IntroductionRoughness evaluation has been an important inspection for the quality of product surface or precision manufacturing in the industrial field. Conventional measurement devices are probe-type sensors with a stylus, but they are hard to be applied to curved surface or small product's surface. Human fingers are still necessary for such object surface evaluation in the industrial field since human fingers are flexible and applicable to narrow space and complex shape. However, the rating by human is subjective and is difficult to record and convey. Hence, this study is aimed to develop a tactile sensor capable of evaluating roughness on curved surface and small product surface as well as flat surface to replace the inspection by humans.Various tactile sensors and image processing technologies have been proposed for roughness or micro surface profile measurement. Image processing technologies have an advantage in wide area measurement, whereas they are not suitable for curved surface due to the limitation of height range (Li, et al., 2010). Many of tactile sensors for roughness or texture evaluation are active sensors, which include actuators for moving sensors and/or objects. For example, Sinapov et al. (2011) developed a tactile sensing system utilizing a robot arm with a three-axis accelerometer and showed that by applying several different exploratory behaviors on a test surface, the robot could recognize surfaces with different texture. In such tactile sensing system, robot arms/fingers with tactile sensors were controlled to Roughness evaluation by wearable tactile sensor utilizing human active sensing Abstract Humans can evaluate roughness on various shaped surfaces. Conventional roughness measurement sensors are difficult to apply to curved surface or small product's surface. In this paper, a simple tactile sensor utilizing human ability based on haptic bidirectionality is developed for the roughness evaluation. Humans can move their fingers while perceiving tactile sensations and change exploratory movements like contact force, scanning velocity, direction, etc. according to haptic perception and task objective. Our developed sensor is composed of two microphones and is mounted on a human fingertip. It allows users to touch the object without haptic obstruction. Users can apply the sensor while retaining their normal haptic perception and simultaneously obtaining vibrations and sound based on the mechanical interaction between the finger and the object. First, influence of contact force and scanning velocity on the sensor output is investigated. The experimental results show that the sensor output increases with a rise in the contact force but the influence of the scanning velocity varies between individuals. Then, on the basis of the results, experiment of roughness evaluation is conducted for flat surface and curved surface. A constant normal force and scanning velocity are exerted and the collected sensor output is calibrated by using the sensor output for the middle-roughness sample. The results...