Detecting track problems
Artificial intelligence could allow rail operators to more accurately detect issues on their tracks, creating a safer and more reliable rail network. GE chief technology officer Wesley Mukai discussed some of the firm’s technology with VentureBeat, explaining how AI-enabled locomotives could deliver massive improvements in efficiency.
GE’s current “smart freight” locomotives are equipped with a set of high-definition video cameras that continually capture the track ahead of the train. A densely architected onboard datacentre processes the camera signals, using AI to find physical defects on the rails.
If a deformation is identified, it’s analysed in real-time to determine whether it could derail the locomotive. If it’s found to be a risk, a warning can be issued before the train’s automatically slowed down.
More efficient rail networks
Beyond identifying track problems, automation offers rail operators a way to streamline their network and eliminate inefficiencies that represent financial losses. A 1mph rise in train velocity translates to a $2.5 billion gain in industry value. Reducing locomotive stationary times can offer even more lucrative savings.
READ NEXT: NVIDIA launches license plate-sized chip for self-driving cars
GE’s applying AI to a range of rail network issues. It wants to create a more effective industry that runs efficiently and reliably.
AI could determine the optimum schedule for different services, monitor locomotives as they run and predict failures before they occur. This all leads to substantial economic benefits and productivity gains for the entire industry.
Optimisation through assistance
Some of the AI’s features are focused on assistance rather than all-out automation. One of GE’s software tools helps the driver to save fuel by reminding them of simple best practices.
As the locomotive ascends a hill, the AI will prompt the driver to ease off the throttle before they reach the crest, letting momentum pull the train over. Mukai told VentureBeat that basic optimisations like this could save “tens of millions of dollars” worth of fuel.
In the short term, it’s systems like this that will be adopted first by rail operators. Autonomous networks may not be too far off though. Self-driving shunting locomotives are already being used in some rail yards, replacing human drivers for low-speed operations. An expansion onto mainline tracks is the logical next step, once the technology is in-place to facilitate full onboard AI processing.
Detecting track problems