Complications arise when a vision system on a robot cannot acquire an image of the end effector. When this happens, the robot has no information on which to make small adjustments. A clever solution to the problem is to acquire an image of the robot arms and use that to control the end-effector position. The position adjustment based on an arm observation is equivalent to controlling a hidden robot in the controller. The concept is simple but the execution is not.
Researchers at the Institute in Communications and Cybernetic of Nantes (IRCCyN) are addressing the problem with Adams multibody dynamics simulation software. They are validating the theoretical works done on a visual servo system to control a parallel-arm robot based on observation of the arms.
Most robots are controlled through the use of encoders that measure joint rotation. But even when encoders with high levels of accuracy are used, the ability of robots to move to an absolute XYZ position and ABC orientation is limited by deflection, thermal expansion, and manufacturing variation. Some applications, such as placing the read head of a disk drive during assembly require higher levels of positioning accuracy that can only be achieved with an expensive, special purpose robot. This positioning challenge is being addressed with visual servoing technology that uses a vision system to acquire an image that determines the relative positions of the robot end-effector and the target. Visual servoing can achieve placement accuracies of just a few microns without requiring an expensive robot.
IRCCyN used Adams simulations to validate this concept of a “hidden robot model” to improve visual servoing accuracy. The hidden robot is a virtual robot with kinematics to represent the mapping between the arm direction space and the end-effector position and orientation space. The researchers decided to attempt to determine a general method to define the hidden robot model for any type of parallel robot controlled by visual servoing based on observation of the arms.
“Adams simulations have played an important role in validating our theoretical work on hidden robot models,” said Sébastien Briot, Researcher at IRCCyN. “The integration of Adams with Simulink through Adams/Controls eliminated the need for us to write complex equations for predicting the dynamics of parallel arm robots. It also provided graphical results that gave us a better understanding of robot behavior.”
The Research Institute in Communications and Cybernetic of Nantes is a scientific organization linked to the French National Centre for Scientific Research. Its purpose is to innovate in several fields, among which include robotics, automatic control, production theory and image processing. For more about IRCCyN: http://www.irccyn.ec-nantes.fr/fr/
Adams software is developed by MSC Software, a company that helps product manufacturers advance their engineering methods with simulation software and services. For more: www.mscsoftware.com