bobhaha
0
- Joined
- May 31, 2009
- Messages
- 3,239
- Points
- 63
So a couple of days ago.. My girlfriend and I got to play in the Mars rover yard at Sydney's Power House Museum. We were testing the accuracy of a camera system that will be implemented the next Mars rover (If they make another...).
So basically this is the camera...
Inside the shell is a High definition camera, 6 axis accelerometer, some sort of intel processor, and a heating unit. It weighs about 58 grams and consumes 1.7W of power. Update rate of 180Hz with a positional accuracy of 2mm error.
Basically what it is going to do is pick out constellations in the sky and track its X,Y,Z coordinates within 2mm error of where it is on Mars. It also has the ability to track its Yaw, Pitch and Roll coordinates as well. The most brilliant thing about this camera is its ability to track it's self with approx 5mm of positional error without any outside stimuli after it has picked out a constellation and calibrated its self.
So this was the yard we were testing on...
It's not exactly big... but it pretty much matches the type of terrain the rover will be experiencing.
As you may notice... there is a small prototype of a solar powered rover in the shot... We didn't get to play with the big rover cause it was being used by some people who are going for their PhD in astronautical engineering.
Here are some close ups of the prototype that the first year students get to work on. There have been two made so far...
2011 model
2012 model
Another interesting thing that was at the exhibit was actual celestial rocks. They are extremely expensive and we were instructed to not touch them.... I'm talking 50k+ per rock.
These two rocks...
And this one...(sorry for shaky pic)
So anyway... the experiment called for 250 data points along the terrain. We needed a known position to measure from, so we can cross reference our positions and find the cameras accuracy.
So we used one of these bad boys...
Built by Nikon, it's a X,Y,Z position calculator. It works off the response time between a shot of IR (from a laser) and the time it takes to come back. It uses this pretty cool reflector... Which I forgot to take a shot of, but is approx 2mm in dia. Pretty cool technology.
So this is my girlfriend setting up one of the platforms for the camera to sit on. You can see our data point positions marked by a stick in the ground with a piece of paper, stating all data collected from the camera and the Nikon XYZ camera.
As she is setting up the cameras position... I am remotely checking its X,Y,Z positions on a laptop and comparing it to the Nikon cameras calculations...
We were simulating the constellation with positional fiducials that were on the roof of the exhibit...
All in all we measured an averaged error of 2.6mm in the camera. We also concluded that its displacement calculations without outside stimuli was an averaged 4.4mm error.
We also found that in one section of the exhibit there was errors with the fiducial markings giving us errors of up to 4 meters. We pointed that out to the exhibit director and within one day they had fixed it and emailed us a written thank you...
It took a painstaking 9 hours to do a complete study of the camera in this terrain but it yielded some very nice results. Hopefully we get to play with the real mars rover soon.
Hope you enjoyed this.
-Adrian
So basically this is the camera...
Inside the shell is a High definition camera, 6 axis accelerometer, some sort of intel processor, and a heating unit. It weighs about 58 grams and consumes 1.7W of power. Update rate of 180Hz with a positional accuracy of 2mm error.
Basically what it is going to do is pick out constellations in the sky and track its X,Y,Z coordinates within 2mm error of where it is on Mars. It also has the ability to track its Yaw, Pitch and Roll coordinates as well. The most brilliant thing about this camera is its ability to track it's self with approx 5mm of positional error without any outside stimuli after it has picked out a constellation and calibrated its self.
So this was the yard we were testing on...
It's not exactly big... but it pretty much matches the type of terrain the rover will be experiencing.
As you may notice... there is a small prototype of a solar powered rover in the shot... We didn't get to play with the big rover cause it was being used by some people who are going for their PhD in astronautical engineering.
Here are some close ups of the prototype that the first year students get to work on. There have been two made so far...
2011 model
2012 model
Another interesting thing that was at the exhibit was actual celestial rocks. They are extremely expensive and we were instructed to not touch them.... I'm talking 50k+ per rock.
These two rocks...
And this one...(sorry for shaky pic)
So anyway... the experiment called for 250 data points along the terrain. We needed a known position to measure from, so we can cross reference our positions and find the cameras accuracy.
So we used one of these bad boys...
Built by Nikon, it's a X,Y,Z position calculator. It works off the response time between a shot of IR (from a laser) and the time it takes to come back. It uses this pretty cool reflector... Which I forgot to take a shot of, but is approx 2mm in dia. Pretty cool technology.
So this is my girlfriend setting up one of the platforms for the camera to sit on. You can see our data point positions marked by a stick in the ground with a piece of paper, stating all data collected from the camera and the Nikon XYZ camera.
As she is setting up the cameras position... I am remotely checking its X,Y,Z positions on a laptop and comparing it to the Nikon cameras calculations...
We were simulating the constellation with positional fiducials that were on the roof of the exhibit...
All in all we measured an averaged error of 2.6mm in the camera. We also concluded that its displacement calculations without outside stimuli was an averaged 4.4mm error.
We also found that in one section of the exhibit there was errors with the fiducial markings giving us errors of up to 4 meters. We pointed that out to the exhibit director and within one day they had fixed it and emailed us a written thank you...
It took a painstaking 9 hours to do a complete study of the camera in this terrain but it yielded some very nice results. Hopefully we get to play with the real mars rover soon.
Hope you enjoyed this.
-Adrian
Last edited:
