The system has to have a robust behavior against the different disturbances the user will face during the time they are using the prosthesis. The electrodes will not be always placed in the same location, the EMG patterns will change for the same movement depending on the arm position, fatigue or other external conditions will affect the system. The prosthesis cannot have an erratic behavior depending on external states unavoidable for the user. Thus, an optimal system should be usable in almost all common conditions with a high performance and not only in controlled environments. At the same time, the easy use of the prosthesis is also important. Tedious training protocols and complicated control structures, in order to achieve a robust performance, are not a solution and will lead to rejection. The use of a prosthesis has to be intuitive and natural, and for this a clear communication between the two agents is essential. Here the training paradigms play a key role and researchers will have to give them the attention proportional to their high relevance in the final outcome. For this, it is essential that the training procedures are clear for the user so the learned model is consistent. Coadaptive models are potential candidates to achieve these requisites. So the model is shaped by the user's learning and therefore by their comprehension of the system. These models will be then more logical for the own user and adapted to their way of understanding the control, increasing the final usability.

Chapter Contents:

• 8.1 Machine adaptation
• 8.2 Human adaptation
• 8.3 Recalibration
• 8.4 Coadaptive prosthesis control
• 8.5 Conclusion
• References

Inspec keywords: prosthetics

Other keywords: user-prosthesis coadaptation; training procedures; prosthesis

Subjects: Prosthetics and orthotics