Skip to content

2020 03 12 Team Meeting

Jump to bottom
Tarik Viehmann edited this page Apr 6, 2021 · 1 revision

Venue: AGVR, Auf der Hüls 197

Date: Saturday, 12.03.2020 10:00

Participants:

  • Cindy
  • Daniel S.
  • David
  • Mostafa
  • Sebastian
  • Tarik
  • Till
  • Victor
  • (Nick)

Definition HotPhase: meet 1.5 times ever week, give notice, if you cannot attend Definition very HotPhase: adjust schedules to visit more frequently Definition extremely HotPhase: requires a reason not to come

Agenda:

  • Intro
  • Update on GO cancellation and adjusting the Schedule
    • on hosting a competition in Aachen:
      • official vs inofficial: try to make it official and don't get involved too much
      • regardless of outcome, we prepare as usual
      • at very least make a 1 week hackathon
      • try to get workpiece scanners from festo
      • get more access points (the new ones, also from festo)
      • check if any other team is interested
  • Schedule until Bordeaux
    • GO:
      • Start of VERY HOT PHASE: 06.04.2020
      • Last Testday: 20.04.2020
      • Pack: 21.04.2020
      • Departure: 22.04.2020 9:30
      • Return + Arrival: 27.04.2020
    • Bordeaux:
      • Start of EXTREMELY HOT PHASE: 01.06.2020
      • Last Testday: 19.06.2020
      • Pack: 20.06.2020
      • Departure: 21.06.2020 ca. 20:00
        • maybe start earlier
        • find something on the way to rest?
      • Return: 29.06.2020
      • Arrival: 30.06.2020
        • tight schedule, Till has to be in Munich by the first of july
        • if it is pushed back, Till might travel alone
    • Discuss drivers
      • 2 Drivers per car for GO - nope
      • at least 3 Drivers per car for Bordeaux
        • check who is allowed to drive (also students?)
      • definitely come by car, train and plane are no option
    • Roadmap until GO
      • Migration Robotino 4
        • upgrade to Fedora31
        • switching to Robotino 4 should be as easy as mounting the towers
      • Yolo Workpiece Detection
        • Sebastian wants to finish it
        • currently data preparation
        • no interface yet (probably most work, Yolo has c++ interface)
        • apply to replace workpiece sensor, afterwards maybe extend to shelf detection, slide and conveyor belt (in that order)
        • Daniel will join on this
      • go through github issues
      • working SS in Bordeaux?
        • Sebastian has a HiWi working on it
      • make agent more robust, get rid of hacks
        • rework goal lifecycle
      • agent optimization
        • handle early finished projects (put in SS or DS)
          • also check the case of one robot
          • maybe multiple workpieces on the same belt
        • probably no promises
        • tune order selection
        • fill rs from base oder cap station???
      • Hardware
        • get stuff professionally printed
          • starting with our disk
        • endstop
        • cable management
        • gripper construction: same same, but different
        • check if there are spare lasers
      • Inventory list - when?
      • improve vision documentation
      • change laptop setup?
        • NUCs would draw additional power
        • maybe run everything on base (+ wifi antenna), also potential power issues
      • spare money to spend: 1500€
        • spend it next month (no hard deadline)
        • lots of tools (multimeter, cable cutter, ...)
        • USB sticks for NNs
        • better storage boxes for gripper
        • ODrive
    • first Roadmap until Bordeaux
      • see above
    • Misc:
      • short recap of meeting with LARG (UT)
        • no names for the products that are built
          • resembles a hamburger
          • find some storyline
        • soccer leagues have 2050 goal, what is our goal?
        • TensorFlow light could be promising
      • presentation aftermovie 2019
      • bbq/pizza + booze

GitHub issue recap (old to new): Fawkes-Robotino:

  • branch cleanup - keep up the good work
  • enter field - wait if other robots are in the insertion zone on agent level (Cindy)
  • CI tests: for agent only, for now
  • Conveyor pose, HV failed - low priority, but Victor will have a look
  • prefer back-laser over front - probably not worth, close it with comment
  • remember tag for conveyor pose TF - maybe just to boost initial estimate
  • wrong function min() - nitpicky - Victor takes care
  • revisit action selection - Mostafa
  • docker causes problemns on shutdown - Till
  • revisit exec monitoring - Till Mostafa Tarik
  • record images at start of gripping - Daniel (taking over Daniels work)
  • revisit half finished products facing product tracking - discuss later
  • rework arduino plugin - hard task, not urgent but we could gain flexibility from having a good design
  • adapt arduino firmware to new microcontroller - probably already fixed
  • TensorFlow plugin - might become obsolete with Yolo (Sebastian, Daniel)
  • stop filling ring stations during late game - closed, will be handled via promises and requests
  • improve manipulation speed (icp iterations) - probably fixed, Victor takes a last look
  • Calibration distance of gripper movements - David is working on it
  • enhance timeout on DOWNed MPS - rework timeouts and DOWN mps handling (Tarik)
  • monitoring failed gotos - no agent concern, but revisit discarding policies
  • monitor points on slide put - revisit if problem appears
  • reduce gripper movements - already merged
  • reduce mps turning when searching tags - Sebastian
  • laser line lag when drive_to_machine_point - Sebastian
  • remove hardcoded shelf offset - Sebastian via Yolo
  • early products block CS - Cindy
  • naming bots - HUGE problem
  • waiting position naming conventions - Tarik
  • waiting positions on same point - no issue, should be handled by tolerances
  • gripper timeout in simulation - core issue Hardware:
  • wifi setup - closed
  • ODrive - future work (ask Nick, maybe get one to test it)
  • barcode scanner for workpiece detection - just use RealSense picture

Home

Guides

Other Stuff

Clone this wiki locally