June 19th Pool Test

By juin 25, 2016Pool Test

Hi guys !

As the competition is coming really soon, we would like to let you guys know what’s going on with our pool test.

Here is a quick wrap up of our last test:

Huge step on the mechanical side: The sonar rack and the camera dome have been tested and are 100% functional!!  The field of view of the camera is really wide and  the objects appear a bit smaller. The long range detection of the obstacle might suffer from this change and a modification of few vision filters will then be necessary. Taking into account the problems that we discovered last week, we worked a lot on the stabilization of the vision server as well as the mapping system.

On the vision side, the very majority of the bugs have been fixed. We were happy to note that despite all the features that have recently been added to the vision platform (complete integration of the dynamic parametrization), we had a really stable platform throughout the test. We had a small crash, therefore we have to be careful with the crash context in order to sort out the steps to reproduce. But we are very confident that the vision platform will be frozen and released soon!

On the mapping side, we wanted to benefit from a Region of Interest logic integrated in the system, as well as a way to detect the walls and discard the data outside of the boundaries. We struggled out on architectural problems, and a huge refactoring of the mapping system has been required to let us continue the development. Regarding this refactoring, we could not test any missions with the mapping, but the problems seem minor and the next test will be dedicated to this purpose.

On the mission side, I will simply quote Jay:

“ The pipe mission worked great with the crab walk, this implies that we can find the bottom target easily. In addition, the crab walk uses a trigger that can be re-used for the buoys and the fence. The front alignment has been concluding, we don’t oscillate around the targets and we don’t surface while trying to align. However, while aligning, we step backward from the targets, we have to fix it. I already know how to do it. I also tested a new method that align and move forward in the same time. It does not work yet, but it’s pretty cool, I have to try it again.

These are all great news, and we can’t wait for the next pool test, let’s wish that it will be as good as this one!!!

Leave a Reply