Haptic exploration of unknown objects

摘要

Haptic exploration is a key mechanism humans use to learn about the surface properties of unknown objects. With specialized fingers and sensors, and the appropriate planning and control, robots can also be enabled to explore the world through touch. Haptic exploration has applications in many areas, including planetary exploration, undersea salvage, and other operations in remote or hazardous environments. This thesis develops an approach for haptic exploration of unknown objects by robotic fingers. Because haptic exploration is coupled with manipulation, a procedure for combined manipulation and exploration using a sequence of phases is presented. Fingers alternately grasp and stabilize the object while other fingers explore the surface with rolling and sliding motions. During an exploratory phase, the goal is to move a finger's tactile sensors over the surface in a way that will elicit useful data. There exist many possible objectives for haptic exploration. This work concentrates on the detection and identification of fine surface features. In the context of exploration with spherical robotic fingertips, fine surface features and macro features such as bumps, cracks and ridges are defined. Using different types of sensor data, various algorithms and experimental results for fine feature detection are presented. There are also many potential methods for actively exploring a feature on a surface in three dimensions. After a feature has been encountered on a surface, tactile sensor and position data may be used to determine the next direction of finger travel, guiding the finger around and over the feature in a way that will efficiently extract surface properties, Shape skeletons are used to create a map of features and regions on a surface.