Abstract:

We are at the developing stage of service robotics now. Robots provide services for people in the public and home environments. Before that, we need to let robots understand the social behaviors in order to letting robots know how to interact with human correctly. Furthermore, maybe we can achieve the step that robots will cooperate with people and join the coexistence society of humans and robots in the future. The way in which robots are going to live with humans has become an important issue. One of the most common human social behaviors is standing in line. Therefore, to integrate robots into human society, queuing is an important ability.

In this thesis, behavior understanding mainly targets on a queuing task consisting of the interactions between people and the group. A pedestrian queuing model was developed that can help robots to detect and learn crowd effects in human society. I have added the queuing model environment constraint of entrance detection on an occupancy grid map, and bus stop detection. The entrance detection algorithm detects entrances as represented by all kinds of doors and allows the robot to identify the location as a traffic artery with an SSE, thus facilitating robot behavior that is acceptable in human environment. Using the SIFT algorithm makes the identification of bus stops easier to perform.