Control of a lower extremity exoskeleton for human performance amplification

Abstract

The lower extremity exoskeleton is a wearable robotic device that should enable a human to walk with a heavy load for a prolonged period of time without reducing the human's agility. The exoskeleton is comprised of two anthropomorphic legs and a spine that provides a versatile loading interface. The device is to be designed and controlled in such a way that the human can conduct a wide spectrum of activities without feeling the device. This thesis presents a control scheme that does not require sensor interfaces between the human and the machine, or explicit command signals from the operator. The control law makes use of dynamic models of the exoskeleton to estimate the human-machine interaction force, thus minimizing the importance of the human interfaces in the performance of the system. Nonetheless, the hardware architecture and the control algorithm mutually affect one another and great care must be taken in harmonizing the two.