Satellites are like any other technology. At some point, they break.

If the problem is software related, you can upload a patch or a fix.

But if you need to change up the hardware on an object that is tens of thousands of miles above the Earth, that usually means that the satellite is done for good, says Ou Ma, an engineering professor at the University of Cincinnati.

"They don’t fix it. That’s it," Prof. Ma told Tech Briefs.

Astronauts on the shuttle have managed to make repairs to satellites in low Earth orbit. The objects higher than 1,200 miles, however, are often left alone.

But what if you could send up some help to the satellites in high Earth orbit — like, say, another satellite?

"The industry now is going in the direction of launching some kind of service satellites to orbit and capture a malfunctioning satellite and then do some repair using some robotics technology," said Ma.

In the near future, Ma envisions a service orbiter being able to connect with a faulty one to provide refueling or minor repairs like battery replacement.

NASA's Restore-L  mission, set for 2020, will launch satellites that attempt to fuel up other satellites in low Earth orbit.

Before you can repair a satellite however, you have to catch it.

Ma's Intelligent Robotics and Autonomous Systems Lab is currently doing just that: developing algorithms to identify target satellites.

The professor and his students are testing the algorithms using a variety of tools: simulation software, a shoebox-sized robot, and a setup that almost looks like an air hockey table. The frictionless, granite surface and pressurized air help to mimic the microgravity environment of space.

The robot, guided by actuators and eight directional thrusters, contains range sensors, cameras, and an inertial measurement unit to identify the object of interest.

A University of Cincinnati robot in Ou Ma's lab tests robotic capture and repair of a satellite
UC aerospace engineering students work on a robotic hovercraft that floats on a cushion of air. The lack of friction helps mimic conditions in space. (Image Credit: Andrew Higley/UC Creative Services)

The models in Ma's lab also feature industrial-sized robotic arms with seven joints and a full range of motion.

In an edited interview below, Prof. Ma tells Tech Briefs how soon he expects to see service satellites going from the test spaces of Earth to the highest orbits above it.

Tech Briefs: How has a satellite traditionally been fixed?

Prof. Ma: In most cases, if there's a hardware problem, they just leave it. You cannot do much about it.

Sometimes if you have an antenna which is supposed to be deployed or unfolded, but it gets stuck due to friction, you might command a satellite to spin; the centrifugal force may possibly spark this mechanism to open. It's very unlikely to work.

There are many examples of repair using human assistance. Hubble , for example, has been fixed several times. Intelsat VI was fixed in one of the shuttle missions . Those satellites, though, are low Earth orbit, and can be reached by shuttle, by human. All of the other satellites, especially the big ones in high orbits, geosynchronous orbits, are very expensive. If anything goes wrong with these satellites, [space agencies] cannot help them. It’s difficult.

Tech Briefs: How do you envision robotic satellite repair to work?

Prof. Ma: In five to ten years, there will be some real fixing missions — maybe starting with just a refuel, because the satellite needs fuel to stay in orbit. NASA and DARPA are developing ways to send another satellite to space, catch the older satellite, and transfer the fuel such that the lifespan of the older satellite can be extended for another few years. That would be the first thing. I would say that kind of demonstration could happen very soon — a few years.

Tech Briefs: How long do you think it will take to have robotic satellites fixing hardware?

Prof. Ma: The replacement of broken parts would be even longer — maybe 10 years. Some of the hardware problems will be difficult to fix, because many older satellites were not designed for servicing. New satellites, however, have been designed so that they can somehow be serviced in the future.

an illustration of the NASA Restore-L mission
An illustration of NASA's Restore-L mission to launch satellites that can repair or refuel other satellites in low Earth orbit. (Image Graphic: Maxar)

Tech Briefs: What are the kinds of satellite hardware problems that robotic satellites would be able to someday repair?

Prof. Ma: The battery of a satellite can last about 10 years. After that, the battery either dies or significantly deteriorates. So, with a robotic arm, replacing the battery and plugging in a new one would be a really easy fix. Anything to be replaced or repaired by a robot, however, has to be designed in a way to be relatively easy to take apart and remove from its main structure. If you bolted it very strongly, then it might be difficult.

I’m a robotics researcher. I know that robots currently are still not very capable. They can do some tasks, but not as well as a human worker. Their capabilities still need to be improved. The capabilities right now are only good for very simple repair. You open something; you unplug something. Even having a robot unscrew something would be difficult.

Doing repair in space is not like in the garage; it’s very difficult. Nothing holds you up. If the robots do the work unsuccessfully, it’s floating up in space. Even a tiny applied force can create quite a drastic motion unexpectedly.

Tech Briefs: What is the work that your lab is doing to enable robotic satellite repair?

Prof. Ma: For a robot to repair a satellite, it first has to capture it. Space capture is not very easy. Satellites are spinning and tumbling. We are developing very basic technology and control algorithms to capture a rotating object in space.

Before you capture something, you have to first know the size, the components, the texture, the mass distribution. You may want to dispose of a bad satellite or space debris so that it doesn't collide with good satellites. If the object that you want to service is a new object, a piece of space junk, for example, the first thing you have to do is understand the target. You can use a camera to determine the size and how the object is moving. We try to identify the mass and mass distribution of each object through the camera, so when you plan robotic capture, you can do the job very smoothly.

Tech Briefs: What is most challenging when you’re trying to design these algorithms to identify objects?

Prof. Ma: The first challenge is the sensors. When you identify an object, you try to have situational awareness.

But space is a very harsh environment. When there is reflection from the Sun on the surface of the satellite, the camera will not "see" the object, just like with a human’s eyes. Sometimes it’s completely dark in space; during the nighttime, you don’t see anything.

Tech Briefs: What about the ability to grasp? How is one satellite able to grasp the other?

Prof. Ma: That’s the most difficult part – physical contact and interception. If the robotic arm doesn’t touch the target properly, if the target is rotating, you can make it tumble, or make yourself tumble.

You’re just floating in space. When the servicing satellite and the bad satellite intercept and physically touch each other, that moment is very critical. The physical-contact operation has to be done very very carefully and gently.

Tech Briefs: How do you test out these algorithms?

Prof. Ma: It’s a very difficult test, since we're on the ground. In a space environment, there’s no gravity.

There are several ways to test the algorithms. First, computer simulation. Simulation relies on your knowledge of the model and the object. You have to use very high-fidelity simulation on the computer, to simulate the operation. This physical contact is a big problem. I did a lot of research on how you can model the impact to make a more accurate simulation.

We also have an air-bearing table in the lab. It’s a flat granite table. The granite is very hard and doesn’t deflect or change shape over time; it’s very stable and smooth. You have pressured air go into the table, much like the table you use in air hockey. You have fresh air coming out; we can put the satellite model on top of the air table, and it can float without friction. You push the object, and it can easily move around and rotate. This is a 2D test.

Tech Briefs: How will the information gathered about satellite navigation be used to support NASA missions?

Prof. Ma: Our research is all fundamental research, so NASA, engineers, and industry can look at it and apply it. We do some research for NASA. Most of their major projects, like going to back to the Moon or exploring Mars, are contracted to industry and the big companies. We at the university develop individual pieces of technologies. Industry is integrating these technologies and applying them to the real situation. We are on the different side of the coin.

We tackle the most difficult, unsolvable problems – the contact/impact problem, or how to capture a tumbling object, for example. There is no one in the world who has even tried to capture something tumbling in space. Hopefully in the future, that kind of capture will become a reality.

What do you think? Share your questions and comments below.