Researchers at Queen’s University in
Currently, when the high-flying celestial objects malfunction – or simply run out of fuel – they become ‘space junk’ cluttering the cosmos.
But because they are many thousands of kilometres away, the satellites are beyond the reach of an expensive, manned spaced flight, while Earth-based telerobotic repair isn’t possible in real time.
To help address the issue, Queen’s electrical and computer engineering professor Michael Greenspan is developing tracking software that will enable an autonomous space servicing vehicle (ASSV) to grasp the ailing satellite from its orbit and draw it into the repair vehicle’s bay.
Once there, remote control from the ground station can be used for the repair.
‘The repair itself doesn’t have to be done in real time, since everything is in a fixed position and a human can interact with it telerobotically to do whatever is required,’ he said.
The Queen’s team is now working to develop the ASSV with the aerospace company MDA (McDonald-Detweiller Associates) Space Missions, which earlier built the Canadarm and has been responsible for all Canadian systems in the international space station.
Developing a computer vision system for the job is challenging, Professor Greenspan said.
Since the objects circle the globe in ‘geosynchronous’ orbit, their speed is synchronised with the Earth’s rotation.
The robotic system must recognise the satellite first, then determine its motion and match that motion before grabbing it.
Due to the harsh illumination conditions in space, conventional video cameras are of limited use.
The preferred sensor is a form of light-based radar called LIDAR, which provides a set of 3D points that accurately measure the surface geometry of the satellite.
The Queen’s team, which includes electrical and computer engineering graduate students Limin Shang, Babak Taati and Michael Belshaw, has developed software that allows such a system to identify a satellite, determine its position and finally track it in real time.
They have recently received funding from the Natural Sciences and Engineering Research Council (NSERC) to continue looking at fundamental aspects of the technique.
Another potential, terrestrial application of their findings is in the area of flexible manufacturing, said Professor Greenspan.
Using vision systems and algorithms, objects can be recognised and tracked as they go down a conveyor belt or assembly line.
‘Once you can do that, automated manufacturing systems can interface much more flexibly with the objects,’ he noted.
‘The result will be a much easier and more cost-effective manufacturing process.’