The Robot Round-Up

During the robot round up, most robots followed a fairly similar approach to solving each of the puzzles. Most robots explored as many squares as possible to get as much information as possible before determining a square to shoot the Wumpus or where the gold was located. Some of the robots followed their original path to get back home while most of them (including us) used a pathfinding algorithm to find a shorter path.

In the first round of the roundup, it became apparent that there was something wrong with our robot. Our robot explored 50 percent more squares than the others before settling on where the gold was. This was unexpected behavior for our robot. As soon as our robot had explored two squares adjacent to the gold, it should have (and it has in the past) determined where the gold was. It was pretty disappointing when it became apparent that there was at least some bug in the code. Such a crucial feature pretty much spelled our defeat.

One thing that stood out to us was that there were not many groups that used a danger system like ours did. The problems stated that there was a findable solution to every single puzzle. Many groups coded their robots with this in mind. Our robot had a danger system that allowed it to take a chance on a square. In a real-world application, such a risk-taking approach would be less than optimal. If our robot had made it to the last stage, however, we would have been able to solve it, even if it was in a suboptimal way.

In the future, some more rigorous testing and fewer last-minute additions would have made our robot a lot more robust. We would have been able to catch the error before it was showcased to the entire class. A little bit more planning ahead is always a good thing.