External Perception for Locomotives: Systems and Algorithm Development
The Federal Railroad Administration sponsored a research team from Aurora Flight Sciences to develop the External Perception for Locomotives (ExP-L) system, which uses the FRA Cab Technology Integration Lab (CTIL) to demonstrate prototype object operator augmentation. The system can detect and localize objects of interest in and around the track, interpret the state of the detected objects to determine if there is any urgent and actionable information, and communicate this information to the operator. Similar to adaptive cruise control in modern cars, the system provides information to the operator allowing them to actively monitor the environment, fight distraction and fatigue, detect potential obstructions, and determine the best course of action. This algorithm achieved 71 percent accuracy while running at 10 frames per second in FRA’s CTIL. The approach and lessons learned from this program can be leveraged to transition the system for use in an operational locomotive. The existing CTIL-based system also can be used for human factors research to refine the human/machine interface to a point where it is most useful for engineers and conductors.