Dynamic Pattern-Recognition-Based Walking-Speed Adjustment for Stable Biped-Robot Movement under Changing Surface Conditions

Stable, efficient and robust walking is an essential biped-robot capability since they are supposed to tackle all the challenges in real environments. In this paper, we propose the use of a pattern-recognition scheme based on measured force-sensory data from various walking surfaces for dynamic walking-speed adjustment of the biped robot. A walking-pattern-recognition accuracy of over 98% for is obtained on a rigid surface, ensuring the pattern recognition basing on force sensors as a valid indication for walking-speed adjustment.