Payload Test results, Competition Demo announcement

Announcements
1

This post will combine a couple of posts we planned to make, so please read it carefully as it may seem to be all over the place otherwise.

First of all - Payload Test results can be found here. Despite only few of you managed to achieve complex tasks, we’ve decided to be rather generous with the points - especially because many of you have prepared really solid presentations of reasonably complex systems. Couple of remarks after the Test:

  • H.264 camera streams (/ffmpeg topics) should be decompressed on the target machine where data is ingested/analysed. Raw FullHD images are bound to saturate the network link otherwise.
  • ROS bags used with Nav2 might need to have some topics filtered out to keep up as the amount of data published is very large
  • When transporting large streams use best effort QoS. This setting can be used to work around network getting overwhelmed by the retransmitions.

Next - as the set of Teams does not change between the Tests this time - here you’ll find an announcement for the next Test. Couple of notes here too:

  • Test will be held in an outdoor environment that will not be the Marsyard yet. There will be Markers, there will be objects to be detected, there will be obstacles to test the safety systems. Ground will be rather flat (think grass, gravel and pavement) which should make the first autonomous runs easier.
  • Keep in mind that we’re still inviting 10 Teams for this Test so time will still be limited to 30 minutes. This is why we both do not expect you to be able to finish the whole challenge perfectly and will try to somewhat limit the browsable area to something smaller than the Marsyard used for Finals
  • Screen share will be mandatory for this Test. It’s much easier to score this way, and if you do so we’re able to troubleshoot any issues much faster - which helps with any delays significantly

Also - we’ve been to Marsyard a couple of times so far to do some testing on our own and we’ll be introducing some small tune-ups here and there, namely:

  • Some Markers will be placed directly on the ground - it makes sense to place them on the outer edge, and most edges have hills on them - which made the original placement unnecessarily high - to the point that they were hidden from cameras, unless robot was unsafely tilted
  • We may end up changing the angle of the cameras to be pointing slightly higher. Current placement makes finding items easier, but makes the marker detection barely usable.
  • Even though we do not know the exact shape of the Marsyard for the competition this year - we already know that it’s shape will be most certainly rather complicated, this is why we want to stress right now: do not try to traverse the whole available area - realistically some spots left on the field will not be traversable autonomously - we will be including them mostly to test the safety systems in various ways, but we will not be placing any items to be found in spots only visible from inaccessible areas (such items may be i.e. visible from entrances to such places, as distant items).

Finally - during our trips to Marsyard we’ve recorded a ROS bag for you. It should contain all the data you need to prepare for the event next week. You find it here together with an “overhead” video recording of the path taken - so you can more quickly reference the raw data and something “human readable”/high level. Please have in mind that this most likely will not be the exact Marsyard setup for neither the Finals of Remote nor Onsite Competition - so please refrain from over-optimizing for those particular conditions/routes/etc.

We hope that this message answers more questions than it creates :wink:
Paweł on behalf of Husarion team

4 Likes