Block diagram illustrating physical pathway and learning mechanisms by which a robotic device might augment the learning experience. By altering task dynamics, robotic devices can make a task practicable or alter the flow of sensory information into learning centers in ways that provoke learning.

Effect of robot guidance on learning a desired trajectory (13). A. Subjects held the tip of a lightweight robot that guided them through a desired movement, which was a curve on the surface of a virtual sphere. B. Subjects learned to make the desired movement about as well simply by watching the robot make the desired movement (i.e., with visual training episodes alone, with their hands on their laps), without being physically guided by the movement. In this experiment, there were nine training episodes of seven movements each, with test episodes of seven movements interspersed between the training episodes. Figure based on data from Liu et al. (13).

Accelerating motor adaptation by transiently amplifying the dynamic environment. Unimpaired subjects walked with a lightweight robot attached to their lower shank (Fig. 3A). A. At step number 10, the robot applied a force field to the lower shank that pushed the leg upward during swing with a force proportional to the forward velocity of the lower shank. Subjects adapted to this force field in about 15 steps (A, dashed line). By roughly doubling the force field strength on only the first step in the force field (B, solid line), subjects adapted more quickly to the force field (A, solid line). Figure based on data from Emken and Reinkensmeyer (4).

Example of dynamic adaptation during golf putting. A lightweight robot was attached to the head of a putter as a golfer tried to make a 1.5-m putt on a flat surface. For trials 0–40, the robot did not apply any force to the putter. After 40 practice putts, the robot resisted the putter head movement with a viscous force field for 60 putts. Initially, the subject tended to putt short of the hole (the “direct effect” of the field), but decreased his putt length error with practice. When the force field was unexpectedly removed at trial 101, the subject exhibited an after effect and putted too strongly, but this “after effect” gradually decreased in size. Data from a University of California, Irvine student project by Dr. Eric Wolbrecht, currently at University of Idaho.

Using aftereffects to learn a desired trajectory. A. Subjects held the tip of a robot and reached to targets in the horizontal plane, with the arms supported. B. Using after effects to teach a desired trajectory to a healthy subject. Subjects initially reached straight to target. During training, the robot perturbed the reaching movements in the opposite direction from the desired movement using a force field. When the force field was removed, the after effects roughly followed the desired target trajectory. C. Using after effects to straighten the curved reaching path of a stroke patient. The results were similar to those in (B) for the healthy subject, except that for the stroke patient, the initial movement was curved and the after effect was straight. [Adapted from Patton J.L., and F.A. Mussa-Ivaldi. Robot-assisted adaptive training: custom force fields for teaching movement patterns. IEEE Trans. Biomed. Eng. 2004;51:636–646. Copyright © 2004 IEEE. Used with permission; and from Patton, J, Kovic, M, Mussa-Ivaldi F. Custom-designed haptic training for restoring reaching ability to individuals with poststroke hemiparesis. J. Rehabil. Res. Dev. 2006;43:643–656. Used with permission.]

Robot assistance can reduce performance errors while still allowing learning. A. The robot applied a force field to the subject’s ankle that forced the leg upward during swing, with a force proportional to the forward velocity of the ankle. B. Without robot assistance, application of the force field caused the subjects to step abnormally high, but they learned to step at a more normal height after several steps. Removal of the force field caused an after effect in which the subjects stepped lower than normal, indicating formation of an internal model of the robot force field. C. With robotic assistance, subjects never experienced large step height errors, and yet still learned a model of the force field, as evidenced by a similar after effect when the field was removed. D. The robot assistance started at a high level to cancel the force field but decreased from step to step according to an error-based learning law with forgetting. [Adapted from Emken JL, Benitez R, Reinkensmeyer DJ. Human-robot cooperative movement training: learning a novel sensory motor transformation during walking with robotic assistance-as-needed. J. Neuroeng. Rehabil. 2007;4:8. Copyright © 2007 BioMed Central. Used with permission.]