The idea that a physical architecture reduces fragmentation of vision systems must be based on objective measures, not just assumptions.
What is the n= of this study for the Festo architecture, and what are the fragmentation thresholds measured before and after?
I don't see how a robotic arm, even sophisticated, directly unifies data protocols or API standardization.
For example, for image processing, using OpenCV or ROS has a much more observable impact on data consistency than mechanical design.