The complexity of upper extremity control of the paralyzed patient offers a great challenge to the rehabilitation engineer, orthotist/prosthetist, or others. However, researchers can approach the problem of disabled arm motion, for the sake of simplicity, as a series of links with minimum pivots and minimum gravitational forces. In addition, when the degrees of freedom are reduced from four to three as in the case of this study, then some daily functions can be achieved by using simple braces and splints that are powered externally and internally.
In our design of a hybrid arm orthosis (HAG), we emphasized simplicity and active contribution of the user. The orthosis takes full advantage of the wheelchair frame without changing its profile. The HAG offers body powered shoulder abduction and elbow flexion of one degree of freedom. A motor-powered wrist supination and three-joint jaw chuck pinch offer the remaining two degrees of freedom.
The target population for this device consists of patients who are wheelchair bound due to post-poliomyelitis, high level spinal cord injury (SCI), or stroke. This population is one whose needs are the most difficult to accommodate when it comes to fitting an external device (orthosis) that will assist in performing the activities of daily living. This target population, once limited in number, has recently increased due to improved intensive care and surgical methods. Once the patient survives, nursing care becomes necessary due to the complete dependence and need for continuous assistance of the majority of high level (C3-5) quadriplegics. While low level (C6-8) quadriplegics are rarely in need of orthoses to provide useful upper extremity function because of residual hand/ arm control, C3-4 patients with the aid of orthotics may become more productive, motivated, and less dependent on nursing care. A 1965 review by Chyatte et al of four patients with muscle disease compared performance in ADL with and without orthoses. The orthosis used was a mobile arm support (MAS). 1 They reported that performance in self-hygiene, grooming, feeding, and diversional activities improved with MAS, and concluded that the orthosis, when properly selected and adjusted, could increase the functional capacity of muscle disease patients.
Current HAG designs most often take advantage of the wheelchair to minimize the hardware attached to the patient, and to conveniently house power sources and actuators. These devices typically employ a control system that uses a joy stick,2,3 sip and puff switch, tongue switches,4 biological (myoclectric) signals,5 or residual motion of the patient's limbs.6
While the literature shows that current HAG designs are adequate, an acceptable control system has not yet been found and is central to the issue of maximum compliance by the user. The literature also states that, as with any adaptive device, simplicity and cosmetics dictate immediate and long-term acceptance of an HAG by the patient, and still need to be improved upon. The researcher maintains that recent advances in the industrial development of thermo-plastic materials, miniaturized power sources, and micro-processor controls encourage fresh, new attempts in the design and control of HAGs.
The purpose of this paper is to describe a novel HAG design for C3-4 patients that restores hand/arm functions. The proposed solution aims to integrate a patient's upper extremity movement with rehabilitation intervention and to withstand the test of time and use by the patient.
The proposed HAG achieves two major functions by using two different power sources. The shoulder and elbow joints are interconnected and simultaneously abduct and flex, respectively, by contralateral shoulder elevation. The wrist supination and three-point jaw chuck pinch is generated by two separate switchable DC motors in sequence. This section details the materials and calculations used to achieve the four de sired motions of the HAG: shoulder abduction, elbow flexion, wrist pronation/supination, and finger prehension.
Figure 1A and Figure 1B F shows a single pivot mechanical shoulder joint mounted on the upper frame of the wheelchair. This pivot houses a torque spring (7.5 in*lb. at 180 degrees). The center of the pivot is adjustable so that it can be aligned with the center of rotation of the patient's shoulder joint. Attached to the pivot bearing is an upper arm support bar (UASB), which is motivated with a cord. A motivating cord is attached to the UASB at an extension (lever arm) 3" above the pivot center. This lever arm rotates 450 with respect to the vertical when the upper arm is at rest along the body (Figure 2) . The motivating cord travels through a system of pulleys and a compression spring on the contralateral side of the target arm to minimize the force needed to elevate the motivating shoulder (Figure 3A and Figure 3B ).
The following formula summarizes the torque analysis of the pulley system.
T [2Fs + Kc(0.051) (45-RGM)] 3SIN(THETA) + [(110-RGM)/180] Kt
where Kc is the compression spring constant (6.2 lb.*in) and (0.051) (45-ROM) is its deflection over a 450 range of motion (65-110). The total displacement generated is 2.31" or 0.051"/° (2.31"/45°). RGM is the range of motion 1,2,3, . 45° and 3" is the extension (lever arm) length of UASB. Kt is the torque spring constant. Adding the shoulder force 2Fs, the left portion of the formula () describes the tension in the cable. The remaining part of the formula describes the contribution that is due to the added torque from the torque spring located at pivot, assuming an initial position of 300 that corresponds to an arm along the body, that is, resting position. Once the torque due to external forces is calculated, the required Es can be solved, as follows:
Fs= T/6 (SIN(THETA)) - (110-RGM/180) Kt/6SIN (THETA) - Kc(0.051) (45-RGM)/2
Adapting from Dempster (1955), the location of average centers of gravity and the weights and lengths of body segments, we generated a graph that represents the optimal required shoulder torque vs. abduction angle (theta) (Figure 4) . Because the force characteristics of both abduction and torque spring have negative linear slopes with respect to theta, we matched the required shoulder torque curve over theta for a given shoulder displacement by varying the spring constants, loading the torque spring and varying the lever arm length.
We assumed no energy loss due to inertia force or friction. The elevation force produced by the motivating shoulder transfers to the shoulder pad that is attached to a shoulder lever arm. The motivating shoulder lever arm travels along a shaft via a linear bearing and pulls the mobile pulley block which also travels along a shaft on a linear bearing (Figure 3) . The compression spring positioned between the base of the shaft and the mobile pulley block assists this pulley unit. The compression spring initially is chosen to balance some of the gravitational forces due to the weight of the whole arm at rest. However, we also wanted to take advantage of the stored energy in the spring, to aid the patient in abduction. Once the mobile pulley system moves upward it pulls the cable attached to it, which travels through the lower and upper fixed pulleys on its way to the lever arm above the UASB. For 2" of shoulder elevation the patient generates 380 of horizontal abduction of the upper arm.
A push-pull cable, on one end attached to the upper seat bar at 4.5 cm away from the center of shoulder (Figure 2a and Figure 2b ), and on the other end fixed to the lower arm support bar (LASB) 5 cm away from the elbow joint (Figure 5a and Figure 5b ), offers a simultaneous flexion in relation to shoulder abduction. The LASB is attached to a custom made elbow support positioned at the end of the UASB (Figure 2a and Figure 2b ). As for shoulder abduction, we used a torque spring (3.5 lb.*in at 1800) inside the bearing block of this joint. We found the contribution of flexion force to overall external forces that act on the shoulder to be negligible. The torque spring works to balance the weight of the lower arm in the main activation range and to assist in flexion as its stored energy is released.
A miniature 6V DC motor (Portescap) drives the spur gear fitted to a wrist brace (Figure 5a and Figure 5b ). The location of the motor and the design permit normal, unrestricted pronation/supination at the rate of 30°/sec. The motor is powered by the wheelchair battery, together with the control circuitry. The control circuitry consists of CMOS logic gates and a current feedback motor controller. The logic gates are switched to activate the motor by air pressure sensitive relays. The relays are controlled by three air buttons fixed on the head rest of the wheelchair (Figure 6) . To activate the pronation/supination motor the patient slightly presses the center button (>12.1 mm Hg) which powers the control circuitry, then depresses one of the more sensitive buttons (>3.7 mm Hg) at the side of the central button to activate the selected motor. Each consecutive depression of the button changes the direction of the motor.
A second miniature 6V DC motor located on the Velcro™ strap that secures the lower arm to the LASB, energizes the prehension gearbox (Figure 5a and Figure 5b ). The gearbox via the flexible shaft translates the motor shaft rotation to the fingers. The transfer of rotation to finger prehension is accomplished by a worm and spur gear combination (Figure 7a and Figure 7b ). The index and middle finger are guided from their interphalangeal joints to oppose the fixed thumb. The thumb harness, made of thermoplastic materials, is custom fitted to the patient. The maximum opening is adjustable up to 10 cm and the pinch at finger tips when it closes reaches up to 5 lbs. of force. This force is adjustable at motor controller level of circuitry. The second half of the head control circuitry is devoted to the control of finger prehension motor with a similar excitation method. The air button which signals the "go" and "direction" is opposite the pronation/supination button.
The benefits of an upper extremity orthotic system for severely paralyzed patients have been proven. However, a system previously reported elsewhere (Lehneis and Wilson, 1972), such as the IRM electric arm orthosis, suffers from cumbersome control and bowden cable power transmission. The Ranchos electric powered orthotic system (Nickel et al., 1969) and the Rehabilitation Institute of Montreal orthotic system both suffer from conspicuous control sites. The Burke modular orthotic system (Stern and Lauko, 1975), on the other hand, suffers from overly complex control modules and an unfavorable cost-benefit ratio. The researchers of the proposed HAG feel that the use of commercially available materials and a design that is easily assembled, fits most wheelchairs, and uses the patient's residual body power, would be desirable and affordable to patients. We have attempted to use a minimum and the simplest of controls. Our clinical evaluation so far was limited to two quadriplegics at C3-4 level. The HAG was compatible with our expectations of providing simultaneous shoulder abduction and elbow flexion and sequential wrist and finger manipulation to the patients. The patients tested accomplished activities of daily living, such as self-feeding, with relative ease after minimal training (an hour to two hours). In developing our HAG, we also tested the feasibility and ease of modifying parts of the system for individuals with different needs with success. Modularity of the system permitted a wide range of flexibility as to location of various parts and their size.
To summarize; the HAG enables the user to carry out some activities of daily living previously not possible, attempt others, and feel increased independence, self-confidence and self-esteem.
We are currently investigating a means to provide simple pinching force feedback display mounted on the transmission box of the prehension unit.
The authors wish to acknowledge the contributions of Jonathan Dietz and Vincent Durso, C.G. to this report.
This work was supported by the Veterans Administration RR&D Service
Nisim Benjuya, Ph.D., is project director at Rehabilitation Engineering R&D Lab, VA Medical Center, W. Roxbury, MA 02132.
Steven B. Kenney, BSME, is currently attending the Graduate School of the University of Arizona.