Human Aware Robot Navigation in Semantically Annotated Domestic Environments

In the near future, the seamless human robot cohabitation can be achieved as long as the robots to be released in the market attain socially acceptable behavior. Therefore, robots need to learn and react appropriately, should they be able to share the same space with people and to adapt their operation to human's activity. The goal of this work is to introduce a human aware global path planning solution for robot navigation that considers the humans presence in a domestic environment. Towards this direction, hierarchical semantic maps are built upon metric maps where the human presence is modelled using frequently visited standing positions considering also the proxemics theory. During the human's perambulation within the domestic environment the most probable humans pathways are calculated and modeled with sequential, yet descending Gaussian kernel's. This way, the robot reacts with safety when operating in a domestic environment taking into consideration the human presence and the physical obstacles. The method has been evaluated on a simulated environment, yet on realistic acquired data modeling a real house space and exhibited remarkable performance.