Engineering & simulationsupervised machine learning200,000 devices deployed; 1.8 billion data points a month
The team used supervised machine learning on data from a smashed rail car to distinguish the specific cause of physical impacts, enabling them to determine whether damage was caused by crane operations or shunting yard procedures.
“we covered the rail car incensors, and we dropped things on it, and we derailed it, and we shunted them together, excessive speeds, and we really trashed it. But through that experience, we were able to do supervised machine learning onto our algorithms and we were able to actually distinguish the cause of the impact.”
Daniel MacGregor · Hear it at 9:13