Improved modeling and design of soft robots using sensitivity analysis

Abstract

The key feature of soft robots is the use of soft materials. On the one hand, the softness is advantageous to fulfill tasks for which adaptability is important. On the other hand, the softness is challenging in modeling and design. We are sure that both modeling and design can benefit from sensitivity analysis. To illustrate this, we present two examples of a pneumatic multi-chamber actuator with load. The first example is based on global sensitivity indices and focuses on finding the model parameters of the actuator with the most influence. With this information, the model as well as the design of the actuator can be improved in an iterative process. The second example is based on Monte Carlo filtering and focuses on evaluating the influence of an eccentric external load on the actuator tip position, and the interaction of the parameters characterising the eccentricity position. In this case, the sensitivity analysis improves the general understanding of how external loads influence an actuator.