The EAGLE has NOT landed

Eagle/FUSION 360/Autodesk need sort out their authentication system.

I have EAGLE. They opted to use two part authentication. It was set up for a time but is now broken.

So to fix it I need log into help. Can’t because I can’t get in to set up authentication AND I never ever ever receive the codes by email. Sure I can set an an authentication app IFF I first log in.

So I am in this dopey position I cannot talk to help desk unless I log in and cannot log in since their authentication system won’t send me the codes by email.

Scott Adams must be working at Autocad, since that is a very VERY Dilbert way of doing authentication.

Zoom, zoom, zoom!

2.8mm lens

6mm lens

So, as I expected. A lens change, with a narrower FOV, “zoomed” into the image to better fill out the sensor realestate.

Focusing now seems really straight forward as I worked out I can set the focus of the lens and then pop the 360^o lens, rather than try focusing via the 360^o lens.

Next step, play!

Found an example that included a linear polar conversion. Hmmm.

HOLLY HELL!

Who needs to write code?!

As simple as:

img = sensor.snapshot().linpolar(reverse=False)

AND, a little twisting of the 360^o lens (probably to move the centre of the image to the centre of the sensor).

As a check I carefully took off the bloggie leans, and then re-ran the hello-world, and here is!

Not quite centred but close.

So, less work to play with looking at means to do obstacle avoidance. Based upon the roborealm example, here is a first attempt using canny edge detection, with the thresholds from the tutorial page.

Canny edges

Really simple, with:

img = sensor.snapshot().linpolar(reverse=False).find_edges(image.EDGE_CANNY,threshold=(50,80))

Needs more playing, but a start. Obviously, me desk may, I say may, be somewhat too cluttered. Out on the lawn, with an expanse of green, may be a better proposition.

Although, canny edges over the starting image may also be useful. At least to, say, run find_circles and so help programmatically find centre. Can’t use the centre (x,y) at the moment. Waiting on update of lin_polar() to take centre input.

Polar canny edges

Or you can find someone has already done it, and hence.

Each vertical should now be seen as a radial plot in 360 degrees, range from bottom to top, centre top is looking forward (range from bottom)

Slight adjustment

Notice the difference?

Just playing with minor mods to the uStepper 4 model to get around a really irritating design decision by the original author. That is using 15mm diameter aluminium tubing. Since, I am finding multiple places aluminium tubing jumps from 14mm to 16mm! Same pattern in carbon fibre tubing.

So, just fiddling with hacking this part from 15mm to 16mm fitting. Using the Manipulator add-in to great effect to reverse engineer centre point of cut, based upon the radius of the cut. Then using the centre provided to site the new cylinder to cut a new chunk out. The trick is that as the rod is thicker, it also had to be lifted a tad (0.5mm) away from the teeth of the belt drive. So a similar fiddle in the bracket.

I’ll do a test print at some stage, before moving onto some of the other parts. In fact, the bracket is on the printer right now, burning through that horrible green colour I opted into out of an urge to homage.

UPDATE

So, turned out not bad in draft print.

Draft print for 16mm shaft

I need to grab some 16mm shaft to try this. I did find 15mm carbon fibre on Aliexpress but don’t really want to wait. But, with the bank up of project, I guess I could.

While the green filament is loaded, I am knocking up a few 8mm feeders for my PnP as well.

Very Flash!

Just starting to use the new V5 of FlashPrint from FlashForge.

I noticed something sweet. I had put a suggestion in about colourising the axis annotations in the rotation wizard, to help mapping the wizard entries to the rotational axis in the graphical display. It looks like that idea was incorporated. Not earth shattering BUT it makes for a easier time relating the wizard entries to the graphical represntation.

Simple human factors improvement

I am fiddling with some of the other features, though mostly driven by a pesky problem. I had decided I would print the PnP parts in blue. Black and blue then the colour scheme. The T-Slot 2020 are in black.

The problem is I am having all sorts of problems getting the print to stick to the plate. So, I have tried out the trick to use different temps at different layers, slowed the initial layers down, heated them up (first 4 at 225^oC), put pauses in to let the bottoms layers cool a tad, to then do layers 20 upwards at around 210^oC.

Fingers crossed, I am up to forth attempt. I also added a wall, which I did not appear to need for the black draft prints. I am also printing fine and not standard. So whether I need the smooth look may be a question. It may be fine using a standard setting. I was just wanting a little more polish in the finish.

I also noticed that the 19mm thick base idea was a mere 1mm off as the MDF I will likely use is actually 18mm. So, some minor changes and now we’ve the four stanchions printed.

It’s SOOOO HAPPENING!

Late run to the PO, the HEMERA 3D print head from E3D has turned up!

That is, E3D Hemera Direct Kit (1.75mm) – 24V / Kit. KOOL!

The final sprint for my Bear Build is now in site.

It’s HAPPENING!

I picked up a package from the PO today. Turns out its the 2020 T-Slot extrusions for my corexy PnP idea!

AND …

… they were as 1) ordered (phew) and therefore 2) as required.

Exiting since the 8mm linear rail and corner brackets are fast on their trail. They seem to be in country … but not at my PO yet.

The 8mm linear bearings ended up on 50% off sale so I ordered them ahead of schedule.

Go figure though, the z-axis and the pnp head were ordered well and truly earlier AND appear to cleared but still awaiting international departure …. grrrrrrrrrrrrrr.

I am otherwise already at a place where I could start a 3D print run for the corexy bits, sans the final X-Carriage design (mitt the .

However, I still have a couple of builds in parallel going, and so time poor. But it’s going to be sooooo much fun!

ESP32 v SMT32

A late find is that the floating point performance of the ESP32 is a bit sad. The current design of the bloggie lens based idea, for the MHS-FS2 type lenses, was a long lead ov2640, to snake out from the adapter to a TTGO Camera board (see Figure 1). This will still be viable, and there are examples of this setup doing image processing. But whether that can be at a rate for converting the bloggie image to a panoramic, and then lay processing over the top as well, is a question.

Figure 1

I have a couple of OpenMV cameras as well, that that the MHS-PM5 lenses might be adapted to.

So, the ESP32 system may be less capable but dependant on what the processing needs are, it will be still a useful option.

Strategies exist, anyway, to get around likely lower frame rates. Including streaming the data without processing up to a server, which is apt for smaller robots.

To get the drift, the PM5 lens has a more or less clip onto lens coaxially approach (Figure 2 and Figure 3), whereas the FS2 has a clip on at 90 degrees approach (Figure 4 and Figure 5).

Figure 2

Figure 3

Figure 4

Figure 5

The PM5 lens (Figure 2) thus is a much better mate for the OpenMV M7 I have (Figure 6).

Figure 6

To connect the MP5 lens to the OpenMV M7 appears to only need a simple 3D printed adaption (Figure 7 and then Figure 8)

Figure 7

Figure 8

I am finding it works (needs some fiddly focusing), but I will need try a lens with a narrower field of view perhaps (Figure 9).

Figure 9

Bloggie lens hack

So, the point of all the bloggie 360^o lenses is in Figure 1.

Figure 1

Here is a hollow 3D camera hood based upon the MHS-FS2 carcase. It has an ov2640 in the back. The lens from the MHS-FS2 clips over the top. A little software hack converts the image into a flat panoramic and then you can add other image processing over the top.

Maybe not …

… so sad, too bad.

It would seem I have the following going on with the gaggle of bloggies.

Bloggie	Triage result
MHS-PM5#1 (Black)	Removeable Battery dead
MHS-PM5#2 (Black)	Removeable Battery dead
MHS-PM5#3 (Black)	Removeable Battery dead
MHS-PM5#4 (Blue)	Removeable Battery ALIVE! BUT, ON/OFF button dicky.
MHS-FS2#1 (Black)	Internal Battery dead
MHS-FS2#2 (Black)	Internal Battery dead

Test Results

So, one of the black MHS-PM5 is actually the one my wife bought me, so sentimental value? The blue one very swish, but the ON/OFF switch dicky. Replacement batteries for the MHS-PM5 available at AUS$16 but do I? When I bought the additional ones (again to get hold of 360o lenses), they were all still on the market. There are a few people trying to sell a scant view for sometimes up to $150. I guess the warning is check the battery. As I was after lenses, I did check a couple of the cameras, but the age of these really is the problem. The batteries simply burned their shelf life up.

Still trying to find batteries for the MHS-FS2, I am assuming you can open them up with a screw driver to change them? Not sure I am that fussed. I am not throwing them out yet, as I need model them sufficient to allow for clipping the lens onto a 3D printed facsimile of the camera region. I downloaded model of the MHS-PM5 but will need start from scratch I think.

Yep, I did previously opt for blue tack to hold an MHS-PM5 lens on, but obviously not an optimal solution.

I feel I may keep the best of the black MHS-PM5 and buy a new battery, assuming the only live removable battery is likely soon to fail. Ditch the rest in the bin, or give them away. MHS-FS2 definitely for the bin.

Bloggies for sale

So I have a collection of Bloggies, Figure 1, one is not in the shot as I am trying to recharge them before sale.

Figure 1

I say recharge, two failed to charge! Though that is fine, since I can recommend BetterBat for batteries for old warn out tech. I am sorting them out to sell them off, so I’ll have to either buy new batteries OR sell them after testing them. I have one that charged, and I’ll likely keep, BUT I may still opt into buying that a new battery anyway. I can use that battery to test the others that didn’t charge, etc.

Why did I buy so many (all off ebay so not full price, just so you know). I did so because the price I was paying for them was well less than 360^o lenses off the shelf, Figure 2.

Figure 2