Reporting from the cabin 1

We had another show and tell update meeting to report progress and lots came out of this, so much so that we have two blog posts to report it all, otherwise it gets very long and boring! 

First off is the new vision system. We had originally been working with a borrowed StereoPi system but our chassis master wanted better performance and started work with OpenCV on the new Raspbery Pi Zero 2, the technical details recounted in our last blog post. Here are a couple of pictures of the new camera solution.

Two Pi Zero 2's with two cameras, they've been setup at 65mm separation to facilitate usage with VR if we want to go down that route at sometime. Next up is time synchronisation and distance measurement based on the stereo image.

On a more prosaic subject, we looked at some more cardboard models of trees and sheep.

It makes for a very useful exercise to actually build the challenge props and look at them instead of theorising online, and so we had some cardboard sheep made up which we pushed around, knocked over picked up etc to see how they could be moved around. Similarly, we looked at the 'wolf' pieces and what they would be like to operate around, how easy to move or pick up.

This led to the decision to use balsawood to construct accurate 'sheep' to which we will add separate decoration in the future.

It is a simple thing but just getting it right will make the challenge easier to design for and prototype.

The 'apples' on the tree had already been designed, they're polystyrene balls, but a bit of thought for attachment (and detachment) was needed. We're attaching via magnets, which while apparently straightforward threw up a set of issues. First off, they have to be strong enough to hold the apple easily, this also helps with setup between runs so that attaching apples is easy.

Secondly, they have to be detachable, obvious but the more attachable it is, the less detachable it becomes!!! Silly observation perhaps, but there is definitely a balance to be found, we did find with some quite small magnets that the cardboard tree was easily pulled over, and when the apple came free, sprung back dislodging other apples. A cardboard tree isn't necessarily the best subject so a thin MDF one will be made for the next tests.

The attachment fitting for the chassis had been decided upon at the last meeting, and a trial plate made to test it. Unfortunately, the measurements for this were off and a second had to be made.

The main chassis fixings are on a 80mm grid for M4 bolts, with the outside dimensions 100mm. Onto this are moulded four lugs to fit the attachment securely, together with four raised supports for a PCB (dimensioned for a .254mm pcb matrix board) to be fitted.

This is the mounting fitted to the chassis. Not the most exciting item but essential.

That's enough for this entry, see the second blog entry for what else we got up to.