Texture roughness estimation using dynamic tactile sensing

摘要: Roughness estimation can help with improving tactile prehension and distinguishing slippage events during object manipulation a robotic hand. Humans are able to estimate roughness from small contact area an object, adapt strategies using this information [1]. In order do the same hand fitted sensors, article focuses on how data sensor. We propose learning algorithm that estimates scale 1 5, which was inspired by human capabilities. For more adapted parameters values, is optimized genetic algorithm. To initialize scale, we asked 30 people classify 25 textures 5. The results were used feed After testing our those textures, conclude even if there errors certain itself new provide approximates one.