Collecting robot training data is dirty, unglamorous work. Some AI labs are already paying XDOF to do it.

# Training Robots Requires Massive Amounts of Real-World Video—and It's Expensive Building AI robots that can actually work in the physical world requires thousands of hours of video showing how humans perform tasks, but collecting this footage is tedious and costly work that most AI labs would rather outsource. Some leading AI companies are now paying specialized firms like XDOF to handle this unglamorous data collection so they can focus on building better robot brains. The bottleneck isn't clever algorithms anymore—it's getting enough real-world examples to learn from.