Book Detection and Grasping in Library Scenario

Total or shared autonomous behavior is still an open challenge for service robots. An example of such system is FRIEND (Functional robot arm with user-friendly interface for Disabled people) (Figure 1). In the project ReIntegraRob care robot FRIEND will carry out all necessary manipulations for a completely paralyzed person who with support of FRIEND will be able to work in library. The task of the robot is to autonomously detect books on a shelf, grasp them and place them on a book holder and later place them onto the shelf again. The environment is cluttered since it includes book cart, books and book holder. The books have unknown size and color and no markers which could be helpful in detecting them. The 7 Degrees of Freedom robot arm receives environment information from overhead stereo and gripper mounted hand camera. The paper describes new approaches for detecting and grasping the book reliably. The proposed approach combines two algorithms for book detection and grasping and uses stereo vision together with hand camera to achieve a high rate of success. Keywords: Visual servoing, book detection, library scenario, planar segmentation, motion planning, grasping