| Summary: | Controlling multi-joint prostheses intuitively and effortlessly has been a research topic since the appearance of the first electric elbow prostheses. Researchers mainly focused on single handed tasks, however in daily life these are mostly executed with the healthy hand and the prosthetic arms only become relevant for two-handed manipulations. Thus, a new approach is presented in this paper addressing bilateral tasks. A taxonomy for bilateral tasks is elaborated in order to categorize and prioritize bilateral manipulations involving a prosthetic arm. Five different key figures for rating bilateral movements are introduced and used to form two quality criteria, which allow evaluation and comparison of different control strategies. Based on the proposed taxonomy and quality criteria, a generalized benchmark test environment is developed with five different evaluation scenarios and realized in virtual reality in an exemplary manner. Furthermore, a new controller-agent strategy, greatly facilitating the usage of prosthetic arms, is presented. The effectiveness of the criteria for evaluation of different control strategies is demonstrated with healthy subjects. With this evaluation concept, we provide the community a means to explore and compare controlling methods and inputs, facilitating the progress and development of new strategies.
|
|---|