Adaptive control and system identification for direct brain control of artificial limbs

Developing a non-invasive direct brain control of artificial limbs is both challenging and desirable. Such a sensory and control system, if successful, will have profound impact on the disabled. In this paper, we present the design of a control algorithm capable of automatic reconfiguration to account for changing sensor conditions, selection of an appropriate transfer function based on input characteristics, and adaptation to adjust output strength based on the user's activity. This algorithm is designed for use with our developed hybrid non invasive brain monitoring technique for the purpose of artificial limb control. The development of the algorithm and its initial performance together with our sensory system are presented and discussed.