The railroad industry expends significant effort to increase operational safety and efficiency by using a variety of sensors for machine health monitoring and inspection purposes. This paper proposes to advance rail technology even further to use similar sensor data for the control of a robotic system designed to automate the uncoupling of freight cars, a hazardous operation that currently requires human operators to interact with moving trains. To automate this process, an intelligent robotic system was developed to detect, track, approach, grasp, and manipulate semi-constrained objects on equipment in motion. This work presents a system prototype that utilizes machine vision, force feedback, and complex end-effector technology capable of autonomously uncoupling full-scale freight cars using visual and tactile feedback. Laboratory tests have proven that modern robotic and sensing hardware can be used to reliably separate pairs of rolling stock at 3.25 kilometers per hour. The results to date suggest that speeds of up to 7 km/h are feasible for a system deployed in a rail yard.

This content is only available via PDF.
You do not currently have access to this content.