KHOA NGUYEN

The final demonstration for Summer Studio was coming close, we had only one week left to finalize all design prototypes and test all control algorithms. Until Monday of the last week, there were some accomplishments that I have made:

• Control algorithms for the UR3 that takes in a set of x y z matrices alongside with orientations for each set of x-y-z. From the given inputs, the UR3 will move on a trajectory formed by those inputs.

• Computer Vision algorithm that can classify different color blocks and identify the position of each edge of the given block.

Meet has finally come up with a final design for his multi-tool hub which has some changes in comparison with the model I was using to develop the control algorithm for the UR3. The main difference was the way he positioned the tool hub and the drill bit, which caused a slight offset from the centre line of the end effector. I had to fix and make some adjustment to my code according to the new design of Meet. Luckily, it was not so difficult as I tried to modulize my code as much as possible and keep track of the changes in the position of the robot working area coordinate frame.

Meet and I wanted to mount two new tools, a laser engraver and a cutter. However, the way he designed the hub was difficult for doing cutting as the blade was always parallel to the ground.  The laser engraver needed a separate circuit to control it, which we believed to be a very long task to work on and also not very safe. Finally, we decided to focus on making the existing features more robust and accurate. My task was to make sure that all control algorithms used for the drilling and the drawing work according to the new design perfectly.

Another task that was in my initial plan for the final week was to do the visual servoing. However, it failed when I was trying to implement it with the real robot. I believed that the main reason that took responsibility for my failure was Computer Vision. The edges changed from frame to frame even though the camera and the cube were stationary. My test was conducted on Wednesday and I had little time to try to come up with an alternative method for the current issue. Therefore, I decided to skip that section and returned on polishing the drilling and drawing tasks, which will be my main demonstration alongside with the Computer Vision.

Finally, I created a UI for the whole program to make the interaction between the visitors and the system becomes friendly. The visitors can use the buttons to choose the mode that they want to see and input some parameters and positions to guide the drill bit or the pen to that exact location. Here are some images of the GUI: