Tableau in FIRST Robotics
During high school, I participated in the FIRST Robotics Competition, an international high school robotics competition. Teams have 6 weeks to design and build a robot to compete in the year's challenge. This year, our team was provided with 5 copies of Tableau software, which was an incredible visualization tool to strategize with our alliance partners.
In 2015, the game was FIRST Stronghold, where alliances of three robots take on each other by breaching the opponent's defenses, firing boulders at the enemy tower, and finally scaling the tower wall. Take a look at the game animation for a brief overview of the rules.
The competition is divided into qualification and elimination rounds. During the qualification matches, teams are randomly paired with alliance partners and opponents. The top 8 performing teams are then chosen to be alliance captains, and can each select 2 other teams to accompany them to the elimination round. The challenge is to have a dynamic strategy optimized for the random qualification rounds, as well as to demonstrate desirable abilities to be chosen in the top 24. The full seminar presented at the 2016 FIRST World Championships can be found here.
It's important to look at the opposition before the competition. By gathering publicly available data from every team and converting their home city to latitude and longitude, each team can be mapped geographically. We use Offensive Power Rating (OPR) as a quick metric for how many points a robot is expected to score, but it's not effective for complex strategy. This map can answer questions such as which regions have the best teams, where is my opposition, and for the world championships, how do international teams compare?
Now that we're at the competition, let's take a look at our first match. It seems that it's in our favor, with the OPR expecting us to win with 20 more points. This chart serves as a quick glance at the outcome before more sophisticated analysis.
A robot's offensive capabilities can be grouped into two categories: breaching defenses and scoring goals. This plot compares the offensive capabilities of the robots in the next match relative to other robots in the competition. It indicates which robots should focus on scoring, breaching, or defending our tower, as well as which robot to best defend against. It looks like the opponent has one sharpshooter, 5114 - a good candidate to defend against.
The box and whisker plot is excellent for comparing consistency. The shooting score in every match is spread out in the plot. A wider box (high variance) means less consistent shooting, and vice versa. Robots who score more low goals than high are marked with orange. The plot could be used to gauge the amount of balls expected to be scored. It requires around 8 high goals, 40 points, to take down the tower, so adding up our team's median goals gives us 55 points, so capturing the enemy tower is a possibility.
Box & Whisker Plot
Now looking at the robots' breaching capabilities, a bar chart can be used to make sense of FIRST Stronghold's breaching system. It is necessary to take a look at the capabilities of every robot in order to identify possible weaknesses on our alliance and our opponent's. 6 of these plots for every alliance member and opponent can be used to map out exactly which defenses our alliance will breach during the match, as well as which defenses to put against our opponent. Here, team 2834 has demonstrated capabilities in breaching all of the defenses except the drawbridge and sally port.
For a comprehensive overview of each robot's capabilities, a stacked bar plot is best. The autonomous, breaching, shooting, and scaling categories are color coded. Unlike the earlier stacked bar chart, this chart uses data from the current competition, so this chart can be uses to double check the performance of the OPR metric. In this particular match, as the Blue Alliance, we would decide that 2834 and 3655 be the goal-scoring robots and 6221 defend our tower against 5114.
Stacked Bar Chart
There is more to strategy beyond charts and plots. After crafting a master strategy, it's useless if no member of your alliance executes it. Persuasive visuals and effective negotiation help keep your alliance partners confident in your strategies and make them more likely to stick to a plan, rather than shoot randomly. If you're interested, take a look at an example competition with 2834 and its alliance partners 33, 1756, and 329 at the 2016 FIRST World Championships!