Are those some type of calibration targets on the side of coring drill?
You must log in or # to comment.
These are Fiducial Markers (or fiducials for short). Fiducials basically allow the rover’s computers to measure itself.
The markings are a staple of engineering that can serve two parallel purposes, to calibrate cameras and to calibrate various mechanisms on the rover.
They’re a common feature in high-tech photography and robotics here on Earth, but the use of fiducials on extra-terrestrial robots is relatively new.
The technique started informally on the Mars Exploration Rover (MER) mission, which landed Spirit and Opportunity on the Red Planet’s surface in 2004.
JPL engineers noticed that, over time, the accuracy of the MER robotic arm was degrading, and they weren’t really sure why. So some of the JPL engineers came up with an algorithm that would just recognize a circular feature on the end of the arm. Then every day, it could track where that actually was versus where the robotic rovers thought it was. The idea blossomed into more than just an engineering hack with Curiosity. JPL engineers came up with the design for the fiducials for Curiosity, as they needed something that would be easy for either a computer program or a person to accurately pick the center of. Having that intersection in the middle makes it easier for a person, as they can zoom in on the image and click exactly on that intersection, and the design also makes it easy for a computer, because it can compute the center of a circle. From there, basic trigonometry lets engineers piece together the positions and orientations of the various parts of the rover.
To measure all of the mechanisms on the rover, fiducials are installed all over the turret, on the end of the rover’s arm, as well as on the tops of the steering actuators and on the top deck of the rover.
They were so useful they are were installed on the Mars Insight lander and the M2020 rover.
M2020 also uses the smaller ‘April Tags’. They’re fiducial markers for machine vision systems and are “robust to lighting and viewing angle”.
Looks familiar and relevant to this conversation. Thank you all for the great feedback… And those two, bouncing off each other the way they did, a great example of scientific deduction
Thank you all for the great feedback… And those two, bouncing off each other the way they did, a great example of scientific deduction
I only read the other replies after I posted my answer (some good stuff there :) It’s great when everyone chips in, shows we have an active community :) Mine was a stock answer that I posted to Reddit several years ago (not long after Perseverance landed IIRC).
It’s a huge pity the team on M2020 have not followed in the footsteps of MSL for posting regular blogs. I’d love to hear that the team had used the Fiducial Markers at some point in the mission. We may even see them in use soon on MSL if the drill checks show any sort of damage to the drill, as the original method of changing a drill bit probably wont work as well as designed / planned. We can see the bit boxes were designed to be used with the drill stabilizers for alignment. Those stabilizers will hardly come into play now (since the feed motor brake failed / locked), so they are going to have to align the arm very carefully to capture a new DBA. The fiducials could really help…
These are aruco indices. Each represent a unique index. They are popularly used in robotics for all sorts of things. I’d imagine they serve some kind of recalibration purposes for this camera.
Since they know the exact position and rotation of the camera in relation to these markers, they can check if the markers move by small amounts, which would mean lens distortion.
I could be wrong though, since you would typically need quite a lot of them at different depths for a full camera calibration.
Actually maybe they’re used to measure vibrations in the drill? The vibration data would be a lot smaller to download than a full video file, so they might process that info on site and send it back
They might also be AprilTags which look almost identical to Aruco markers but in either case they are so called fiducial markers.
They are typically used for pose estimation, which is the process of determining the relative position and rotation of the marker compared to one or more cameras.
The drill head is on a relatively long arm which can cause the kinematics (determining the head position by adding up angle sensors in the joints) to become a little imprecise.
Attaching markers to the head allows you to determine its position relative to the main body which you can use to verify/correct/supplement the position given by the kinematics.
In the end you know your drill head position with higher precision and certainty, which I would imagine is important when you want to take drill samples or shoot lasers at a tiny target.
While they could be used for camera calibration,
camera intrinsics usually don’t change andyou’d typically have a calibration target with a much denser pattern on it. With only 3 measurement points you’d need to take hundreds of different pictures to get enough data for a good calibration.While I believe you are correct and I don’t think they are used for camera calibration, camera intrinsics are known to vary depending on external conditions. Cameras on autonomous ships need regular recalibration for instance.
Yeah I guess a little expansion due to temperature change would wreak havoc on an optical system.
I don’t really know how they’d recalibrate the navigation cams in the main body then. The mast and arm cams can look at calibration targets on the rovers back.
I have updated my comment to remove that statement.
It’s for martians, so they know in which bin to recycle
Wrong answers only, please.
I think so. I know at the robotics lab in my uni they had lots of those mini qr codes on the walls.
Iirc there was something about a self steering trailer that looked at a grid of four of those codes on the leading vehicle.
