1240 Part F Field and Service Robotics 53.4 Smart Prostheses and Orthoses 53.4.1 Grand Challenges and Roadblocks 53.4.2 Targeted Reinnervation In 2005,the Defense Sciences Office (DSO)of the Standard prosthetic arms and hands are commonly con- US governmental research agency,the Defense Ad- trolled with a cable drive or by EMG signals from vanced Research Projects Agency (DARPA)launched residual muscles.For example,to open and close an arti- a program to revolutionize prosthetics in a four ficial hand,one common technique is to place a Bowden year timeframe.According to the agency website cable around the shoulders in a harness,and connect the (http://www.darpa.mil/dso/thrust/biosci/revprost.htm),cable directly to the artificial hand.The user can then this program will shrug the shoulders to move the cable and open and deliver a prosthetic arm for clinical trials that is close the hand.Alternately,electrodes can be placed on far more advanced than any currently available. a muscle in the residual limb or on the user's back,for This device will enable many degrees of freedom example,and then used to control a motor on the ar- for grasping and other hand functions,and will be tificial hand.The cable technique has the advantages rugged and resilient to environmental factors.In four of simplicity,and of having the property of extended years,DSO will deliver a prosthetic for clinical tri- physiological proprioception(EPP),which refers to the als that has function almost identical to a natural fact that the grip force is mechanically transmitted back limb in terms of motor control and dexterity,sen- to the user's shoulder muscle force sensors so that the Part user can gauge the strength of the grasp.Because of sory feedback (including proprioception),weight, and environmental resilience.The four-year device their simplicity and EPP,cable drives(or body-powered will be directly controlled by neural signals.The prostheses)have been more popular than myoelectric 53 results of this program will allow upper limb am- (or externally powered)prostheses.However,the body- powered technique is amenable to controlling only one putees to have as normal a life as possible despite degree of freedom at a time,although chin switches, their severe injuries. for example,can be used to switch between degrees of This program announcement lays down the grand chal- freedom in a somewhat cumbersome way.The myoelec- lenge for prosthetics research in an ambitious timeframe: tric approach can be used to control multiple degrees of develop an artificial limb that has functionality and freedom.but such control is nonintuitive and cumber- durability at least as good as a natural limb. some.Also,multiple control sites for reading out EMG There are several roadblocks to meeting this chal- are not available for people who have lost their entire lenge.First,providing an intuitive way for individuals arm.Thus,prosthetic control systems are typically lim- to control and coordinate multiple joints of a robotic ited to one or two degrees of freedom,while functional limb is challenging.Second,robots do not yet match the arm and hand movement benefits from at least four de- human arm in terms of the combination of range of force, grees of freedom (three to position the hand,and one to weight,and duration of use with a portable power source. open and close it). Third,human limbs are rich with tactile and movement Kuiken et al.[53.132]recently developed a novel sensors.Installing artificial sensors on a robotic limb,approach to improving control of a multijointed pros- and then returning information from those sensors in thetic arm.In this targeted reinnervation technique,they a way that is usable by the user is challenging.Thus, rerouted the nerves that previously innervated the lost solving the grand challenge will require better sensory- limb to a spared muscle,and then read out the user's motor interfaces for prosthetic limbs,as well as lighter intent to move the limb using electromyography at the stronger actuators and better power sources. spared muscle.They demonstrated this technique in a bi- Substantial progress has recently been made in im-lateral shoulder disarticulation amputee who had lost proving sensory-motor interfaces for prosthetic limbs,both of his arms in an electrical power accident.They and this progress is the focus of this section.For the took the residual brachial plexus nerve for the left arm, current state of robotic actuators that could be used in which normally innervates the left elbow,wrist,and prosthetic devices,the reader is referred to Chap.62 on hand,and moved it to the pectoralis muscle.The subject neurorobotics.For an overview of the design of conven- could still contract his pectoralis muscle,but this mus- tional prosthetic hands and arms.,the reader is referred cle was no longer useful to him since it used to attach to[53.131]. to his now-missing humerus.A surgeon dissected por-1240 Part F Field and Service Robotics 53.4 Smart Prostheses and Orthoses 53.4.1 Grand Challenges and Roadblocks In 2005, the Defense Sciences Office (DSO) of the US governmental research agency, the Defense Ad￾vanced Research Projects Agency (DARPA) launched a program to revolutionize prosthetics in a four year timeframe. According to the agency website (http://www.darpa.mil/dso/thrust/biosci/revprost.htm), this program will deliver a prosthetic arm for clinical trials that is far more advanced than any currently available. This device will enable many degrees of freedom for grasping and other hand functions, and will be rugged and resilient to environmental factors. In four years, DSO will deliver a prosthetic for clinical tri￾als that has function almost identical to a natural limb in terms of motor control and dexterity, sen￾sory feedback (including proprioception), weight, and environmental resilience. The four-year device will be directly controlled by neural signals. The results of this program will allow upper limb am￾putees to have as normal a life as possible despite their severe injuries. This program announcement lays down the grand chal￾lenge for prosthetics research in an ambitious timeframe: develop an artificial limb that has functionality and durability at least as good as a natural limb. There are several roadblocks to meeting this chal￾lenge. First, providing an intuitive way for individuals to control and coordinate multiple joints of a robotic limb is challenging. Second, robots do not yet match the human arm in terms of the combination of range of force, weight, and duration of use with a portable power source. Third, human limbs are rich with tactile and movement sensors. Installing artificial sensors on a robotic limb, and then returning information from those sensors in a way that is usable by the user is challenging. Thus, solving the grand challenge will require better sensory￾motor interfaces for prosthetic limbs, as well as lighter stronger actuators and better power sources. Substantial progress has recently been made in im￾proving sensory-motor interfaces for prosthetic limbs, and this progress is the focus of this section. For the current state of robotic actuators that could be used in prosthetic devices, the reader is referred to Chap. 62 on neurorobotics. For an overview of the design of conven￾tional prosthetic hands and arms, the reader is referred to [53.131]. 53.4.2 Targeted Reinnervation Standard prosthetic arms and hands are commonly con￾trolled with a cable drive or by EMG signals from residual muscles. For example, to open and close an arti- ficial hand, one common technique is to place a Bowden cable around the shoulders in a harness, and connect the cable directly to the artificial hand. The user can then shrug the shoulders to move the cable and open and close the hand. Alternately, electrodes can be placed on a muscle in the residual limb or on the user’s back, for example, and then used to control a motor on the ar￾tificial hand. The cable technique has the advantages of simplicity, and of having the property of extended physiological proprioception (EPP), which refers to the fact that the grip force is mechanically transmitted back to the user’s shoulder muscle force sensors so that the user can gauge the strength of the grasp. Because of their simplicity and EPP, cable drives (or body-powered prostheses) have been more popular than myoelectric (or externally powered) prostheses. However, the body￾powered technique is amenable to controlling only one degree of freedom at a time, although chin switches, for example, can be used to switch between degrees of freedom in a somewhat cumbersome way. The myoelec￾tric approach can be used to control multiple degrees of freedom, but such control is nonintuitive and cumber￾some. Also, multiple control sites for reading out EMG are not available for people who have lost their entire arm. Thus, prosthetic control systems are typically lim￾ited to one or two degrees of freedom, while functional arm and hand movement benefits from at least four de￾grees of freedom (three to position the hand, and one to open and close it). Kuiken et al. [53.132] recently developed a novel approach to improving control of a multijointed pros￾thetic arm. In this targeted reinnervation technique, they rerouted the nerves that previously innervated the lost limb to a spared muscle, and then read out the user’s intent to move the limb using electromyography at the spared muscle. They demonstrated this technique in a bi￾lateral shoulder disarticulation amputee who had lost both of his arms in an electrical power accident. They took the residual brachial plexus nerve for the left arm, which normally innervates the left elbow, wrist, and hand, and moved it to the pectoralis muscle. The subject could still contract his pectoralis muscle, but this mus￾cle was no longer useful to him since it used to attach to his now-missing humerus. A surgeon dissected por￾Part F 53.4