Mohammad Salah

Imitation-based Skill Learning Simulator for Mobile Robots: Hypotheses of Imitation Learning and Simulator Design for Skill Transfer Mohammad Salah

In this text, a framework that allows a layperson to transfer skills to a simulated robot is developed. The simulated robot is played by a simulated truck with a trailer. The elements that make up the framework are imitation-based learning and primitives that describe the robot?s motions. The skill transfer is accomplished using imitation learning where the demonstrated movements are mapped to a set of primitives, which are combined in a desired sequence to implement novel skills. In this monograph, imitation technology is developed in a simple two-dimensional world of a truck. Enabling technologies for this simple world will be very similar to the technologies that will enable manufacturing, assembly, and maintenance as well as service robots to utilize imitation learning. Creating a framework to transfer skills to robots that could be used by lay people would have a tremendous effect on our everyday lives. The effect would be similar to the effect that software applications for PCs had on our society.