Jankytron

Primary Project Goal:

LEARN! :)

Background and The Game

Build a Big Autonomous Recylcling-Garbage Extractor (BARGE) that demonstrates the ability to collect Gunk And Randomly Bobbing And Getting Everywhere (GARBAGE) for proper waste separation and disposal.

The goal was to build a bot that could navigate the field shown on the left, collect balls that are randomly scattered on the field and deposit them into a landfill (C and D in the picture) or recycling center (A and B in the picture) in competition with another bot that is on the field at the same time. Each bot has its own landfill and recycling center. Whether a ball is landfill garbage or can be recycled is decided by the color of the ball. Additionally the balls are blown in a clockwise vortex on the field boundaries.

Each of the landfills and recycling centers emit IR at a characteristic frequency in all directions. The landfill is open at all times, both at floor level and at the top (so balls could be shot in), but the recycling center can only be opened at the floor level, when an appropriate IR signal is flashed at its detector.

More details can be found in the project description handout and the protocol for the communication with the master game controller.

The strategy adopted to play this game is detailed below, and the other pages detail the implementation of these strategies.

Points Strategy

We took inspiration from the GOAT basketball team The Golden State Warriors. Using economic averaging, we determined that the more advantageous cost-benefit approach to win would be to score a few points into the recycling center, and the major challenge was then locating and navigating to our recycling center in as little time as possible. However, through the project development, we changed our strategy to scoring in the landfill as well.

Navigation

We decided early on in the development to try to have a redundant array of sensors to better be able to localize ourselves and navigate to our goal locations. To this end we worked to getting three types of sensors to work:

Ultrasound sensors for wall following and distance measurements
IR beacon detection and subsequent triangulation for localisation
Limit switches to detect the presence of another bot

All these sensors were tested and code was written and successfully tested to run them and perform the expected task, but due to time constraints, they could not all be integrated into the bot.

Ball Collection

Several strategies were considered and prototyped. The two most promising ones were:

Using a funnel in the direction of motion of the balls to direct them into a conveyor belt that took them up to be classified into recycling or trash.
Using a funnel in the direction of motion of the balls that directed them into the stream of a fan blowing perpendicular to the motion of the bot, which pushed them up a chute to be classified as recycling or trash.

We decided to go with the fan as it proved to be more reliable; however it did consume significantly more power than the conveyor belt.

Overall Game Strategy

Our strategy evolved during the entire duration of the project build-up. In the beginning, our strategy was to win. Then our strategy was to get a robot that worked. Then our strategy relegated again to not embarrass ourselves during the competition.

Our initial strategy is summarized in the block diagram below:

Initial Strategy

However, although all the individual modules were completed and software tested individually, they could not be integrated and tested on the bot in time and hence we had to revert to a more simplistic strategy at the end, which is detailed below: