Abstract: In order to meet the needs for large structural component machining in situ, this paper presents a vision based approach for determining the relative pose between the workpiece frame and the base frame of a hybrid robot mounted on an automated guide vehicle(AGV). Drawing mainly on screw theory, the deviation between the poses respectively obtained by a laser tracker and by a 3D vision sensor is modeled as a linear function of the motion errors of the actuated and virtual joints. These motion errors are then approximated by the second-order response surfaces expressed in terms of nominal actuated joint variables, whose coefficients can be estimated by solving a multiple linear regression problem in an off-line manner. Consequently, once the pose of the workpiece frame relative to the vision frame is captured by vision measurements at one pose, the pose of the robot base frame relative to the workpiece frame can be determined in an on-line manner. Experiments on a prototype machine are carried out to verify the effectiveness of the proposed approach.

Key words: mobile machining robots, kinematic calibration, vision localization