返回Google圖書搜尋

Intelligent Pose Control of Mobile Robots Using an Uncalibrated Eye-in-Hand Vision System

出版	INTECH Open Access Publisher, 2006
ISBN	386611284X9783866112841
URL	http://books.google.com.hk/books?id=tD2xoAEACAAJ&hl=&source=gbs_api

註釋Mobile robots commonly need a guidance control system to navigate the mobile base and a method for performing the pick-and-place operations. The guidance control system and the non-planar ground inevitably cause the position and orientation errors of the mobile base. Therefore, this study proposes a behavior-based control strategy that employs an uncalibrated eye-in-hand vision system to control the end-effector of the manipulator to approach and grasp the target workpiece. All the designed behaviors are defined from the perspective of the camera and are mediated through fuzzy rules with a look-and-move control structure. The presented neural fuzzy controllers map image features in image space to relative motion commands in the camera frame. These motion commands are then transformed to the end-effector frame by the proposed rough motion transformation. This work differs from the references (Wasik & Saffiotti, 2002, 2003) as follows. (1) A backpropagation algorithm is used to reduce image feature errors through the adjustment of the singletons of the consequent parts in the fuzzy singleton rules. (2) Two additional image features, the ratios of the lengths of the two pairs of opposite sides on the quadrangular image, are extracted to guide the rotation of the camera when the workpiece has a tilt angle. (3) The control values defined in the camera frame generated from the controller in (Wasik & Saffiotti, 2003) are transformed to joint controls by a Jacobain transformation, revealing that either a hand-eye calibration process has been implemented or the hand-eye relationship is known beforehand. However, this study proposes a rough motion transformation to replace the time-consuming hand-eye calibration. Only a general ruler and naked eye are required to estimate hand-eye configuration. The inaccuracy of this estimation can be ignored by adjusting on-line the singletons of the consequent parts in the fuzzy singleton rules. (4) A more human-like control strategy is.