| Summary: | 碩士 === 國立臺灣大學 === 機械工程學研究所 === 106 === With the development of technology and science, robots are becoming more and more essential for society. For these robots, understanding the environment is important so that they can provide suitable responses. To realize the use of space, an environment model is one of the principal aspects for robots. As a result, this thesis focuses on developing an environment model based on people’s behaviors. In order to recognize the behaviors, a new structure of neural network called dynamic highway networks is proposed. Besides, 2D pose estimator, Laban movement analysis and fuzzy integral are employed. With these methods, two new behavior recognition algorithms are developed and a method to record the relationship between behavior and environment is proposed based on the theory in psychology. Using this environment model, robots can judge whether a behavior is abnormal or not and can consequently react correctly and guide a person to the proper place to perform their behaviors.
|
|---|
