Categories

SwARM, week 18, 1/2

Hi,

D-2.

Since Arnaud’s implementation of the new algorithm works, we decided not to take the chance to use my better optimized version: as we say in French, “better is the enemy of good”.

We now have moving robots, sort of, and are integrating the imu. Hopefully, we’ll have integrated the decawave too, for now the precision is horrendous.

The presentation is getting really good, only minor changes were made recently.

We are really close to getting something good looking, but as of now it is highly disappointing, especially since now bugs seem to spawn hourly.

Ad unum

Alban

SwARM, week 17, 2/2

Hi,

Sorrow shall mark these days. We simply could not integrate Paul’s and my work, far too many bugs spawned everywhere. Despair overcomes us.

I am currently re-writing my pathing algorithm to clean the ugly mess it had become, and in the hope that this new elegant code will suck less: I stick to KISS.

Alban

 

SwARM, week 17, 1/2

Hi,

Busy times lately, a big end-of-the-semester final tomorrow, my HR interview on Monday morning, and quite a lot yet to do.

I adapted and debugged the path algorithm, both in C and coffee script: now, only one edge case remains in which the GUI fails to draw the path, otherwise it is all set in stone, or at least I dare hope so. We built and welded several motors this week-end to anticipate the short delay to make the robots, but we still need to do quite a lot of 3D-printing, welding and assembling to have fully fledged robots. Hopefully it all seems to be converging towards a rather satisfyingly functioning project, even though the process is quite messy.

Our pace has become quite harsh, but our recent successes help us endure this, although quite frankly the stress is building up. Time is of the essence!

Alban

SwARM, week 16, 1/2

Hi,

So I implemented the algorithm that gives a trajectory with an arrival date and orientation, it took quite some time since I many an incredible amount of silly mistakes and we changed the architecture of code quite a bit, which required a few rewrites. I also made a test file to check it, but Arnaud’s GUI will prove to be the best tool for the job.

I now need to get back to the camera.

Alban

 

SwARM, week 15, 2/2

Hi,

I finally finished implementing the first version of the algorithm that gives the trajectory to be followed as well as the proper orders to follow it. It does not yet handles all cases, but soon will. I deemed more important to have a working version that can be tested as soon as possible.

Alban

 

SwARM, week 15, 1/2

Hi,

we now have the music we will use for our choreography, at last. This will allow us to be sure that we do exactly what will be used in the choreography, no more, no less.

I have been researching IR cameras, but right now and until Friday, I am working on the algorithm to be used to find the trajectory of the robot once we have an order: before, we did not set the arriving orientation, changing this means changing our algorithm.

The batteries we intended to use are not available in Europe any more, so we will use some batteries that were not being used plus the ones of the BreathUp team, with the addition of a little more than a dozen of another type, we thus will have to modify the connector. All in all, this is an issue that was promptly and neatly resolved. Scared me to be honest.

Alban

SwARM, week 14, 2/2

Hi,

So the LED detection issue was just a matter of finding the proper parameters to tweak, so good news, nothing is to be discarded. Yet.

On Friday, we had to give a presentation on the advancement of our project, and the least I can say is that it is all happening scaringly fast. We’ll need a fully fledged prototype before we begin production, and hopefully this will be soon.

Currently struggling with which camera to buy, a problem that actually did slip our minds.

Alban

SwARM, week14, 1/2

Hi,

This week I had an exam I had to work for, seems like a neverending cycle.

I worked with a LED that Perceval had soldered on an E407 to try and reliably detect it with a webcam. Long story short, it is rubbish. I am currently trying to determine how much of it is due to the poor conditions (visible spectrum with much noise around) and improve it as much as I can, but it is by no mean satisfying at the moment.

We also discussed the server’s architecture and how we intend to merge the different ways of measuring the position, that is yet quite an ongoing process.

Also still thinking on the choreography.

Alban

SwARM, week 13, 2/2

Hi,

After much debate, we have decided to use the camera as a precision enhancer more than a fully redundant system to the decawave. As such, it becomes far easier to implement our ideas. As of now, the plan is to have the decawave identifying each robot and providing a more or less precise position, then the camera will pinpoint a robot which will be associated to the closest one detected by the decawave. This system is heavily reliant on the decawaves’ ability to locate a robot with such a precision that it will not be confused with its neighbouring one. Tests will define how close we can put our robots to each others.

I also double-checked Arnaud’s maths and reached the same conclusions by different means (which helped notice a little mistake in what results had been given to Paul).

Perceval and I are currently working on the choreography to be sure that we implement exactly what we need to realise it, and neither too much nor too little.

Alban

SwARM, week 13, 1/2

Hi,

First of all I wish you a happy new year.

I had lovely holidays, went completely off the grid and had the opportunity to rest; I believe this replenished energy will be fully needed in this final stage.

I had to re-undo some work in order to go back to having opencv as a git submodule, and this raised some yet unresolved issue, surely they will prove easy enough to overcome. I worked a bit with a webcam, and there is not much difference from working with a recorded video.

Tomorrow, I will document myself on the shutter issue, and will decide with the group what to do depending on my findings.

Alban