NASA’s Perseverance Mars rover
Discussion
Smiljan said:
SpudLink said:
There’s an interview here with the senior software engineer on the project:
https://spectrum.ieee.org/automaton/aerospace/robo...
“We use a cellphone-grade IMU, a laser altimeter (from SparkFun), and a downward-pointing VGA camera for monocular feature tracking. A few dozen features are compared frame to frame to track relative position to figure out direction and speed, which is how the helicopter navigates. It’s all done by estimates of position, as opposed to memorizing features or creating a map.”
“ We also have an inclinometer that we use to establish the tilt of the ground just during takeoff”
Brilliant, thanks for that info https://spectrum.ieee.org/automaton/aerospace/robo...
“We use a cellphone-grade IMU, a laser altimeter (from SparkFun), and a downward-pointing VGA camera for monocular feature tracking. A few dozen features are compared frame to frame to track relative position to figure out direction and speed, which is how the helicopter navigates. It’s all done by estimates of position, as opposed to memorizing features or creating a map.”
“ We also have an inclinometer that we use to establish the tilt of the ground just during takeoff”
they could always use gps ;-) ;-)
The Perseverance rover captured by da choppa as it zoomed past on its 3rd flight.
https://spaceflightnow.com/2021/04/25/nasas-ingenu...
https://spaceflightnow.com/2021/04/25/nasas-ingenu...
bmwmike said:
Absolutely mind blowing tech.
Be awful if the chopper crashed and took out the rover, how bad a day would that be for the pilot. First accident claim on another planet? Or I guess they are on a joint policy anyway. Crazy stuff.
Does it have a pilot? Given the time that signals take to get there I assumed it has various flight programs that are loaded up and then it goes itself. Be awful if the chopper crashed and took out the rover, how bad a day would that be for the pilot. First accident claim on another planet? Or I guess they are on a joint policy anyway. Crazy stuff.
Autonomous.
It's impossible to directly control a rover or helicopter from earth due to the time delay involved.
There are two reasons for the time delay - one is the distance and the limitations of the speed of light. The other is the fact that the rover/chopper receives and sends its signals to and from earth via Mars orbiting relay satellites. These have to be in the right location in their orbits to be able to make the link with earth.
It's impossible to directly control a rover or helicopter from earth due to the time delay involved.
There are two reasons for the time delay - one is the distance and the limitations of the speed of light. The other is the fact that the rover/chopper receives and sends its signals to and from earth via Mars orbiting relay satellites. These have to be in the right location in their orbits to be able to make the link with earth.
Yeah of course, figured there was no "realtime" pilot, was also a slightly tongue in cheek comment, but there must be someone issuing instructions; spin up at X time, elevate to 20m, go left, etc? Sorry to drag the conversation down to technical matters again (ha!) but I wonder how high level the instructions are. I mean, can they say, do a 30 second flight, max alt 50m, max distance from rover 25m, and it chooses the flight plan itself? In which case, some logic to avoid landing or hitting the rover would be needed, i'd think.
Did I miss something? Since the failed attempt, it had it's 4th flight earlier today (well, yesterday now) 133m there, 133m back, 5m altitude, 118s flight duration.
https://www.youtube.com/watch?v=XbnS967lJqY
https://www.youtube.com/watch?v=XbnS967lJqY
Leithen said:
That does read as if they intend to try using it for scouting ahead.xeny said:
Leithen said:
That does read as if they intend to try using it for scouting ahead.Article about the anomaly during the sixth flight
https://mars.nasa.gov/technology/helicopter/status...
https://mars.nasa.gov/technology/helicopter/status...
MartG said:
Article about the anomaly during the sixth flight
https://mars.nasa.gov/technology/helicopter/status...
Navigation images taken 30 times a second, and a single lost image causes 20 degree oscillations. Impressive that it coped with this to complete its flight. And a useful experience to help improve the software. https://mars.nasa.gov/technology/helicopter/status...
It shows the value of extended testing in the real world! This is the sort of thing that would have been difficult to catch in a relatively small vacuum chamber. Very odd though. I imagine one lost image wouldn't normally be a massive problem, the big problem was that the timestamps seem to have become corrupted and the analysis software was reading the images in the wrong order. Attempting to correct this perceived motion just made it worse. I guess if they have to then they can default to inertial navigation, which seems to be working well.
I imagine some software engineers are probably having an interesting weekend analysing just what happened and how it can be patched around.
I imagine some software engineers are probably having an interesting weekend analysing just what happened and how it can be patched around.
Zad said:
It shows the value of extended testing in the real world! This is the sort of thing that would have been difficult to catch in a relatively small vacuum chamber. Very odd though. I imagine one lost image wouldn't normally be a massive problem, the big problem was that the timestamps seem to have become corrupted and the analysis software was reading the images in the wrong order. Attempting to correct this perceived motion just made it worse. I guess if they have to then they can default to inertial navigation, which seems to be working well.
I imagine some software engineers are probably having an interesting weekend analysing just what happened and how it can be patched around.
My interpretation, they're using a counter rather than time stamps, system got out of phase due to error reaction behaviour.I imagine some software engineers are probably having an interesting weekend analysing just what happened and how it can be patched around.
Counters make the code simpler and faster, but as a result the downstream system would not know that the upper system was corrupted, especially if things like end to end protection are also affected.
The fix should be relatively simple if this is the case, increment the counter prior to the picture validation.
If they are using timestamps, the fix would be similar if slightly more complicated, ensuring the phase alignment of picture to IMU timestamp.
Gassing Station | Science! | Top of Page | What's New | My Stuff