Giving part of a robot's AI less information helped it learn a new task
What to look for
An OpenWAM experiment suggests that stripping context from a motor component can make it easier to reuse.
Read the full analysis: Giving part of a robot's AI less information helped it learn a new task ↗