Specknet infrastructure for Cyber-Physical Systems

In an iconic sequence in the 1930´s film ldquoModern Timesrdquo - a commentary on the banality and boredom of life on an industrial production line, Charlie Chaplin is seen working on a conveyer belt, trapped in an endless loop of repetitive actions. In more recent times, the human on the conveyer belt has been replaced by robots corralled in their own pens, and their interaction with the human is a distant one, if at all. The vision of the future is altogether different: robots and humans living cheek by jowl and interacting as part of a wider cyber-physical system. The talk will explore the fabric of such as a system based on specknets: wireless sensors on the human for real-time motion capture is networked wirelessly with humanoid robots to provide unprecedented levels of flexibility and freedom in real-time interactions. The resulting ldquophysical avatarrdquo acts as a physical presence, either near or afar, taking the place of the human operator for purposes of communication, exploration and object manipulation in collaboration with other such robots or humans. An in-built online learning and classification system allows motions of a human operator to be learned by the robot. We discuss whether everyday use of a motion-capture physical avatar may eventually provide enough training data to adequately prepare robots for autonomous operation. The talk will be interspersed with videos of the implementation of the infrastructure.