progressed enough for practical prototype testing, one of the main challenges remaining is further neural connectivity. As one of the driving forces of the DARPA RP09 program, neural integration is a key component in allowing for greater function of the next generation of upper-limb prosthesis.

Without more advanced neural connectivity between the brain and the prosthesis to initiate intended control, a multi-joint system has little practical application. Conversely, as the neural connectivity of the brain with the prosthesis is enhanced, the need for a multi-joint controlled prosthesis will be greater. The user of the prosthetic device must have the ability to control the multiple degrees of freedom. At the same time, the need for sensory feedback (sense of touch, proprioception, etc.) may be far greater when using a system with capabilities of multi-joint simultaneous control.

Full cortical and/or peripheral implanted devices are actively being investigated. These strategies stand to provide a near-seamless integration for control between the human brain and the prosthesis—and the electromechanical limbs being developed will be suitable for achieving an anatomically similar level of control—hence 28 degrees of freedom.

Until those strategies are completed, alternative or possibly complementary approaches are actively being investigated as well. By using pattern recognition or targeted nerve reinnervation strategies, a far greater level of control of the prosthetic device is possible. Even enabling for three to four joints to be simultaneously controlled creates enormous benefit to the user through greatly enhance dexterity and functionality, especially when sensory feedback is provided as well.^3,4 At the same time, however, this also creates enormous challenges for socket interface integration never before faced.

Socket Interface Requirements and Challenges

Figure 2

Having a prosthetic limb system that is capable of multi-joint simultaneous control and strength capabilities comparable to that of an adult male offers new challenges to the socket interface environment that are drastically different from conventional design.

In conventional socket interface design, the loads experienced on a high-level upper-limb prosthetic user are relatively low. Conventional elbow systems, for instance, have the ability to lift up to six kilograms.⁵ Requirements for the developments on the DARPA RP09 program, however, call for an elbow to lift 27 kilograms. This, in combination with an actively powered shoulder joint with simultaneous movement with the elbow, further compounds the issue.

If a high-level upper-limb prosthetic user is able to control such an arm effectively (neural integration, targeted nerve reinnervation, etc.), the forces experienced on the prosthesis and resulting forces on the residual limb or torso will be far more complex than with conventional designs. Users will no longer be experiencing predominantly static forces, but rather will be experiencing many dynamic forces, many of which will be larger than the static load limits currently experienced with conventional interface designs.

Conventional interface systems for a shoulder disarticulation prosthetics user typically can allow for roughly 50 pounds of static in-line gravitational force. On a clinical basis, many of these users have demonstrated the ability to lift heavy loads, but these loads are largely hung axially from the body; they are not actually being lifted dynamically by the prosthesis through various planes. Loads that are carried by powered actuation of conventional prosthetic designs, such as an elbow joint lifting an object, are relatively low force due to the prosthetic elbow components' inherent strength limitations. Further, there are currently no electronic shoulder joints used on a consumer clinical basis that work simultaneously in conjunction with the elbow. All movements within a conventional upper-limb device are also at one joint at a time—greatly simplifying the load dynamics.

With the incorporation of this new generation of prosthetic components and systems capable of multi-joint simultaneous control, such as on the DARPA RP09 program, the loads experienced on the body will be significantly different. Further, while unknown at this point, the incorporation of neural control strategies and components with greater functionality may likely lead to increased use of the prosthesis—additionally affecting the interface dynamics with the body.

New management of interface forces is required. Managing these new loads on the body must be met with comfort and control, and be cosmetically appropriate. Many creative approaches have been investigated on the RP09 program that attempt to address these unique needs.

For shoulder disarticulation-level amputations in particular, the torso takes up the majority of the weight and forces of the prosthesis. Because these loads are so great, it is important to disperse these loads over as great an area as possible. This portion of the body in particular has many sensitive areas that cannot incur these new dynamic forces. Further, since this part of the body has so much range of motion and movement (torso anterior, posterior, and lateral bending, and inhalation and exhalation), it is important to allow for dynamic movement of the interface while maintaining a solid connection between the body and prosthesis.

Using Compliant Force Dissipation Techniques, these seemingly competing requirements can be met. This is accomplished by providing an interface frame that is relatively low profile and specifically contoured to help maintain the position of the interface in correct relation to the underlying anatomy. The frame spans from the prosthetic shoulder attachment point medially and distally to provide anterior and posterior stability about the torso as dynamic loads are experienced. However, the predominant force distribution of the interface is not taken up by the frame itself, but rather the Compliant Force Dissipation Garment. This garment, much like the axilla strap in conventional shoulder-level prostheses, takes up much of the load; however, instead of hanging the load of the prosthesis off one strap, this garment is made up of thousands of tiny straps that encapsulate the entire torso. The result is a broad load distribution around as great of an area as possible—including about the prosthetic side torso through posting mechanisms built into the garment. As an amputee holds an object with his prosthetic hand and then travels that object through space using shoulder and elbow joints in dynamic movement, the resultant force on the torso is quite large and rapidly changing about various aspects of the torso.

Since each of these tiny micro straps carries a very small load and is extremely flexible, the perceived and experienced load on any one part of the residuum is very low. In effect, the use of the prosthesis may become more comfortable than with conventional designs despite the greatly increased load on the body with the use of more capable components.

Figure 3

From a cosmetic standpoint, using this ultra-low-profile design adds to a more cosmetically appealing system. With current prototypes, the interface and its associated attachment system are no more than three millimeters thick at any one point. With the slight soft tissue compression of the interface about the body, there is typically no more than a one- to two-millimeter extension outside of the body's envelope. Additionally, the shoulder disarticulation level socket system including attachment garment weighs less than 400 grams—staying within the RP09 spec requirements.

The incorporation of various nanotechnologies into the socket environment additionally helps to provide a more suitable environment for user comfort, hygiene, and durability.

Conclusion

The rapid progression of engineering design in recent years has opened the door for prosthetics evolution. Today's designs have allowed for prosthetics technology to hit new milestones, resulting in a whole different set of development requirements and challenges.

Two years into the program, the DARPA RP09 team has conquered many of the challenges and attained many of the program's goals in prototype form. While much work remains before the program's development culmination in 2009, and further clinical testing begins, many milestones in prosthetic engineering development have already been met, providing prototype limbs capable of multi-joint simultaneous control, significant strength capabilities, and dexterity comparable to that of an intact limb.

The future of upper-limb prosthetics is changing—and will continue to change—very rapidly. Designs in research labs today will soon have a major impact on the upper-limb amputee community. As this takes place, practitioners will be faced with requiring new approaches to fitting—enabling comfortable use of such capable systems. As enabling technologies continue to shape how far prosthetics design can be pushed, the end users will ultimately benefit through increased function, dexterity, and control—rivaling the abilities of the anatomical arm.

Acknowledgments

Johns Hopkins University Applied Physics Laboratory DARPA Revolutionizing Prosthetics Management team DARPA Revolutionizing Prosthetics Development team

References

1. Touch Bionics. The i-Limb Hand. 2008; www.touchbionics.com

2. Otto Bock. Sensor Hand Speed. 2008; www.ottobock.com

3. Kuiken, T.A.; Marasco, P.D.; Lock, B.; Harden, R.N.; Dewald, JPA "Redirection of Cutaneous Sensation from the Hand to the Chest Skin of Human Amputees with Targeted Reinnervation," PNAS, 104(50):20061-20066, Dec 2007.

4. Miller, L.A.; Stubblefield, K.A.; Lipschutz, R.D.; Lock, B.A.; Kuiken, T.A. "Improved Myoelectric Prosthesis Control Using Targeted Reinnervation Surgery: A Case Series," IEEE Trans Neural Sys Rehab Eng., 16(l):46-50, Feb 2008.

5. Otto Bock. Dynamic Arm. 2008; www.ottobock.com

Joy Martin, BS, CP, LP, is the director of Advanced Systems Group, OrthoCare Innovations LLC, and founder of Martin Bionics LLC Send correspondence to Jay Martin, CP, LP; 800 Research Parkway, Suite 395; Oklahoma City, 0K73104;phone: 405.2712466 ext. 33002; fax: 405.271.2467; e-mail: jaymartin@aol.com

This project is funded under the DARPA Revolutionizing Prosthetics 2009 program in conjunction with the Johns Hopkins University Applied Physics Laboratory.

COMING SOON

Earn PCE Credits from this article