Ahuja receives T. A. Stewart-Dyer and F. Trevithick Prize
ECE Professor Narendra Ahuja was recently honored with the T. A. Stewart-Dyer and F. Trevithick Prize for his work on the use of computer vision for railroad engineering. The award, which is presented annually by the Institution of Mechanical Engineers, recognized Ahuja’s work on an interdisciplinary project on automated visual inspection of the loading of intermodal freight trains to improve their energy efficiency. The Illinois researchers on the winning team are from the Department of ECE, the Coordinated Science Laboratory, and the Beckman Institute (Ahuja, John Hart), and the Department of Civil Engineering and the Railroad Engineering Program (Professor Chris Barkan and students). Other participating institutions are IIIT, Hyderabad, India (Professor P. J. Narayanan and colleagues/students), and BNSF Railroads, the corporate user as well as sponsor of the project (Mark Stehly and colleagues).
The project involves development and deployment of an automatic train-inspection site. “Our vision system gets automatically triggered by approaching trains, inspects them for loading efficiency--to minimize drag--and prepares loading reports for railroad administration,” said Ahuja, a Donald Biggar Willet Professor of Engineering.
“It’s a great feeling to win the award because this project is not only academically rewarding, but also provides short-term benefits,” said Ahuja. The analysis it performs on a daily basis is of direct and immediate value to railroads, in helping improve their day-to-day operations.”
Working with the Railroad Engineering Program in the College of Engineering, Ahuja’s research has focused on tasks that are normally performed by human experts in the railroad industry. “We’re trying to improve the execution of tasks that are often critical but extremely demanding on human workers, in terms of hours, attention, and sustained performance at the required levels of reliability,” said Ahuja. “We are either automating the execution, through a combination of cameras and computers, or developing assistive tools that would carry out a lion’s share of the work, while identifying challenging cases and forwarding them for handling by the more experienced, human inspectors.”
Ahuja and his team have worked to design computer vision software that works on a variety of different necessary functions. By positioning cameras in specific places alongside train tracks, they collect the video of a moving train, and utilize computers to analyze the load that the engines are pulling. Because of the massive weight of a train’s freight and the long distances that they travel, small imperfections in the way a train’s load is pulled can have dramatic effects on the train’s performance. “The efficiency of the fuel a train runs on depends on wind dynamics and the flow of air between the train cars,” said Ahuja. “Professor Barkan and his students have developed a method to evaluate the dynamics of a loading pattern. If you load the train better--and this may mean shifting the loads by just centimeters--it can add up to hundreds of gallons of saved fuel over a coast-coast run.”
The new computer vision software is essentially a load inspector who never gets tired, takes sick days, or forgets to check up on something.
Ahuja’s work has already started to have an impact on the railroad industry. One trackside inspection station has already been installed on a busy BNSF line in Joliet, near Chicago, and according to Ahuja, BNSF Railroads are in the process of installing another analysis station on one of the largest national railway crossroads, a BNSF line in KansasCity, Missouri. And Ahuja is hopeful that his system will become an industry standard. “The industry is such that if one company benefits from technology, than everyone has to do it or they will be less competitive,” he said. “The award is nice but the impact the system is capable of making on the performance and functioning of railroads is truly rewarding.”
The T. A. Stewart-Dyer and F. Trevithick Prize consists of a certificate and a cash prize of up to £1,000. Professor Ahuja is also a recent recipient of an HP Labs Innovation Research Program Award for his project “3D Reconstruction of Dynamic Real-World Objects and 3D Motion Aided Gesture Recognition.”