Computer Vision-Based System Gauges Fuel Levels

Michael D. Wheeler

Most conventional fuel gauges rely on a sensor with low-voltage electrical leads that come in contact with the fuel, which some suspect have contributed to recent explosions on airplanes. Former graduate student Srivatsan Chakravarthy and two of his professors at Pennsylvania State University theorize that a much safer method would be to use a form of computer vision to monitor fuel depths.

The researchers customized a fuel tank by equipping it with two glass portholes on top. Through one port, a light source -- in this case a red laser pointer -- flashes a crosshair pattern on the surface of the liquid inside the tank. An off-the-shelf CCD video camera attached to a comput-er records the results through the other port.

Laser triangulation

The image processing software designed by the researchers is calibrated by filling the tank slowly and noting the position of the pattern of the laser pointer at various depths. The digital map shows the correlation between these depths and the three-dimensional scene. Hence, when any of the pixels of the crossed lines change, the computer calculates the change of depth of the fuel in the tank.

"We used the principle of triangulation -- in which the position of a projected light pattern on the fluid surface is measured in the captured image -- to determine the depth of fluid," said Rangachar Kasturi, a professor of computer science and engineering, and one of Chakravarthy's advisers on the project. "The position of the reflected light pattern in the captured image changes as a function of the depth of fuel. This is because the light source is placed at an angle with respect to the surface normal of the liquid. Hence, there is a direct relationship between the fuel depth and the detected position of the line pattern in the image."

Although the principle of triangulation for depth measurement is well-known, Kasturi said that he is not aware of any other effort to use this procedure for measuring fuel levels. The team performed its tests with a stationary system, but the researchers say their approach can be adapted to systems affected by turbulence or vibrations. They are interested in exploring opportunities for commercializing the system.