| Summary: | 碩士 === 國立成功大學 === 機械工程學系碩博士班 === 98 === The robot technology contributed much to the growth of manufacturing industries. In recent years, however, it is extending the activity from the manufacturing site to our daily life. The objective of this thesis is to create a feeding humanoid robot to support those who are unable to eat independently. The constructed feeding humanoid robot comprises mainly an omni-directional wheeled base, a fixed torso mounted on the mobile base, two seven-degrees-of-freedom arms, two robot hands and a head equipped with a panoramic camera and a pan/tilt/zoom camera. In the task of feeding, the humanoid robot moves in two stages. In the first stage of movement, a vision-guided control strategy is first utilized to determine the pose of the meal tray to be carried, and a developed intelligent coordination impedance controller is then applied to drive two robot arms to carry the meal tray. In the second stage of movement, the same vision-guided control strategy is first employed to find the pose of the human mouth to be fed, and the right arm of the feeding robot is then controlled to grasp solid foods in the meal tray and feed the person with the foods. The utilized vision-guided control strategy is based on the back-propagation neural network, which is applied to achieve the mapping between image features of the target and the pose of the target. Finally, a set of experiments are conducted to verify the performance of the constructed feeding humanoid robot.
|
|---|
