Robots Can Train Humans Using Principles of Operant Conditioning Through Visual Reinforcement Tools

Researchers have established new techniques to study human-robot interactions based on current knowledge in interspecies communication and comparative psychology. Studies on animal acceptance of robot conspecifics in complex social environments has led to the development of robots that adapt to animal and human behaviors. Using a robot with adaptable algorithms developed by the authors, the researchers hypothesized that, by using familiar visual rewards as positive reinforcement, robots could use operant conditioning principles to teach humans a basic task. The robot in this study independently determined optimal control of construction equipment by capturing the motions from an expert operator. The robot then attempted to teach those same skills to novice operators using familiar, yet simple, visual reinforcement tools. In this study, participants were asked to manipulate a model excavator using feedback from the guidance system on a nearby computer screen. Participants were assigned randomly to one of 3 groups: simple visual reinforcement, complex guidance, and no visual feedback (blank screen). To measure learning, participants returned a day later to repeat the task without the guidance. The group using simple feedback resulted in cycle times that were closer to the expert times than both the complex or control groups and had significantly different end times (p˂ 0.05) than either group. This result supports the hypothesis that, similar to what has been found in vertebrates and invertebrates, robots can shape behaviors of humans using visual positive reinforcement.