Online Variable Recruitment for Pneumatic Artificial Muscles with Springs

Sammanfattning: Pneumatic artificial muscles (PAMs) have gained attention in the realm of soft robotics for their high power to weight ratio, low manufacturing cost, low weight, and relatively high compliance. This makes them appear as a great candidate for exoskeletons. An area of recent research involves variable recruitment, the process of successively activating individual PAMs from a set to improve overall system efficiency. While a few simulation and quasi-static studies exist, very little research has investigated real time switching with a physical system. In the quasi-static studies, the buckling of non-activated PAMs has been a consistent issue. In this thesis, a set of six parallel PAMs are connected serially to individual springs to prevent non-activated PAMs from buckling during contraction. The system is run through both a batch and orderly, open loop recruitment cycle to better understand transition effects and energy consumption. It was found that the batch method uses more energy and is prone to disturbances during transitions. The serial elastic elements do prevent buckling at the cost of individual recruitment level movement capability. Recommendations for implementing the switching strategies and how to use springs are given.