GCCF : a generalized contact control framework
MetadataShow full item record
The field of robotics has come a long way since the first reprogrammable robot was able to automate simple tasks on an assembly line. However, many industrial robots are stuck doing similar simple tasks in the field, especially in the nuclear industry. Roboticists can expand the task space of industrial robots by making advanced robot technology reliable, easily integrated, and packaged in a manner that does not require an expert in the field to use. One particular field of robotics that could be used to help this task space expansion is compliant control which is used to execute robotic procedures involving contact with environmental objects. It is especially useful when the position or orientation of the environmental objects is not precise. Examples of industrial procedures that a robot could do with compliant control include material reduction, surface finishing, packaging, assembly, material handling, and many more. This thesis explores the state of the art in compliant control and proposes a Generalized Contact Control Framework (GCCF) that packages compliant control laws in a manner that is easy to use for a non-expert. GCCF splits the control of a robot end effector into separate control of each linear and rotational dimension. The user sets the law that controls each dimension independently to one of three intuitive laws. By specifying laws and stiffness independently for each dimension of end effector control, the user can complete a large variety of contact tasks. We illustrate GCCF’s broad capabilities in two flexible demonstrations. The first demonstration provides a graphical user interface to GCCF with which a user can set and reconfigure the control of the end effector while interacting with the robot. This allows the user to subjectively experience the reconfigurablilty as well as the physical behavior prompted by the control. In the second demonstration, we use GCCF to execute multiple contact tasks with the goal of putting a peg in a hole. These demonstrations prove the feasibility and usefulness of GCCF, using the API and ROS compatible package for the controller.