Wearable-based evaluation of human-robot interactions in robot path-planning

Abstract

In robotics, many path-planning algorithms have been proposed but almost all are evaluated only by their path lengths in PC. However, the evaluation is completely independent of a human ability. Recently, a human intends to live with many robots. For this reason, we should experimentally evaluate many kinds of interactions between a human and robots in our living space, which are supervised by each path-planning algorithm. Unfortunately, it is impossible, expensive and dangerous for us to investigate such interactions in a real world. To overcome such drawbacks, we propose wearable-based system to evaluate a path-planning algorithm by a human being. A human avoids some neighbors of virtual robots by using two eyes (using his visual property) and two legs (using his movable ability). In our wearable system, we can quickly develop smart robots and sensors without considering many irrational technical limitations. Therefore, we can test a novel sensor-based path-planning algorithm in the virtual environment. Using this, their superiorities are checked by visual and movable abilities of a human being.