Not exactly a morning-news sort of read, but it’s fun to look through.
Semi-officially, 42 teams submitted at least one solution to the tasks. I guess I did better than I thought.
Team Rhubarb did not make the cut. As I said before, I had fun, learned some things, and I hope I helped others to learn a thing or two as well. Congrats and good luck to the 20 teams who moved on!
While we wait for the results of the qualification round of the Space Robotics Challenge, lets take a look at a few other competitions:
- The Google Lunar XPrize is moving along with 5 teams scheduled to launch before the end of 2017
- The Agile Robotics for Industrial Automation Competition (ARIAC) is starting up
- … and then there is Hebocon: The Best Worst Robotics Competition
Team Rhubarb ranked at least 92nd place out of 405 teams according to this (official) site.
I like the map and seeing how many other countries signed up, although I wish you could zoom in a little.
When teams asked for clarification, the “92 teams competed” translates to “92 teams were eligible to submit files during the Qualification Round, not the number of teams that competed.” I.e. it is possible not all teams submitted log files. Which means, at the very least, of the 92 teams that were eligible to submit files, I did better than the teams that did not submit log files (duh) and likely ranked better than teams that submitted log files for only one of the two tasks. If I were to make a conservative guess, I’d say I ranked between 75th and 50th. I suspect we’ll get some more numbers once the 20 finalists are announced on (or around) January 31st.
I updated the main project page, including a new write up of the competition so far, videos, my (unofficial) scores, and links to my code. Click here to check it out!
With exactly 5 hours to spare, I was able to complete Task 2. I had a little bit of troubles completing it using C++, so I switched over to Python and things ran smoothly. I’m currently trying to fix up a few things that should be somewhat easy to fix and will result in a better score. I’ll make another post later with pictures, videos, and an after action report.
Over the past weekend I was able to peel away from my day job and devote some time to the competition. Qualification task 1 is now complete. My score isn’t great, but it gets the job done. I’m hoping task 2 will be as easy and that I will have enough time to work on it. So far I’ve spend roughly 3 days (calendar time, not cumulative time) on the technical solution to the challenge, so I’m happy with the results I’m seeing! Will I be able to finish task 2 before Saturday at 5pm? Stay tuned to find out…
Finally! Today was the first time I got to write code towards solving the tasks put forth by NASA. I concentrated on the first of two tasks, which is a computer vision task. Lights on a console turn on and your program needs to determine what color the lights are and where they are located relative to the robot’s head. The work I completed today detected the colors as they lit up as well as where they are located within the incoming video feed from the robot’s camera. The next steps are to determine the button’s location in 3D space as well as refine the accuracy of the button locations as that will result in a higher score. Once these tasks are complete, I simply need to publish ROS messages with the appropriate information. A logging mechanism subscribes to these messages and stores them in a file. This file is submitted to NASA as my final score for Task 1 of the Qualification round.