BROWSE
Detail
Conference Paper
Recognizing Human Touching Behaviors using Neural Networks in Human-Robot Interaction
Cited - time in 
Share 
Share
- Authors
- Joung Woo Ryu, Cheon Shu Park, Joo Chan Sohn
- Issue Date
- 2006-10
- Citation
- International Conference on Ubiquitous Robots and Ambient Intelligence (URAI) 2006, pp.1-5
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- In interactions between humans and robots, touching is an important behavior which can provide humans with emotional safety. To data, however, most studies have been focused on interactions based on speech and vision. In this paper, a method of recognizing human touching behaviors is proposed for developing a robot that can naturally interact with humans through touch. In the proposed method, the recognition process is comprised of a preprocess step and a recognition step. In the preprocess step, a process of converting stream sensor data into the input data of the recognizer is designed and used for the efficient recognition. An input datum is classified as one of touching behaviors by using a neural network called multi-layer perceptron (MLP) in the recognition step. This paper presents the preliminary results of classifying the 3 basic human touching behaviors of 'stroke', 'pat', and 'tickle'. As a result, an average recognition rate of 95.65% could be obtained from input data generated by the touching behaviors of one person.
- KSP Keywords
- Human-Robot Interaction(HRI), Recognition rate, input data, multilayer perceptron, neural network, preliminary results, sensor data