November 19, 2021
Sprint 1 is over, and Sprint 2 has begun!
Sprint 1 was all about laying the foundation. Most of us started with zero experience in ROS, SVL, or other tools at the heart of Nova. Some of us had not used Git or Linux before. But we learned a lot and ultimately finished 67% of the Sprint's story points... as virtual beginners to this whole thing. We:
- Wrote a bare-bones communication framework between the web browser and our ROS code
- Read through lots of Controls papers and arrived at a roadmap for lateral and longitudinal control going forward
- Wrote a bare-bones safety manager, which will ultimately help our system overcome critical failures
- Formed a Finite State Machine diagram for the behavior planner
- Configured our simulation server to create restore points automatically (along with some other neat things)
- Added and interfaced with a state-of-the-art GPS system
- Recorded 174 GB of campus drives
- Formed a nice draft 3D model of Hail Bopp
- Wrote three new "deep dive" articles for our website
- Began refinements to our localization algorithm
- Detected people for the first time using a 3D vision-based AI
- Converted our map data into cubic splines, which our path planner will use
- Lots more!
This is where we address our first checkpoint: "A", where Navigator must localize itself along the entire 2.01 mile course. A tall order! But an achievable one.
While the Perception team works on Checkpoint A, everyone else will be busy working on future checkpoints. For example, BPC should have path planning totally finished by the end of January-- what! And Quinn, Phu, Connor, Avery, Dylan, and Raghav will also have their hands full.
I've added Issues for many of you in ZenHub already. Let me know if you have any questions about them.
Dylan and Quinn, please reach out to me so we can touch base on your Issues. BPC, please meet with Josh to pin down your Issues for this Sprint.
We have a lot to share!
Everyone worked hard this past week, and so this was tough.
The rubber chicken this week goes to Egan, who worked through lots of challenging and messy Lanelet code over Sprint 1. He spent many hours in both the upstairs and downstairs lab, rushed in to record sensor data, and came to nearly all the open labs (or was it every one?).
Craig Robinson, the team lead for Waymo's inertial navigation group, is coming to campus on the Friday we return (Dec. 3). In case you live under a rock, Waymo, formerly the Google self-driving car project, is probably the most advanced autonomous car company in the world. Craig will deliver a talk and Q&A to students, and then he's coming to chat with our team afterwards (!!!).
Though the times are still tentative, Craig will likely chat with us from 5 to 5:30 on that Friday. His public talk and Q&A, which I would encourage you to go to, will be from 3-4pm.
You can find a flier for this event here.
Remember how you ranked your choices? This was used to calculate a ranked choice vote. Using these results, Logo "a" won, with logo "d" in a close second. Ranked-choice accounts for simulated runoffs (since you've already provided your second and third choice).
I've shared the results, along with feedback, with Phu. He may tweak logo "a" slightly per the feedback.
For the nerds out there, here is the pairwise matrix of the results:
- Required hours will be cut in half to 2.5 hours next week.
- I'd like you to check Teams once every two days, rather than twice a day.
- There will be no team-wide meeting next week, but BPC and Perception will each have an individual meeting next week, virtually. Times TBD based on availability. These won't take long.
- No open labs next week
I hope this is sounds fair. We can't afford to quit entirely for a week, but this should still allow each of you to work at your own pace and to get some rest.
This is personal, but since Thanksgiving is next week, I just wanted to thank you all for working on Nova with me. I've said this before, but I was in this lab alone for a year (well, I had Justin), and for that whole year I was dreaming about the day when I could get a group together to work on this as a team.
After a year, we had capacity to welcome five members, and we did a whole lot together. I'm grateful to them for getting this project off the ground. That was really tough, especially with the pandemic, but not a single one of you gave up!
And one year later, our team has tripled in size. We now have so many talented people working together that our weekly meetings easily fill an hour, and I'm sure could fill two if we let them.
It's a wonderful thing to be on a team with such motivated people as you, working toward the same future.
Above: The evolution of our team