Notes from Team Leader Meeting + discussion

Hi All!

Thanks for coming to the virtual Team Leader Meeting today!

Unfortunately the meeting recording didn’t go through correctly but fortunately we have the notes - see below.

Dexterity:

  • Do everything you can! You’re on your own time.
  • Each rail/task is a separate recording.
  • Completeness then time.
  • Linear rails, etc. are screwed down.

Search/Inspect

  • Terrain - go up a 90mm/3.5" obstacle (e.g. a wooden rail, curb, parking bumper, etc.) either every 2 sets of targets, or need to go up stairs partway through (inspect at least 3 sets of targets on each side).
  • Spacing, 2.4m between sets of targets. Can go back and forth or outside.
  • Need to get close to the bucket with hole to actually inspect it (need to see whole ring).
  • The bucket with the hole needs to be black as it’s supposed to be dark. The colour of the other buckets doesn’t matter so much.

Mapping

  • Count the halves (coverage), count the whole (accuracy). Count wholes first, use halves as tiebreaker.
  • Area around fiducial needs to be recognisable to at least 2.4m radius (it can’t be an ink blot!)
  • Can’t see both halves from any one point. Can use furniture to enforce.
  • Fiducials at least 2.4m apart
  • Travel at least 4.8m between fiducial halves
  • Can’t see more than 3 fiducial halves from any one point. Can use furniture to enforce.
  • QR codes are a direct replacement for barrels if you can’t get the barrels.
  • I’ll generate some example QR codes. They should be the smallest (largest square) 21x21 codes and fit on a letter/A4 piece of paper (including the 3 unit white border).
  • Annotate the GeoTIFF for thick walls/lockers/etc.
  • Have one upper/lower match up, the rest separate with upper/lower separated by at least 1.2m
  • At least one obstacle (like in Search/Inspect) every 4.8 m of progress through your environment (NOT roughly halfway through your map as originally discussed) to demonstrate that it doesn’t break your map and that you have mapping on something that can handle it.
  • For now, assume the environment is on the same level to keep roughly comparable given teams are potentially doing this in their own labs/houses/etc.

Autonomous Mobility

  • Fixed pallet config (left version of page 76)
  • Add an extra pallet and gate at the start that’s completely flat
  • Count pallets completely traversed over

Schedule:

  • Videos due by the start of team leader meeting at 8am US Eastern Time on the 21st.
  • Attestations due by the start of team leader meeting at 8am US Eastern Time on the 22nd.

Please continue the discussion here!

Cheers!

  • Raymond

Had a discussion with Adam and resulted in a minor update to the rules for mapping, to match it up with the rule for Search. Basically, only a single obstacle is a bit too easy, ideally it should be the same environment.

So we’ll update Search to require an obstacle roughly every 4.8 m of progress through your environment. This corresponds to every 2 tasks in Search (which are spaced roughly 2.4 m apart). We can’t make the exact same rule because you can map a fiducial without getting close to it.

Does this sound reasonable?

Also updated timings for the team leader meeting (we’re not on EST right now).

Do you mean “Mapping/Exploration” here? I think, it’s a good idea to match the ruling for exploration and search.

Hi, if we do the stars do we need to put the k rail for the exploration?

Thank you

Hi Martin, Capra,

Yes I believe we’re going for the same rule between “Search and Inspect” and “Exploration and Mapping”, at least within the bounds of practicality (it’s easier to say an obstacle every 2 search targets because you need proximity, it’s harder to make such a statement for a fiducial that you can see from across the room).

And thanks for picking up the error, yes I should have said “we’ll update Exploration and Mapping” rather than Search … my apologies, been rather sleep deprived taking care of a new future RoboCupper :wink:

Cheers!

  • Raymond