Surface Recognition and Speed Adjustment of Humanoid Robot Using External Control Circuit

Abstract

We propose a surface-recognition-based speed- adjustment system for humanoid robot during walking on different surfaces. Two different types of surfaces are considered in the reported experiment (example, rough and smooth surface). Force-sensors, classification unit and external- controller-circuits are applied for surface-recognition and speed-adjustment. For surface-recognition the Euclidean distance is used to calculate the nearest-neighbor reference pattern for the feature of the waking pattern generated online. The mean-absolute-value (MAV) feature vector is used to classify two different surfaces. To distinguish two different surfaces, the hardware accelerated decision-signals are generated across LEDs in the form of analog voltages (maximum peak voltage 212 mV for rough-surface and 147 mV for smooth surfaces respectively with detection time 2.8 s and 1.5 s). The external-controller-circuit is used for speed- adjustment using decision-signal coming from LED. It is observed that, when robot is moving from rough-surfaces to smooth-surfaces, the speed of the robot motion changes from 190 frames/stride (i.e. slow-speed) to 160 frames/stride (i.e. medium-speed) with 4.9 s transition time, whereas from smooth- surface to rough-surface transitions, the transition time takes 4.5 s. The experimentally measurement results of speed- adjustment time after surface transition are useful for fast and stable recognition-system design.