Akademik Veri Yönetim Sistemi
2.INTERNATIONAL BLACK SEA SCIENTIFIC RESEARCH AND INNOVATION CONGRESS, Trabzon, Türkiye, 28 - 30 Haziran 2024, ss.33-42
Pneumatic Artificial Muscles (PAM) are a type of soft actuators well-suited for use in robotic exoskeletons and rehabilitation robots due to their lower cost than electric drives, high power-to-weight ratio, and natural adaptability. On the other hand, the most important factor restricting the widespread use of PAM is its inherently nonlinear properties, creating difficulties in modeling and controlling actions. To overcome such difficulties, there are many modeling studies in the literature, such as geometric, empirical, and phenomenological models. The existing approaches mostly involve nonlinear methods that use direct input-output relationships, such as varying stiffness models or models that roughly estimate behavior. In the experiments aimed at understanding the characteristics of PAM, which we carried out in the test mechanism we developed, it was determined that the dynamic response of PAM to change in pressure input occurred as an integrated and simultaneous force production and change in
muscle length. Due to this complex behavior, existing modeling methods based on direct input-output estimation are not sufficiently suitable for modeling such a nonlinear dynamic behavior. Based on these results, an inverse dynamics modeling approach combining simultaneous force production and muscle length change to use PAM in control applications has been demonstrated by us in our previous work. In addition to our previous inverse modeling approaches, in this study, it is developed a black box inverse dynamics model based on Nonlinear ARX prediction using physical data obtained from the PAM testbed. According to the experimental test results, it is concluded that the Nonlinear ARX-based inverse model shows satisfactory performance and can be used as a simple and effective solution for PAM modeling and control.