Loomo Guiding Robot

At the first time we participate in this project, we are attracted by the cute name of 'Loomo.' So, a week before the competition, we decide to work with Loomo and help him to give a tour in Makerspace.

However, after we went over the sample code on the developer website and familiarized outselves with the complicated circumstance in makerspace, we found that it is hard to use our original algorithm (using RGD-B camera and OpenCV to distinguish the obstracles and identify the working sector's name).

At that time, we are so depressed that we even perpared a back-up plan in case Loomo cannot even move.

But one of my teammates, Qu, decided to try one more hours and he use the most stupid method to finish this projects: draw a articfical line on the 2D space that integrated on Loomo. The question is we find out that both Loomo robots in makrspace cannot work so smoothly: one of them cannot connect to WiFi and the other's speed sensor has some defection which prevents it from traveling in a straight line. Since we did not find instructions on methods to interface directly with the motor speed controller, we had to stick with the sensor defection throughout our project. After a long time of testing and trying to fix the problem, we decide to use regression analysis to adjust Loomo's routine. In the testing, we find that it is hard to estimate the routine for Loomo as long as the input or original location has a little difference, the x and y line would have a great distortion. In the other word, the 2D space is not the same we observe and there are some hidden variable that various based on each test.

Because of the sensor's defection and the uncertainty in motion due to Loomo's own self-balancing sysem, most of our time was spent on trying to determine the best location and number of checkpoints. We made a series of tests to determine the maximum uncertainty of each tour.

In the end, we picked several coordinates that reduced the random errors the most as checking points and managed to finish a portion of the makerspace tour.