PROJECT SUMMARY/ABSTRACT Simple tasks such as using a computer feeding oneself personal hygiene and grabbing objects areimpossible for high level tetraplegics without assistance. Remarkable advances in brain-machine interfaces(BMIs) in the past 20 years however have demonstrated that paralyzed individuals can exert good controlover assistive robotic devices derived from neural recordings using electrodes implanted in the brain. Despitethe promise of such BMIs there nevertheless remain significant shortcomings including 1) the duration overwhich neural recordings remain viable is limited to a couple of years and 2) the procedure requires invasivesurgery associated with substantial risks and costs. Most individuals with high level paralysis however retainthe ability to voluntarily move their head and tongue activate facial muscles and can speak. It seemsreasonable to hypothesize therefore that signals derived from these actions could be used to controlmovements of a robotic arm accurately and intuitively. The main goal of this project therefore is to evaluatethe utility of non-invasive methods to supply the inputs needed to control movements of a robotic limb toperform a variety of tasks. Toward this goal we will carry out three specific aims: 1) evaluate control types(position and velocity) and input modalities (head face/head EMG tongue voice) for regulating robotic armposition 2) assess various methods to control robot arm grasping 3) characterize improvements in robotic armcontrol performing standardized real-world tasks with practice. Importantly data collected here using non-invasive methods during standardized tasks will provide a crucial benchmark needed for evaluation of andjustification for using BMIs developed in the future. Moreover this project will provide a major advance towardthe development of non-invasive and readily controlled assistive robotic arms that could greatly increase theindependence and well-being of individuals stricken with high-level paralysis.