Unlike standard factory robots, tomorrow’s autonomous manufacturing systems will have to be able to respond flexibly, appropriately and safely to new requirements and changing situations. With this in mind, a team of researchers led by Dr. Michael Golm of Siemens Corporate Technology in Berkeley, California, has developed software that generates virtual representations of environments to create a nearly limitless range of potential situations. The software then analyzes how different machines and systems respond to unforeseen circumstances.
The Future of Manufacturing — Inside Siemens’ Labs
Testing Robots’ Robustness
As factory robots gain autonomy, developers must ensure that unexpected events do not result in errors. Researchers at Siemens have developed an “Environment Generator” designed to achieve this.
Creating Situations that Humans Might not Think of
Known as an “Environment Generator,” the system “creates situations that humans might not think of,” says Golm. For instance, it might generate a situation in which a technician’s flashlight inadvertently produces reflections on a robot’s work piece. “The system looks at the robotic system’s model and the environmental model, selects a range of combinations of novel events, and creates tests designed to uncover potential hidden bugs.” If the robot’s sensing software is sufficiently robust, the robot will continue to function normally in spite of the anomalous reflections. On the other hand, if the robot were to stop, or did something it was not supposed to do, then this would be identified by the Generator and a programmer would have to go back and modify the robot’s code.
Now in its prototype phase, the goal of the Environment Generator is “to make Siemens products more robust in a very cost- and time-efficient way,” explains Golm.