Automated Polishing and Grinding _ Advanced Robotic Manufacturing
Comments are disabled for this video.
Polishing and grinding are important tasks in manufacturing. The current approach is to have
human workers manually do the job, which is time-consuming, expensive, and potentially dangerous.
Here we propose a robotic solution for automatic polishing to improve efficiency, safety, and
flexibility of the task. The system follows the state transition graph to perform measurement,
polishing, and final examination. The workflow is illustrated on a wild bead removal task
in the interior of a workpiece. First, the robot performs visual alignment to locate the target
workpiece. Then the robot generates a path for wild bead scanning and uses a laser scanner to
measure the wild bead. Once the scan result is obtained, the robot plans a polishing trajectory
and then executes the polishing trajectory through real-time motion control. As a last step,
the robot will take another laser scan of the interior of the workpiece to evaluate the polishing
performance. The robot motion planning problem is challenging. We need to deal with not only the
collision avoidance constraint between the robot and workpiece, but also the task constraint,
which requires that the end effector to maintain contact with the wild bead for polishing.
Existing motion planning methods either fail to consider these two constraints simultaneously
or cannot generate desired trajectories in real time. We propose a novel motion planning algorithm
using iterative convex optimization that can successfully handle both the collision avoidance
constraint and the task constraint. Most importantly, the algorithm is suitable for
real-time computation as it takes only 10 seconds to find a feasible solution shown in the animation.
For real-time motion control, we need to address different objectives. The trajectory tracking for
task performance, the compliance control for quality, and the collision avoidance for safety.
How to address these objectives simultaneously is the main challenge in the controller design.
A decomposition framework is proposed based on a global task coordination framework.
As shown in the figure, in the normal direction of the workpiece surface, we design a force
compliance controller to control the polishing quality. In the tangential direction and the
binormal direction, we design a tracking controller and a safety controller respectively.
These controllers are seamlessly integrated into the system and successfully addressed
the three control objectives. This project is a collaboration between CNU, Siemens, and Nisgawa.
We anticipate that we will apply this technology to real shop floors in the near future.

