Evaluation of a modular prosthetic arm

Introduction In the ToMPAW arm (1) we demonstrated the use of a digital network in a modular prosthesis. The network was built of a number of Neuron™ microcontrollers communicating through the LONWORKS protocol (Echelon Corp.). The hand in the demonstration prototype was based on the Southampton hand. The wrist and elbow used Edinburgh Modular Arm components, with carbon fibre tubes housing geared d.c. motors. Although this was a field prototype and not a fully engineered version, it stayed in regular service for ten years, from 2001 to 2011. Method The user possessed a short humerus following traumatic amputation 20 years previously. Initial training developed muscle bulk and control of two EMG signals. Control of the two channels was not fully independent, so a ‘winner takes all’ algorithm was adopted, with the larger signal assuming control. The joints were controlled sequentially, using scapular abduction to select them with a switch built into the harness; hand, wrist rotation, elbow flexion/extension, returning to hand, and defaulting to hand after a fixed period of inactivity. The hand had sensors for touch and slip and was capable of automatically selecting precision and power grip patterns. Results The user soon mastered control of the prosthesis despite long term lack of use of the relevant muscles, and was a keen participant in the arm’s development. One observation that he made was that control appeared to be improved when the metal and carbon fibre structure of the prosthesis was touched. This suggested that electrical interference was involved, leading to further study into electromagnetic compatibility, as reported in (2). Adding a conductive plastic earth to the inside of the prosthesis socket resolved this issue without affecting safety. No problems were encountered with the network operation, which proved to be highly reliable, though occasional maintenance was needed to remedy mechanical faults. A common problem was wires breaking where they crossed joints. This was ameliorated by using an extra flexible type of cable with a large number of copper strands and silicone rubber insulation, as used for items such as test meter leads. There was also a failure of the wrist due to gearbox seizure and consequent burning out of the motor. The supply voltage to the wrist was reduced to help avoid a recurrence. Maintenance on the hand was mostly limited to changing FSR force sensors as they wore with age. The elbow was entirely reliable, needing no maintenance over the ten year period. Initially a stack of 10 NiCd cells was used, giving 12 V for the wrist and elbow motors, with a tap at 6 V, later moved to 7.2 V, for the control electronics. The Neuron devices required 5 V and this was supplied by regulators on the controller circuit boards. Later, the cells were changed to NiMH types, with a smaller size for similar capacity. This reduced the weight of the battery pack from 550 g to 350 g. Conclusion The concept of a modular prosthesis with digital networking was demonstrated in a ten year trial. References (1) Kyberd, P.J., Poulton, A.S., Sandsjo, L., Jonsson, S., Jones, B., Gow, D. (2007). The ToMPAW modular prosthesis - a platform for research in upper limb prosthetics. Journal of Prosthetics and Orthotics, 19, (1), 15-21. (2) Poulton, A.S. (2010). Electromagnetic compatibility in myoelectrode amplifiers: isolation, impedance and CMRR. ISPO 13th World Congress, Leipzig, Germany