This project took place at the German Aerospace Center (DLR) in Munich, in conjunction with Technical University Munich. The aim of this project was to program a DLR light weight arm (LWR III) to do some task in the medical field (e.g. prepare medicine), and provide a human friendly interface for users to interact with the robot.
A combination of kinect and projector projecting onto a tabletop was used to provide an interface the user could interact with. The camera from a second kinect was used to perform facial recognition to identify the user. The user would then be asked to place his mug on the table, which would be localized using the kinect. The robot arm would then fetch the correct medicine and poor it into the mug on the table.
ROS was used in the development of this project. OpenCV and the eigenfaces method was used to do the facial recognition. The Pointcloud library was used to localize the cup on the table. Bubbles and python was used to program to high level state machine coordinating the entire process.
The practical application of this project could be used to mix or prepare medicine for patients, to avoid human errors in administering the incorrect medication.