Autonomous Navigation by Mobile Robots in Human Environments: A Survey

With the service robots are well used in more and more indoor environments, autonomous navigation in such a human environment has been explored in recent decades. Different from the traditional navigation schemes, the new scenarios pose challenges about how to deal with the dynamic obstacles, especially the humans. To overcome the challenges, researchers need to consider: 1) the uncertainty of humans motion, 2) the interaction between human and robot, 3) the group information of the people. Also, the energy cost in the navigation process is of vital importance. In this case, the navigation requirements go far from the shortest path. In this paper, we reviewed the related works in the past decade, which can be roughly divided into four categories: reactive based, predictive based, model based and learning based. For each category, we analyzed some state of the arts, and listed the pros, cons and open problems. In the last of the paper, we summarized some evaluation metrics and corresponding methods.