Design of human-machine system for estimating pattern of white cane walking

Mobility training uses a white cane for the blind or the visually impaired. Such training, which enhances their own walking ability, is different from an approach that enables them to walk with mechanical equipment. During training, one trainee has one particular trainer, and they work together. After the training, however, trainee anxiousness tends to increase, so some trainees can´t walk on paths or routes on which they previously walked. Since trainee/trainer interaction can offer psychological care, trainers must maintain contact with trainees and provide advice. In this study, a new support system is proposed that helps trainers estimate their trainee walking patterns so that they can offer advice without meeting in person when trainees get worried and lose confidence. The system has the following two characteristics: wireless compact acceleration sensors to gather white cane walking data and a pattern recognition technique whose study data are labeled by trainers so that the system can evaluate the trainer judgments and experiences. A subject experiment evaluated the system´s effectiveness with subject pairs comprised of actual trainers and trainees from the Kyoto Lighthouse.