Personally I feel this sprint ran a lot better than the last one. I think that the shuffle of teams worked a lot better as there seemed to be a lot more communication between each team member, While everybody was assigned their own tasks, there was more help available. My task was to create a shape detect program, which at the start I found difficult as I had never managed to program a visual detection program before. I got the task completed however, it would’ve been nice to see it operating with more of the lamp modules on the day of the test. I did feel however that the test day was a lot less stressful than the previous one as everybody seemed a lot more prepared.
As I mentioned above, the teams seemed to work a lot better together in this sprint than in the previous sprint. I felt that there were a few leaders in our team, those being mainly Kamil, along with Luke and David. Kamil was always available to help each member of the team along with working on his own task also. There was a few errors in my code when it came to the testing and execution during the build stages. With the help of Ian and Luke I managed to find that the amount of arguments being supplied to the function that would count the sides of the shape was incorrect. This meant that the program was crashing out before it could reach the stage of counting the sides, meaning it was not able to identify the shapes. This error was fixed with some help from the guys.
Project management was quite well organised, I feel, in terms of each team member being assigned a task that they chose from the list on the first day of the sprint. This meant that there was a bit more time available to the team to work on bugs. The idea to split the group into two, one group working on the software side and the other group working on the mechanical side, was a good idea. I feel this was good as it meant we got a choice of what area we liked to work in. Unfortunately, the combination of the modules on the day of the test did not work. This was mainly due to leaving the combining of the modules until the day of the test but each module seemed to work on its own.
Technically, I think my module worked to a certain extent. One of the major problems I found while executing my code was that noise was playing a big part of increasing the amount of bugs. Due to a shake in a person’s hand and the focus distance of the PiCamera the image was easily blurred. At first I decided to test the module with shapes created as a JPG file in paint. This returned the correct name of the shape on each execution of the code. Once the camera was introduced the problems began. I could set up the project in one room and it may work perfectly. Once the project was moved to a different room, the lighting would change, meaning that the image system would be more or less noisy than the previous location. This was a big problem which I think may be hard to fix. One suggestion was to introduce the shape on a black background which I hope to try in the next sprint.
William, ESD, Post #4