SUMMARY

Motivated by the rapid advances of intelligent robotics and the need for safe physical human-robot interactions, this proposed research seeks to develop a distributed-parameter method to design, model, and control a bio-inspired elastic-actuation system that utilizes its physical field to simultaneously actuate and sense the constrained motion/force for a spectrum of robot-environment interaction applications. Due to many unique features (such as high energy density, no backslash/friction, overload protection, and robustness under harsh environments), magnetic devices capable of transferring motions and forces/torques through their magnetic energies are widely used as an alternative to traditional mechanisms to design novel actuation systems. However, most of the existing designs rely on incorporating mechanical springs into off-the-shelf gearmotors and rotary-to-linear transmission, for example, series elastic actuators; the potential utilization of magnetic fields as non-contact elasticity and transmission has been underexplored. The proposed research explores the uses of the inherent magnetic field to achieve both spring-like elastic actuation and motion/force sensing in the design of an elastic-actuation system. Both the forward model characterizing the magnetic elasticity and its inverse solutions for estimating the motion/force directly from the measured magnetic field are derived analytically using a distributed current source method and verified experimentally. While the immediate focus of this research is the development of a magnetic series elastic actuator, the methodology developed in this research can be extended to a spectrum of bio-inspired actuator designs such as ankle joint exoskeleton and downhole robotic applications, where non-contact elasticity and embedded motion/force sensing plays an important role in interacting with human or environment.