Novel Design of a Transradial Socket to Allow Independent Pro-Supination Control in a Myoelectric Prosthesis

Abstract

Background/Objectives: Individuals with transradial limb loss or absence often retain the ability to pro-supinate their forearm, but the traditional design of the prosthesis precludes this motion from being used for direct prosthesis control. Methods: A prosthetic arm was created for a single user that employed a novel split inner socket to allow pro-supination of the residuum to control a powered prosthetic wrist rotator. A total of 14 subjects (13 able-bodied subjects and one prosthesis user) performed the Refined Clothespin Relocation Test. The user performed the test with their own and a novel research prosthesis, which allowed independent hand and wrist function. Movements of limb segments were recorded using a motion capture system and an analysis of limb segment angles and compensatory motion was made. Results: The research prosthesis reduced compensation in the trunk and head and reduced pain in some joints, while the time to complete the test increased. Conclusions: This method has the potential to create additional intuitive control channels for transradial prostheses.

1. Background

The role of the arm in manipulation is to preposition the hand in the right orientation to grasp the object appropriately. With fewer degrees of freedom available than the seven the human arm possesses, a person with an arm injury or a prosthesis needs to compensate for the movements that do not exist or are unavailable. Compensatory motions can use any part of the body [1] and this can lead to over-use injuries [2,3]. Even when an axis is reproduced prosthetically, the device may be too heavy or slow to be used by wearers. A common prescription for unilateral transradial amputation is dual-site control of an electric terminal device [4]. By contracting the flexor and extensor muscles, the user can create two electromyographic (EMG) signals that are used to open and close the prosthetic hand. Adding a wrist unit increases complexity, mass, bulk, length and the moment arm [5]. Studies show these to be some of the reasons for prosthesis rejection [6].

Conventionally, the control input signal is shared between the wrist and hand, so only one can be controlled at a time. This is not intuitive and necessitates concentration for the user to control the device. The cognitive effort required to select the wrist function, rotate the wrist, and then switch to the hand function are deterrents to routine use. The user may decide that the effort required is too great and will find a simpler strategy. They may use over-rotation of the shoulder (humeral flexion and abduction) to place their hand in the correct position [7]. Alternatively, the user may simply reach across and employ their intact hand [8]. Other inputs have been used in an attempt to capture forearm motion, such as mercury switches [7,9], and fiber optic sensors [10], and changes in contraction rate/magnitude [11], with varying degrees of success.

Taylor [12] related the degree of pro-supination a person is able to conduct after amputation of the forearm to the length of the residuum. It varies from almost none with a very short residuum to 100% with a loss through the wrist. Conventional sockets for myoelectric prostheses are composed of a rigid inner socket and a forearm section plus a terminal device. Even if the user can rotate their residuum, once the socket is donned its rigidity prevents movement. This means the motion is not usable as a control input. This project aimed to investigate if that movement could be captured and used as an independent control input. If this input arranged to control the pro-supination of a powered wrist, dedicating surface EMG electrodes solely to controlling the terminal device, it has the potential to create an entirely intuitive control strategy for the user and their prosthesis.

The recognition of patterns of EMG contractions within the musculature of the arm can be decoded by a network of electrodes to obtain the user’s intended wrist motion. Pattern Recognition (PR) algorithms have been successfully applied in the field [13], but the expense and complexity can be a high bar to routine use. What PR offers is seamless sequential control, it still only decodes one motion at a time. The user can move from hand to wrist control without performing co-contractions to switch joints. It is clear that simultaneous control is the ultimate goal of pattern recognition [14]. It is currently only demonstrated in the laboratory [15,16]. Even when practical, there will still be users who are physically, cognitively or financially unable or unwilling to use a PR solution, so there remains good reason to investigate alternative control strategies

Similarly, Targeted Muscle Reinnervation (TMR) surgery has been successfully used to enable simultaneous control in myoelectric users [17]. The control is more intuitive if the signals from the nerves are used to drive the same joint in the prosthesis (e.g., hand) as they did before the amputation. This results in an independent direct control strategy [18]. TMR surgery has more often been applied to absences above the elbow [17,19]. TMR on the forearm has been less successful [20].

Assessing the performance of a prosthetic solution and its impact on patient functionality requires the use of the appropriate outcome measure. As users will perform compensatory motions to overcome the shortcomings of their prosthesis, a comparison with the motions of an able-bodied control group allows observation of the effect of an intervention. Previous studies of transradial prosthesis users analyzed the compensatory motions [21,22,23]. The relationship between the compensatory motions of prosthesis user and repetitive strain injuries has been discussed [1,3,24]. Compensatory motions have been studied using motion capture of body segments. Hussaini [25] has described the refined clothespin relocation test (RCRT), which when combined with motion capture, creates a standardized test to observe and compare compensatory motions. Using this method, improvements in prosthetic technology can then be observed. The question of whether they reduce compensatory motions can then be investigated.

2. Methods

In the novel research prosthesis presented here, an inner socket was devised that allowed pro-supination of the residuum and created a viable control signal for powered wrist rotators. The suitable volunteer required a residuum which was roughly half full length; therefore, a myoelectric prosthesis user attending the Atlantic Clinic for Upper Limb Prosthetics was approached. He possessed a mid-length loss which had the potential to rotate 40–60% of the unimpaired range. The user’s current device used a conventional supracondylar suspension prosthesis holding a single axis Motion Control hand with two electrode/amplifiers and a conventional two-channel control strategy with a quick disconnect wrist. He used his device on a daily basis.

2.1. The Design

2.1.1. Pilot Study

Initial tests of the design were performed on a socket with liner (Figure 1), which was then fitted inside a test rig (Figure 2) that the user could don. This approach aimed to allow the observation of the motion of the distal tip of the residuum.

When casting the volunteer for the socket, the distal tip of the negative mould was compressed by the prosthetist in the mediolateral direction to ensure the socket gripped the distal end of the residuum. The socket was split into two parts: a proximal section around the elbow and a distal end cap to capture the forearm rotation. Fenestration in the liner was located in the region of the split to enhance rotation of the distal end cap. Accelerometers were used to track the distal rotation, which was the rotation of the sensor on the participant’s limb, measured with respect to the sensor affixed to the back of the wrist rotator. EMG electrode/amplifiers were incorporated into the socket and an Arduino captured the data and ultimately sent control signals to a motor driver for the wrist. The EMG electrode/amplifiers were employed in the same way as in a conventional clinical system. The amplifiers use a low pass filter to a create slow moving analog voltage that approximately reflects the tension in the muscle and has an upper frequency of less than 100 Hz. In Figure 2, to improve the acquired rotation signal by reducing any slippage between the hard cap and the soft compressible liner, the distal end cap was removed and the accelerometer was affixed directly to the distal end of the thermoplastic liner. The use of the accelerometer was eventually abandoned.

Once the rig was assembled and donned, the user was asked to pro-supinate their residuum and to contract their muscles to command the hand to open and close whilst placing the test rig in different orientations to allow the motions of the forearm and the independence of the two control signals to be observed. This was evaluated through observations of any unwanted activations in different orientations. The user was asked to move the wrist and any inadvertent motion of the hand (or vice versa) was observed and the threshold for the wrist was altered in the custom electronics. For the hand, the manufacturer’s software was used to increase the EMG threshold to prevent unintended activation.

It was noted that the distance between the distal end of the residuum and the wrist rotator in the rig was not constant. This is the result of the way in which the radial bone rotates around the ulna; there is a small translation. This pistoning is a natural motion and must not be restricted in a final design, or it may limit the amount of forearm rotation that can be captured.

Using the accelerometers required greater signal processing and increased physical electrical wiring, which introduced greater complexity and would have reduced reliability. A potentiometer provided the simplest hardware and most reliable and consistent signal for driving the rotation of the wrist unit. The output of the potentiometer was reflected in the operating range of the wrist rotator. The potentiometer voltage range was input to the Arduino, which mapped that to a value between 0 to 1023 (0 to 5 Volts). The neutral position corresponds to the median value of 512; the pronation value was less than the median and the supination value was greater than the median. In practice, these values were selected to prevent inadvertent activation (e.g., Pronation < 400, Supination > 700), thus creating a ‘neutral region’ around the median value.

The range was divided into three regions corresponding to neutral, clockwise rotation, and anticlockwise rotation. These threshold values were determined empirically. For this first design, the wrist rotation was restricted to one speed (so-called ‘bang-bang control’).

2.1.2. Definitive Prosthesis

To create the definitive prosthesis, a second split socket with liner was made. The distal end cap of this iteration was shaved flat (Figure 3). The socket was re-donned and the user was asked to rotate their forearm. A 1/2-20 bolt was affixed in line with the rotation axis. Separation allowed forearm rotation to be captured. However, the split meant that distal end cap (and distal residuum) no longer tracked with the prosthetic hand and that much of the load-bearing surface of a traditional inner socket is now further concentrated to the elbow section. This needed to be accounted for as it can result in the top surface of the distal end cap colliding with the inner ceiling of the forearm section during elbow flexion, creating a large localized pressure (Figure 4).

Figure 4 shows the two axes. The rotation axis (offset towards the ulna) is where a rotation sensor (potentiometer) was placed to measure the level of pro/supination and act as the command input for the controller. The forearm axis is the axis on which the prosthetic hand and wrist unit are situated. As there was a concern that the forearm would be pro/supinated away from the neutral and more loading would be concentrated over a smaller area, a sleeve bearing was positioned over the bolt on the rotation axis. This provided the surface for load bearing and allowed for rotation while minimizing lateral displacement. A potentiometer was attached to the 1/2-20 bolt, the sensor was set to have the middle of its range as close to the mid-range of the forearm as practical.

The system schematic is shown Figure 5. As the aim of the study was to assess the socket design, the electronics used many standard components as practical to make the comparison with his routine prosthesis simpler. The hand was controlled with standard prosthetic electrodes and the conventional two-input analog controller within the hand (Motion hand—Fillauer). The changes in voltage were used to signal the wrist unit to rotate in the clockwise or anticlockwise direction. The control of the wrist rotation was similar to the myoelectric control of a prosthetic axis. Two inputs (forearm pro/supination, flex/extension for the hand) are used to set the direction of motion with a dead-band in the middle set to prevent wrist (or hand) movement on small or noise voltages. The commercial hand used the controller built into it. For the wrist, the pro/supination was realized using an Arduino Uno (https://www.arduino.cc/ (accessed on 13 March 2025)). This simple control format was well within the capabilities of the microcontroller.

The powered wrist was a Motion Control unit (https://fillauer.com/ (accessed on 13 March 2025)) driven by 7.2 V (max speed 32 rpm). Similar to conventional myoelectric control, the rotation range generated by the user was mapped to prosthesis rotation speed to give maximum sensitivity while not triggering the wrist or hand when instructed separately. This setting was achieved by working with the user, testing their ability to open and close the hand and rotate the wrist with the arm in different positions and adjusting the gains of the electrodes and the controller.

The final design worn by the user is shown in Figure 6. The external wires were repositioned during the assessment (detailed below) to ensure they did not interfere with his motion.

2.2. Experimental Setup

The user attended the clinic to allow for the fitting and adjustment of the system. He visited on two further occasions: the first to perform the RCRT, with his standard prosthesis and then with the research prosthesis.

The RCRT is a training procedure using the Rolyan Graded Pinch Exerciser—Handelnine Global LLC, 16192, Coastal Hwy, Lewes DE 19958, USA. https://ninelife.uk/ (accessed on 13 March 2025) that has been adapted to become an assessment tool [17,26] and further refined by Hussaini for use with motion capture [25]. In this form, it aims to test the range of motion and coordination of the upper extremity with the rest of the torso. A test consists of the subject standing and moving three clothespins between the middle horizontal rod and onto the vertical rod followed by a second run with the pins being returned to the horizontal bar (referred to as ‘upward’ and ‘downward’ assessments) We believe the emphasis here makes the definition clearer, Figure 7. The duration of a trial and the start and end points are recorded through the subject pressing a button on a timer with their prosthesis. From this, the different tests could be temporally scaled and aligned. The RCRT is an ideal test to evaluate the separation of hand flexion and independent wrist control. Compensation through over-rotation of the shoulder is easily identified.

In an earlier study [25], the RCRT was used to compare the user with his conventional prosthesis with thirteen left-hand dominant, able-bodied users. For that study, the prosthesis user participated and employed his usual terminal device (Motion Control single degree of freedom hand with flexion wrist locked in flexed position).

Motion analysis of the subjects was performed using an eight-camera Vicon M-Cam motion system (Oxford Metrics, Oxford, UK). Prior to performing the assessment, the subjects’ arm, forearm, wrist and hand were measured. These facilitated the joint angle calculations. The subject wore a fitted shirt that would allow for the markers to be attached to bony landmarks on the body. The marker set used in this study was modified from Zinck [1] and Ross [27].

Data were exported from the Vicon system into Matlab for processing, marker trajectories were filtered with a zero-lag fourth-order Butterworth filter with a 5 Hz cutoff frequency. Three local coordinate systems (pelvis, trunk, head) and three local measurement angles (shoulder, elbow, wrist) were created. The coordinate system was named after the body segment or limb that it represented and consisted of three orthogonal axes. The motion of one coordinate system relative to another during the assessment tasks was analyzed to compare the prosthesis user to the able-bodied group.

In this study, shoulder rotation was tracked using the lateral epicondyle marker relative to the marker placed on the acromion process of the same side. The epicondyle marker movement relative to the acromion marker was not a direct measure of shoulder movement but was a repeatable measure on all subjects. A three-dimensional angle between the elbow, acromion, and right hip markers was used to define the shoulder angle. The elbow angle was defined as the angle between the acromion marker, the lateral epicondyle marker and the marker on the radial styloid.

The user answered surveys for each day of testing. The surveys were completed directly following the sessions (which were 4 months apart). Questions were concerned with usage, pain and the ability to perform the test. This study and protocol were approved by the Research Ethics Board (REB 2013-114) at the University of New Brunswick (subjects provided written consent before participation).

3. Results

Figure 8 shows the time normalized trajectory of the shoulder angle for the upward and downward assessment in three cases. The trajectory represents the physical three-dimensional angle between reflective markers at the lateral epicondyle, the acromion process, and a marker on the right hip on the subject’s right side. The user’s current device, the research prosthesis, and a typical able-bodied subject’s trajectory (age and gender matched) are shown by the grey, black and red trajectory lines. To facilitate comparisons between different trials, the horizontal axis was normalized between 0% (start) and 100% (end task) when the subject pressed the timer button [28,29]. The trajectories are fit to a cubic spline with 400 data points.

The magnitude of the shoulder trajectories of the able-bodied subject and the current prosthesis peaks at three points (0–20%, 40–60%, and 80–100%, for the Up trial). These are points in the trial when the clothespins are manipulated on the vertical rod. The order in which the peaks appear can also be used to establish direction, as peaks get larger with time in the upward direction and smaller in the downward direction, reflecting the distance between the grasp and placement of the clothespin.

The research prosthesis trajectory had more variation. Much of this was from the change in eccentric loading on the residuum. The center of mass of the hand created an eccentric rotation and loading about the longitudinal axis of the forearm. The user was continuously adjusting his elbow position by lifting or lowering their arm, which influenced the shoulder angle.

In the instances when starting or stopping the rotator was difficult due to their lack of experience with the new device, the user attempted quick/jerky motions to stop the device, which are reflected in sudden spikes in ROM when the device was not stopped at the desired location. They also compensated by increasing or decreasing their shoulder angle, which resulted in jerky motions that appear as spikes in the trajectory plot. Similar differences are found in the Down direction.

Figure 9 and Figure 10 show the trunk and head motions (lateral tilt, flexion, rotation) during the upward and downward assessments. The able-bodied group is displayed as an averaged trajectory within the shaded boundary (95% confidence interval). The routine device trajectory is shown in grey and the research prosthesis is shown in black.

In the Upward assessment, the range of motion (RoM) of the lateral tilt with the research prosthesis was similar to that of the conventional device, though the range of motion is centered about different values (0⁰ for the research device, 5⁰ for the conventional. The research prosthesis’s total RoM was largely a result of perturbation by the user trying to start and stop the wrist, evidenced by the spikes on the trajectory plots, as opposed to the conventional devices where the ROM was a result of using the lateral tilt motion to manipulate the clothespins. The first clothespin was picked at 6% and placed at 14%. The placement resulted in a decrease in lateral tilt that can be seen in the plot. The second clothespin was from 30% to 52%, and the third from 75% and 97%. The user adopted a more neutral stance with the research prosthesis. The able-bodied subjects showed similar peaks during placement but with positive angles at 45% and 85%, corresponding to the placement of the second and third pins. The lateral tilt for the able-bodied user was primarily to achieve hand height, whereas the prosthesis user’s lateral tilt was for the proper position of the hand relative to the vertical rod.

Trunk flexion of the research prosthesis is in sharp contrast to the trajectory with the conventional device. The RoM during trunk flexion reduced from 30.4⁰ to 25.2⁰ between the standard and the research prostheses. The latter also provided a more neutral position for the trunk flexion angle (closer to 0⁰). The decrease in trunk flexion was related to the ability of the user to attain more height with the rotation of the prosthetic hand. The flexion in the head had a similar RoM across all subjects.

The trunk rotation RoM with the current prosthesis was 24.5⁰, compared to the range of 13.0⁰. As with trunk lateral tilt and flexion, the research prosthesis placed the user in a position closer to neutral when performing the task. The largest difference was the trunk rotation for the third clothespin, which resulted in more than 25⁰ to the dominant side. The research prosthesis required rotation of less than 10⁰ at this same point.

Head rotation for all subjects was greatest for the movement of the third clothespin, which required the longest horizontal distance moved as he was following the placement of the clothespin. The RoM of the head rotation increased slightly. As inferred from the analysis of the video, it can be seen that the user’s eyes focus on the timer, which is to their right side. The user’s head rotation for both prostheses starts and ends on the same angle, confirming that the timer’s position relative to the last clothespin on the horizontal rod defines the range of motion in this trajectory.

In the Downward assessment, the lateral tilt trajectories had a RoM between 14⁰ and 19⁰. Peaks in the trajectory of the able-bodied subjects and the standard device’s trajectory correspond to when the top clothespin was removed from the vertical rod. The research prosthesis trajectory was more a representation of the user attempting to start and stop the device with quick, jerky motions of the forearm and trunk. This was confirmed with a video review of the trial. The head lateral tilt with the research prosthesis confirms a synergy in coordinated motion of the trunk and head, evident in the first quarter of the trial (positive trunk lateral tilt matched with negative head lateral tilt).

At each grasp of the clothespin on the vertical rod, the user did not have to bend as far back with the research prosthesis. The RoM was similar between the two prostheses, but the research prosthesis allowed the user to maintain a more neutral stance, as the trajectory was closer to 0⁰ throughout the exercise. The able-bodied subjects had the smallest RoM in trunk flexion, but the coordination with head flexion made it apparent that they compensated for low trunk flexion with an increase in head movement. It can be seen that the user did not tilt their head back as far to grasp the final two clothespins when using the research prosthesis.

The range of trunk rotation with the current prosthesis was 42.3⁰, compared to the research prosthesis (24.7⁰). This was balanced by the smaller head rotation with the current prosthesis and a larger rotation with the experimental. Grasping the initial clothespin at the highest location resulted in a very large trunk rotation with the conventional device, and there was a corresponding larger negative rotation. In the research prosthesis’s trajectory, there was a large positive peak in head rotation (20⁰) when placing the first clothespin (27%). The user over-pronated the wrist and compensated for this by twisting their body towards the dominant side. This can also be seen in trunk and head flexion and the trunk and head lateral tilt. For the remainder of the task, head and trunk rotation was reduced, and the user performed the task in a more neutral position. In both directions, the motions of trunk, head and shoulder flexion of the user coincide with the underlying shape of the able-bodied subjects’ motions. The peaks follow the same order and have the same relative position.

The reduction in trunk flexion was evident with the research device, as it was with the upward assessment (Figure 11). At each grasp of the clothespin on the vertical rod, the user did not have to bend as far back with the research prosthesis. The RoM was similar between the two prostheses, but the research prosthesis allowed the user to maintain a more neutral stance, as the trajectory was closer to 0⁰ throughout the exercise. The able-bodied subjects had the smallest RoM in trunk flexion, but the coordination with head flexion made it apparent that they compensated for low trunk flexion with an increase in head movement. It can be seen that the user did not tilt their head back as far to grasp the final two clothespins when using the research prosthesis.

3.1. Time Taken to Complete the Task

The unimpaired subjects were up to ten times quicker than the prosthesis users with either prosthesis. The user was clearly still learning to employ the rotator and initially took longer as he overcompensated for the device. In every instance, the user took longer to execute the task with the research prosthesis, but with far more reduced compensatory motions. Note, the unimpaired cohort is faster than the user, and the time compression of the research prosthesis trajectory results in low frequency of trajectory changes appearing as high frequency changes in Figure 9 and Figure 10.

3.2. Survey Responses

Feedback from the survey indicated that the research prosthesis alleviated pain in the shoulder, neck and residuum, but there was an increase in pain in the elbow. The user indicated that he did not have to bend his neck as far to complete the exercise with the research device, and this was reflected in the trajectory of head and trunk flexion. The user responded that the wrist rotator also improved the ability to perform the task, even though the device felt marginally heavier. He felt it was easier to use, i.e., that the cognitive load was lower.

The user initially indicated a difficulty in starting and stopping the wrist rotator but was able to adapt to the prosthesis mechanism with larger and quicker rotations of the residuum inside the socket.

They also indicated a preference for the functionality that the research prosthesis provided. The novel independent input source from forearm rotation was one that the user was comfortable controlling and displayed an ability to adapt to this functionality during the testing period. However, the user suggested that this novel input source could be used to instruct powered flexion and extension of the wrist instead of rotation. This was outside the scope of this study.

4. Discussion

The focus of this study was to assess the design of a novel and potentially intuitive way to capture an additional independent degree of freedom for use in prosthetic arm control. A redesigned socket is used to obtain direct forearm rotation. Its impact on the motions of a transradial prosthesis and to see how this change manifested itself in terms of user performance was assessed with a manipulation task. This socket design may lay the foundation for future ideas that can improve functional outcomes for the transradial user. There are few comparable studies to this work; however, two are useful to consider:

In a case study by Carey [22], a bilateral transradial user with mid-length amputation performed two activities using their right side prosthesis. The first activity was a simulated drinking from a cup task, and the second involved opening a door with a standard door knob. They performed the exercise with their myoelectric device first and then with their body-powered prosthesis. There are two observations of that study important to this work: The first is that different prostheses lead to different ranges of motion across intact joints for the same task. The second is that the specific task requires an increase in RoM of some joints more than others. Both of these observations were seen in this current study as well. Different prostheses led to changes in the RoM of the same user for the same task. For example, a reduction in trunk flexion was seen when the user placed the clothespin on the vertical rod when using the research device. Looking at different tasks, there can be a larger change in the range of one motion but minimal change in another. For example, manipulation of the clothespin on the vertical road affected the trunk lateral tilt more than manipulation on the horizontal rod. Manipulation on the horizontal rod affected shoulder motion more than manipulation on the vertical rod. In both these cases, the user was employing the research device.

Metzger’s study [23] considered that the loss of degrees of freedom may contribute to the increase in trunk range of motion to accomplish the task. The users with transhumeral amputations also displayed larger motions in the trunk when compared to transradial users. In our study, the inclusion of a degree of freedom reduced the range of motion in the trunk, which lends strength to Metzger’s discussion. It is reasonable to assume that increasing the number of degrees of freedom will reduce the trunk motions further, so long as the user is able to control them properly. The intuitive motion (forearm rotation) used in this work was a prime example of this, as it seemed to reduce the cognitive effort required to control the two outputs. The effect of overuse (or increases in range of motion) on upper limb prosthesis users is one of the areas that still lacks sufficient research. There are no long-term studies that have been performed on this population, and much of the indications that exist come from anecdotal data and case studies.

The relationship between posture, RoM and repetitive strain is still one that needs to be formally established. The literature review by Gambrell [24] attempts to find this relationship through work external to prosthetics. The survey by Jones [3] combined the overuse problems with other musculoskeletal concerns. In this current study, the post-testing surveys attempted to isolate the effect of decreasing the RoM in more proximal joints (shoulder and trunk) with the addition of increasing the range of motion in distal joints (the wrist) through the use of a pain scale. When comparing the current and research prostheses, the user indicated that there was a decrease in the pain felt in his shoulder. This is an indication of how decreasing the range proximally through the use of an additional distal degree of freedom can reduce the pain associated with overuse. The user also experienced a reduction in neck pain and the trajectory indicates a reduction in range of motion and improvement in posture (closer to 0⁰). Although these results are from one user, the fact that the user feels an alleviation of pain with a reduction in RoM should encourage the use of a pain scale survey in any future work with compensatory motions so that a database can be built.

The intention of this study was to observe how a prosthesis that allows simultaneous control of the prosthetic wrist and hand impacts the kinematic motions of the user. The user’s current device, however, did not have a wrist rotator, and therefore, the results presented here are a combination of adding an additional degree of freedom, allowing intuitive independent control of both outputs, and simultaneous control.

The user did not have much time to learn to use the wrist prior to testing. The wrist was capable of nearly double the speed that was used in this exercise. This is one application where Extended Physiological Proprioception (EPP) would be a significant benefit to the user [30,31]. As the control in this prosthesis was as close to physiological as possible, the additional benefit of EPP could be significant. However, given the space constraints, the design and application of such a device would be challenging. For this application, the measured pro/supination would then be mapped directly onto the rotation of the prosthetic wrist, but this does have implications for the power the prosthesis uses.

The user was able to learn to use the device quickly and found the use intuitive, this suggests that the cognitive load needed to control the device was no worse than a conventional controller. The fact that he could control the two axes simultaneously and not have to separate them, as individuals learning a new skill tend to do [32], suggests that the load is decreased. Future studies with longer term use would be able to assess this. However, the tendency to over-rotate the wrist shows that more practice in use and adjustments to the thresholds are likely to be needed to reduce this effect.

5. Conclusions

This work attempts to face the challenges in control of powered upper limb prosthetic devices that are technologically advanced but lack a control structure that provides intuitive control and decreased cognitive effort. Specifically, it looked at the viability of using forearm rotation to create a user-initiated control signal for a wrist rotator, resulting in independent control of hand and wrist function, the provision of which is still lacking in transradial prostheses.

This study provided evidence that compensatory motions can be reduced if an additional degree of freedom is added to the prosthesis. Motions of the trunk and the head showed improvement and the prosthesis user was able to accomplish the tasks in a more neutral stance. The improvements in posture and reduction in range of motion should decrease the chance of developing a repetitive strain injury. It also demonstrates a novel way to capture residuum motion as an intuitive additional control input for users with a mid-radial absence.