Mounting an OAK-D, or OAK-D Lite on a servo

gregflurry

Has anyone tried to do so?

I am designing a wheeled mobile robot where, for navigation, I hope to recognize objects close to the ground. The objects could be 50 cm to 1000 cm away (cross fingers, I've not even trained a model yet). To do this I could easily do a fixed mount for the OAK; in fact I've done so already using off-the-shelf parts. I'd also like to be able to recognize some faces. Based on some work I've done, with a fixed OAK, the camera would have to be > 250 cm away from a sitting person's face for a face to be in the VFOV. That may or may not be OK; I've not done all the work yet to know.

In any case, the goals got me to thinking about mounting the device on a servo. Sounds like a job for a good mechanical designer and a 3D printer. I am not, and I don't own a 3D printer. So, I'm looking for some input. It seems an interesting challenge with the 1/4-20 screw and potentially stiff (mine are) signal and power cables.

erik

Hello gregflurry , asking the team if anyone has heard about such mount. We might also be able to prototype one (can't promise anything though).
Thanks, Erik

gregflurry

erik Thanks. I should probably add that while I think I'd be satisfied with just a tilt capability, a more universal approach would be both pan and tilt. With what little thought I've devoted to the subject, it might be easier to create the latter.

Also, I notice that the OAK-D Lite has the mount hole on the bottom rather than the back. This might make a design easier, and certainly the Lite would weight less and require a servo with less torque. I'm looking forward to receiving mine!

gregflurry

erik I was going back through some notes I made a couple of months ago and I found this ( entry in an OAK contest. It obviously has an OAK-D mounted on what appears to be a reasonably simple pan/tilt mechanism. I could not extract enough detail to understand what the creators did. I have not yet figured out how to communicate with the creators, but will keep trying. You might have more luck.

gregflurry

erik This entry ( from the same team has better shots of the pan/tilt mechanism, but still not enough detail to understand it.

gregflurry

erik And once again, I wrote before exhausting all search terms. I found this, and this. I think that mechanism is what is used in the project I cited in earlier posts.

Now that I can see the design, I have some reservations about long term viability. It uses micro-servos which are not known for their torque. I have not done the math but 115 g is not trial to move around, and as I said in my original post, the signal cable I acquired is very stiff and would add to the torque needed to move the device.

So, while there is an answer to my question, I'm not sure it is a good answer, and almost certainly isn't the best answer.

erik

Hello gregflurry , thanks for links, we weren't even aware someone has opensourced these designs! We will probably modify these for other models that have tripod mount on the bottom (eg OAk-D-Lite), not the back. From the video it looks like it's enough torque for the OAk-D, but might depend on the cable as you mentioned.
Thanks, Erik