Vision‐Based Tactile Finger Integrating Fingerpad and Nail Sensing for Robotic Manipulation

Abstract

Vision‐based tactile sensors offer rich contact information with simple optical hardware, but most existing designs focus on the fingerpad and treat nail‐like structures mainly as mechanical aids for grasping. This paper presents a compact tactile sensor that senses both the nail and the fingerpad with a single internal camera. Compared with our previous NailTact design, the fingerpad shape is changed from a quarter circle to a quarter ellipse, which reduces the sensor size and makes the sensor better suited to grasping small objects while supporting broader fingerpad contact during holding. The system also combines an external camera with a motorized mechanism that presets the inclination of both fingers relative to the tabletop before grasping. The preset is determined from the recognized object class and a representative height stored for that class. The nail part estimates force from marker displacement, while the fingerpad part estimates force and measures the contact position and in‐plane orientation of the contacted object, so multiple tactile cues can be acquired within a single fingertip. Experiments include a nail‐tip contact test with the table and grasping tests with objects of various shapes.