Several years ago, U.S. Air Force maintenance technicians at Tinker Air Force Base identified a potentially dangerous condition in the fuel tanks of B-52 bombers when they came in for depot maintenance. A new additive designed to remove water from traditional JP8 jet fuel was unexpectedly causing paint to peel off the inside of the tanks. The main concern was that the paint chips could clog the fuel filter and cause an engine flameout.
All 96 active and reserve bombers were experiencing the problem, a development that forced the USAF to undertake a laborious and time-consuming process whereby the fuel-tank cells are hand scraped and touched up. Such a job, however, is tailor-made for automation, particularly industrial robots.
The USAF is now working with the nonprofit Battelle on just such a solution. Called the Multi-Use Robotic System (MURS), the technology has already been demonstrated to conduct autonomous nondestructive inspection and is now being improved to potentially tackle challenges such as the paint problem.
“When we talked to the folks at Tinker, they were skeptical because they had robotics manufacturers in there trying to solve this issue,” said Tim Lastrapes, MURS Program Manager. “The whole premise behind our project was to determine the requirements for working inside the tank and then develop as benign a system as possible. What I mean is that there are currently guys inside the tank doing the work. Our job was to figure out a way to give them a different tool so that the same guys can sit outside and run the robot from a remote control station.”
The robot itself is basically an off-the-shelf small-arm painting and welding robot from California-based Denso Robotics. It was modified with a 3-D laser scanner and mounted on a track for movement. It has a mass less than 30 lb and can be installed in 15 min.
“Because of foreign-object damage, you can’t take any kind of tools inside the aircraft fuel tank if you can help it,” said Lastrapes. “The technician only takes in the parts of the robot, and because it is a tool-less assembly, he doesn’t have any tools to leave behind.”
The remote control station has two screens: one that shows a simulated, rendered version of the robot in the tank and another that displays auxiliary video feeds coming from the robot and from optional cameras that can be clipped in the tank.
Battelle’s biggest engineering challenge on MURS is the software.
“Robotics is always a mechanical challenge up front, but capabilities is almost always a software challenge,” said Darren Krasny, Software and Controls Lead for the program. “The key to making industrial robots work is giving them preprogrammed paths and in having access to computer-aided geometry that describes the section that the robot is working on. The difference with our robot is that it has a scanner that can define that geometry without having access to proprietary CAD geometry.”
Conducting motion planning through point-and-click technology is one of the key challenges being addressed by the MURS team.
“If you think about a six-degree-of-freedom robot, and if you were going to put a sensor at the end of each of those degrees of freedom, you would have feedback that would tell you if the robot was getting close to the wall,” said Scott Newhouse, MURS Technical Lead. “When we started with that simple solution, we ended up with a few hundred sensors that were needed to feed every direction in which the robot was moving.
“We had to look at it from a different direction, and the way we solved that problem was a proprietary software solution. But it’s an effective way of using a robot in a confined space, and because we wrote it with an open architecture we can now use anybody’s robot and interface with their control system. And by having a three-dimensional sensor on the end of the robot that can look around and understand its environment, we can take this robot and do just about anything with it, including mounting it on a vehicle and looking at an IED on the ground without running into the explosive,” said Newhouse.
The MURS system that was originally demonstrated for the USAF is now back with Battelle for improvements. The electronics are a few years old, so they’re being updated. The team is also looking to improve the speed of the robot.
“The robot was not optimized to do continuous trajectories that might be required to paint, strip, or recoat the inside of a tank or the outside of an aircraft; so that’s one capability we’re looking at,” said Krasny. “In addition, the sensor we purchased for this has incredible resolution but is pretty big. The robot itself has trouble getting into really tight places because of the size of the head, and that turns out to be a bottleneck in terms of positioning.
“We’re also looking for another sensor that can give us feedback at video frame rates so we can actually capture the environment in real time, which eventually could lead to being able to scan in the environment, even a dynamic one where a new obstacle appears or possibly a human enters the picture. These are the kinds of things we’re considering,” Krasny said.